Handbook of Practical Medicine Stroke

9781405127660_4_016.qxd 10/13/07 10:51 AM Page 897 References 897 asymptomatic carotid artery disease. Brain 1995; 464 Benade MM, Warlow CP. Costs and benefits of carotid 118(Pt 4):1005–11. endarterectomy and associated preoperative arterial 449 Van Zuilen EV, Moll FL, Vermeulen FE, Mauser HW, van imaging: a systematic review of health economic literature. Gijn J, Ackerstaff RG. Detection of cerebral microemboli by Stroke 2002; 33(2):629–38. means of transcranial Doppler monitoring before and after 465 Rothwell PM, Eliasziw M, Gutnikov SA, Warlow CP, carotid endarterectomy. Stroke 1995; 26(2):210–13. Barnett HJ. Sex difference in the effect of time from 450 Lloyd AJ, Hayes PD, London NJ, Bell PR, Naylor AR. Does symptoms to surgery on benefit from carotid carotid endarterectomy lead to a decline in cognitive endarterectomy for transient ischemic attack and function or health related quality of life? J Clin Exp nondisabling stroke. Stroke 2004; 35(12):2855–61. Neuropsychol 2004; 26(6):817–25. 466 Rothwell PM, Eliasziw M, Gutnikov SA, Warlow CP, 451 Bossema ER, Brand N, Moll FL, Ackerstaff RG, van Doornen Barnett HJ. Endarterectomy for symptomatic carotid LJ. Perioperative microembolism is not associated with stenosis in relation to clinical subgroups and timing of cognitive outcome three months after carotid surgery. Lancet 2004; 363(9413):915–24. endarterectomy. Eur J Vasc Endovasc Surg 2005; 467 Bond R, Rerkasem K, Cuffe R, Rothwell PM. A systematic 29(3):262–8. review of the associations between age and sex and the 452 Lal BK. Cognitive function after carotid artery operative risks of carotid endarterectomy. Cerebrovasc Dis revascularization. Vasc Endovascular Surg 2007; 41(1):5–13. 2005; 20(2):69–77. 453 Bo M, Massaia M, Speme S, Cappa G, Strumia K, Cerrato P 468 Sundt TM, Sandok BA, Whisnant JP. Carotid et al. Risk of cognitive decline in older patients after carotid endarterectomy. Complications and preoperative endarterectomy: an observational study. J Am Geriatr Soc assessment of risk. Mayo Clin Proc 1975; 50(6):301–6. 2006; 54(6):932–6. 469 McCrory DC, Goldstein LB, Samsa GP, Oddone EZ, 454 Lunn S, Crawley F, Harrison MJ, Brown MM, Newman SP. Landsman PB, Moore WS et al. Predicting complications Impact of carotid endarterectomy upon cognitive of carotid endarterectomy. Stroke 1993; 24(9):1285–91. functioning. A systematic review of the literature. 470 Goldstein LB, McCrory DC, Landsman PB, Samsa GP, Cerebrovasc Dis 1999; 9(2):74–81. Ancukiewicz M, Oddone EZ et al. Multicenter review of 455 Schroeder T. Hemodynamic significance of internal preoperative risk factors for carotid endarterectomy in carotid artery disease. Acta Neurol Scand 1988; patients with ipsilateral symptoms. Stroke 1994; 77(5):353–72. 25(6):1116–21. 456 Naylor AR, Merrick MV, Sandercock PA, Gillespie I, Allen P, 471 Golledge J, Cuming R, Beattie DK, Davies AH, Griffin TM et al. Serial imaging of the carotid bifurcation Greenhalgh RM. Influence of patient-related variables on and cerebrovascular reserve after carotid endarterectomy. the outcome of carotid endarterectomy. J Vasc Surg 1996; Br J Surg 1993; 80(10):1278–82. 24(1):120–6. 457 Yonas H, Smith HA, Durham SR, Pentheny SL, Johnson 472 Rothwell PM, Slattery J, Warlow CP. Clinical and DW. Increased stroke risk predicted by compromised angiographic predictors of stroke and death from carotid cerebral blood flow reactivity. J Neurosurg 1993; endarterectomy: systematic review. Br Med J 1997; 79(4):483–9. 315(7122):1571–7. 458 Hartl WH, Janssen I, Furst H. Effect of carotid 473 Pritz MB. Timing of carotid endarterectomy after stroke. endarterectomy on patterns of cerebrovascular reactivity Stroke 1997; 28(12):2563–67. in patients with unilateral carotid artery stenosis. Stroke 474 Eckstein HH, Schumacher H, Klemm K, Laubach H, Kraus 1994; 25(10):1952–7. T, Ringleb P et al. Emergency carotid endarterectomy. 459 Yamauchi H, Fukuyama H, Nagahama Y, Nabatame H, Cerebrovasc Dis 1999; 9(5):270–81. Nakamura K, Yamamoto Y et al. Evidence of misery 475 Fairhead JF, Rothwell PM. The need for urgency in perfusion and risk for recurrent stroke in major cerebral identification and treatment of symptomatic carotid arterial occlusive diseases from PET. J Neurol Neurosurg stenosis is already established. Cerebrovasc Dis 2005; Psychiatry 1996; 61(1):18–25. 19(6):355–8. 460 Visser GH, van Huffelen AC, Wieneke GH, Eikelboom BC. 476 Blaser T, Hofmann K, Buerger T, Effenberger O, Wallesch Bilateral increase in CO2 reactivity after unilateral carotid CW, Goertler M. Risk of stroke, transient ischemic attack, endarterectomy. Stroke 1997; 28(5):899–905. and vessel occlusion before endarterectomy in patients 461 Silvestrini M, Vernieri F, Pasqualetti P, Matteis M, Passarelli with symptomatic severe carotid stenosis. Stroke 2002; F, Troisi E et al. Impaired cerebral vasoreactivity and risk of 33(4):1057–62. stroke in patients with asymptomatic carotid artery 477 Fairhead JF, Mehta Z, Rothwell PM. Population-based stenosis. J Am Med Assoc 2000; 283(16):2122–7. study of delays in carotid imaging and surgery and 462 Markus H, Cullinane M. Severely impaired cerebrovascular the risk of recurrent stroke. Neurology 2005; reactivity predicts stroke and TIA risk in patients with 65(3):371–5. carotid artery stenosis and occlusion. Brain 2001; 478 Welsh S, Mead G, Chant H, Picton A, O’Neill PA, 124(Pt 3):457–67. McCollum CN. Early carotid surgery in acute stroke: a 463 Benade MM, Warlow CP. Cost of identifying patients for multicentre randomised pilot study. Cerebrovasc Dis 2004; carotid endarterectomy. Stroke 2002; 33(2):435–9. 18(3):200–5. .. ..

9781405127660_4_016.qxd 10/13/07 10:51 AM Page 898 898 Chapter 16 Preventing recurrent stroke and other serious vascular events 479 Rantner B, Pavelka M, Posch L, Schmidauer C, Fraedrich G. 492 Ascher E, Markevich N, Hingorani A, Kallakuri S. Pseudo- Carotid endarterectomy after ischemic stroke: is there a occlusions of the internal carotid artery: a rationale for justification for delayed surgery? Eur J Vasc Endovasc Surg treatment on the basis of a modified carotid duplex scan 2005; 30(1):36–40. protocol. J Vasc Surg 2002; 35(2):340–5. 480 Rodgers H, Oliver SE, Dobson R, Thomson RG. A regional 493 Berman SS, Devine JJ, Erdoes LS, Hunter GC. collaborative audit of the practice and outcome of carotid Distinguishing carotid artery pseudo-occlusion with endarterectomy in the United Kingdom. Northern color-flow Doppler. Stroke 1995; 26(3):434–8. Regional Carotid Endarterectomy Audit Group. 494 Rothwell PM. ACST: which subgroups will benefit most Eur J Vasc Endovasc Surg 2000; 19(4):362–9. from carotid endarterectomy? Lancet 2004; 481 Turnbull RG, Taylor DC, Hsiang YN, Salvian AJ, Nanji S, 364(9440):1122–3. O’Hanley G et al. Assessment of patient waiting times for 495 Rothwell PM, Warlow CP. Prediction of benefit from vascular surgery. Can J Surg 2000; 43(2):105–11. carotid endarterectomy in individual patients: a risk- 482 Pell JP, Slack R, Dennis M, Welch G. Improvements in modelling study. European Carotid Surgery Trialists’ carotid endarterectomy in Scotland: results of a national Collaborative Group. Lancet 1999; 353(9170):2105–10. prospective survey. Scott Med J 2004; 49(2):53–6. 496 Mackinnon AD, Aaslid R, Markus HS. Ambulatory 483 Rothwell PM, Warlow CP. Low risk of ischemic stroke in transcranial Doppler cerebral embolic signal detection in patients with reduced internal carotid artery lumen symptomatic and asymptomatic carotid stenosis. Stroke diameter distal to severe symptomatic carotid stenosis: 2005; 36(8):1726–30. cerebral protection due to low poststenotic flow? On 497 Molloy J, Markus HS. Asymptomatic embolization predicts behalf of the European Carotid Surgery Trialists’ stroke and TIA risk in patients with carotid artery stenosis. Collaborative Group. Stroke 2000; 31(3):622–30. Stroke 1999; 30(7):1440–3. 484 Morgenstern LB, Fox AJ, Sharpe BL, Eliasziw M, 498 Mead GE, Lewis SC, Wardlaw JM, Dennis MS, Warlow CP. Barnett HJ, Grotta JC. The risks and benefits of carotid Severe ipsilateral carotid stenosis and middle cerebral endarterectomy in patients with near occlusion of the artery disease in lacunar ischaemic stroke: innocent carotid artery. North American Symptomatic Carotid bystanders? J Neurol 2002; 249(3):266–71. Endarterectomy Trial (NASCET) Group. Neurology 1997; 499 Boiten J, Rothwell P, Slattery J, Warlow C, for the European 48(4):911–15. Carotid Surgery Trialists’ Collaborative Group. Lacunar 485 Eliasziw M, Streifler JY, Fox AJ, Hachinski VC, Ferguson stroke in the European Carotid Surgery Trial: risk factos, GG, Barnett HJ. Significance of plaque ulceration in distribution of carotid stenosis, effect of surgery and type symptomatic patients with high-grade carotid stenosis. of recurrent stroke. Cerebrovasc Dis 1996; 6:281–7. North American Symptomatic Carotid Endarterectomy 500 Inzitari D, Eliasziw M, Sharpe BL, Fox AJ, Barnett HJ. Risk Trial. Stroke 1994; 25(2):304–8. factors and outcome of patients with carotid artery stenosis 486 Rothwell PM, Gibson R, Warlow CP. Interrelation between presenting with lacunar stroke. North American plaque surface morphology and degree of stenosis on Symptomatic Carotid Endarterectomy Trial Group. carotid angiograms and the risk of ischemic stroke in Neurology 2000; 54(3):660–6. patients with symptomatic carotid stenosis. On behalf of 501 Chaturvedi S, Bruno A, Feasby T, Holloway R, Benavente the European Carotid Surgery Trialists’ Collaborative O, Cohen SN et al. Carotid endarterectomy: an evidence- Group. Stroke 2000; 31(3):615–21. based review. Report of the Therapeutics and Technology 487 Rothwell PM, Gibson RJ, Slattery J, Sellar RJ, Warlow CP. Assessment Subcommittee of the American Academy of Equivalence of measurements of carotid stenosis: a Neurology. Neurology 2005; 65(6):794–801. comparison of three methods on 1001 angiograms. 502 Chambers BR, Donnan GA. Carotid endarterectomy for European Carotid Surgery Trialists’ Collaborative Group. asymptomatic carotid stenosis. Cochrane Database Syst Rev Stroke 1994; 25(12):2435–9. 2005; (4):CD001923. 488 Rothwell PM, Gutnikov SA, Warlow CP. Reanalysis of the 503 Executive Committee for the Asymptomatic Carotid final results of the European Carotid Surgery Trial. Stroke Atherosclerosis Study. Endarterectomy for asymptomatic 2003; 34(2):514–23. carotid artery stenosis. J Am Med Assoc 1995; 489 Johnston DC, Goldstein LB. Clinical carotid 273(18):1421–8. endarterectomy decision making: noninvasive vascular 504 Moore WS, Young B, Baker WH, Robertson JT, Toole JF, imaging versus angiography. Neurology 2001; Vescera CL et al. Surgical results: a justification of the 56(8):1009–15. surgeon selection process for the ACAS trial. The ACAS 490 Norris JW, Rothwell PM. Noninvasive carotid imaging Investigators. J Vasc Surg 1996; 23(2):323–8. to select patients for endarterectomy: is it really safer 505 MRC Asymptomatic Carotid Surgery Trial (ACST) than conventional angiography? Neurology 2001; Collaborative Group. Halliday A, Mansfield A, Marro J, 56(8):990–1. Peto C, Peto R, Potter J et al. Prevention of disabling and 491 Rothwell PM, Pendlebury ST, Wardlaw J, Warlow CP. fatal strokes by successful carotid endarterectomy in Critical appraisal of the design and reporting of studies of patients without recent neurological symptoms: imaging and measurement of carotid stenosis. Stroke 2000; randomised controlled trial. Lancet 2004; 31(6):1444–50. 363(9420):1491–502. .. ..

9781405127660_4_016.qxd 10/13/07 10:51 AM Page 899 References 899 506 Bond R, Rerkasem K, Rothwell P. High morbidity due to inflammatory process irrespective of the dominant endarterectomy for asymptomatic carotid stenosis. plaque morphology. Circulation 1994; 89(1):36–44. Cerebrovasc Dis 2003; 16(Suppl 4):65. 520 Redgrave JN, Lovett JK, Gallagher PJ, Rothwell PM. 507 Kresowik TF, Bratzler DW, Kresowik RA, Hendel ME, Histological assessment of 526 symptomatic carotid Grund SL, Brown KR et al. Multistate improvement in plaques in relation to the nature and timing of ischemic process and outcomes of carotid endarterectomy. J Vasc symptoms: the Oxford plaque study. Circulation 2006; Surg 2004; 39(2):372–80. 113(19):2320–8. 508 Halliday AW, Thomas D, Mansfield A. The Asymptomatic 521 Trivedi RA, King-Im JM, Graves MJ, Cross JJ, Horsley J, Carotid Surgery Trial (ACST). Rationale and design. Goddard MJ et al. In vivo detection of macrophages in Steering Committee. Eur J Vasc Surg 1994; 8(6):703–10. human carotid atheroma: temporal dependence of 509 Cuffe RL, Rothwell PM. Effect of nonoptimal imaging on ultrasmall superparamagnetic particles of iron oxide- the relationship between the measured degree of enhanced MRI. Stroke 2004; 35(7):1631–5. symptomatic carotid stenosis and risk of ischemic stroke. 522 Tang T, Howarth SP, Miller SR, Trivedi R, Graves MJ, Stroke 2006; 37(7):1785–91. King-Im JU et al. Assessment of inflammatory burden 510 Nicolaides AN, Kakkos SK, Griffin M, Sabetai M, Dhanjil S, contralateral to the symptomatic carotid stenosis using Tegos T et al. Severity of asymptomatic carotid stenosis and high-resolution ultrasmall, superparamagnetic iron oxide- risk of ipsilateral hemispheric ischaemic events: results enhanced MRI. Stroke 2006; 37(9):2266–70. from the ACSRS study. Eur J Vasc Endovasc Surg 2005; 523 Dotter CT, Judkins MP, Rosch J. Nonoperative treatment of 30(3):275–84. arterial occlusive disease: a radiologically facilitated 511 Baker WH, Howard VJ, Howard G, Toole JF. Effect of technique. Radiol Clin North Am 1967; 5(3):531–42. contralateral occlusion on long-term efficacy of 524 Mathur A, Roubin GS, Iyer SS, Piamsonboon C, Liu MW, endarterectomy in the asymptomatic carotid Gomez CR et al. Predictors of stroke complicating carotid atherosclerosis study (ACAS). ACAS Investigators. Stroke artery stenting. Circulation 1998; 97(13):1239–45. 2000; 31(10):2330–4. 525 McCabe DJ, Brown MM, Clifton A. Fatal cerebral 512 Spence JD, Tamayo A, Lownie SP, Ng WP, Ferguson GG. reperfusion hemorrhage after carotid stenting. Stroke 1999; Absence of microemboli on transcranial Doppler identifies 30(11):2483–6. low-risk patients with asymptomatic carotid stenosis. 526 Qureshi AI, Luft AR, Sharma M, Janardhan V, Lopes DK, Stroke 2005; 36(11):2373–8. Khan J et al. Frequency and determinants of 513 Abbott AL, Chambers BR, Stork JL, Levi CR, Bladin CF, postprocedural hemodynamic instability after carotid Donnan GA. Embolic signals and prediction of ipsilateral angioplasty and stenting. Stroke 1999; 30(10):2086–93. stroke or transient ischemic attack in asymptomatic 527 CAVATAS Investigators. Endovascular versus surgical carotid stenosis: a multicenter prospective cohort study. treatment in patients with carotid stenosis in the Carotid Stroke 2005; 36(6):1128–33. and Vertebral Artery Transluminal Angioplasty Study 514 Markus H, Cullinane M. Asymptomatic Carotid Emboli (CAVATAS): a randomised trial. Lancet 2001; 357:1729–37. (ACES) Study. Cerebrovasc Dis 2000; 10(Suppl 1):3. 528 Naylor AR, Bolia A, Abbott RJ, Pye IF, Smith J, Lennard N 515 Gronholdt ML. Ultrasound and lipoproteins as predictors et al. Randomized study of carotid angioplasty and of lipid-rich, rupture-prone plaques in the carotid artery. stenting versus carotid endarterectomy: a stopped trial. Arterioscler Thromb Vasc Biol 1999; 19(1):2–13. J Vasc Surg 1998; 28(2):326–34. 516 Nicolaides AN, Kakkos SK, Griffin M, Sabetai M, Dhanjil S, 529 Alberts MJ, for the Publications Committee of the Thomas DJ et al. Effect of image normalization on carotid WALLSTENT. Results of a multicentre prospective plaque classification and the risk of ipsilateral hemispheric randomised trial of carotid artery stenting vs carotid ischemic events: results from the asymptomatic carotid endarterectomy. Stroke 2001; 32:325. stenosis and risk of stroke study. Vascular 2005; 530 Brooks WH, McClure RR, Jones MR, Coleman TC, 13(4):211–21. Breathitt L. Carotid angioplasty and stenting versus carotid 517 Takaya N, Yuan C, Chu B, Saam T, Underhill H, Cai J et al. endarterectomy: randomized trial in a community Association between carotid plaque characteristics and hospital. J Am Coll Cardiol 2001; 38(6):1589–95. subsequent ischemic cerebrovascular events: a prospective 531 Yadav JS, Wholey MH, Kuntz RE, Fayad P, Katzen BT, assessment with MRI: initial results. Stroke 2006; Mishkel GJ et al. Protected carotid-artery stenting versus 37(3):818–23. endarterectomy in high-risk patients. N Engl J Med 2004; 518 Cai J, Hatsukami TS, Ferguson MS, Kerwin WS, Saam T, 351(15):1493–501. Chu B et al. In vivo quantitative measurement of intact 532 Coward LJ, Featherstone RL, Brown MM. Safety and fibrous cap and lipid-rich necrotic core size in eficacy of endovascular treatment of carotid artery stenosis atherosclerotic carotid plaque: comparison of high- compared with carotid endarterectomy: a Cochrane resolution, contrast-enhanced magnetic resonance systematic review of the randomized evidence. Stroke 2005; imaging and histology. Circulation 2005; 112(22):3437–44. 36(4):905–11. 519 Van der Wal AC, Becker AE, van der Loos CM, Das PK. 533 Reimers B, Corvaja N, Moshiri S, Sacca S, Albiero R, Di Site of intimal rupture or erosion of thrombosed Mario C et al. Cerebral protection with filter devices during coronary atherosclerotic plaques is characterized by an carotid artery stenting. Circulation 2001; 104(1):12–15. .. ..

9781405127660_4_016.qxd 10/13/07 10:51 AM Page 900 900 Chapter 16 Preventing recurrent stroke and other serious vascular events 534 Mas JL, Chatellier G, Beyssen B, Branchereau A, Moulin T, 550 Powers WJ, Derdeyn CP, Fritsch SM, Carpenter DA, Yundt Becquemin JP et al. Endarterectomy versus stenting in KD, Videen TO et al. Benign prognosis of never- patients with symptomatic severe carotid stenosis. N Engl J symptomatic carotid occlusion. Neurology 2000; Med 2006; 355(16):1660–71. 54(4):878–82. 535 Featherstone RL, Brown MM, Coward LJ. International 551 EC-IC Bypass Study Group. Failure of extracranial- carotid stenting study: protocol for a randomised clinical intracranial arterial bypass to reduce the risk of ischaemic trial comparing carotid stenting with endarterectomy in stroke: results of an international randomised trial. N Engl J symptomatic carotid artery stenosis. Cerebrovasc Dis 2004; Med 1985; 313:1191–200. 18(1):69–74. 552 Warlow CP. Extracranial to intracranial bypass and the 536 CARESS Steering Committee. Carotid Revascularization prevention of stroke. J Neurol 1986; 233(3):129–30. Using Endarterectomy or Stenting Systems (CaRESS) 553 Derdeyn CP, Grubb RL, Jr., Powers WJ. Indications for phase I clinical trial: 1-year results. J Vasc Surg 2005; cerebral revascularization for patients with atherosclerotic 42(2):213–19. carotid occlusion. Skull Base 2005; 15(1):7–14. 537 Major ongoing stroke trials. Stroke 2006; 37(10):e36–e44. 554 Grubb RL, Jr., Powers WJ, Derdeyn CP, Adams HP, Jr., 538 Graor RA, Hetzer NR. Management of coexistent carotid Clarke WR. The Carotid Occlusion Surgery Study. artery and coronary artery disease. Stroke 1988; Neurosurg Focus 2003; 14(3):e9. 19(11):1441–4. 555 Eberhardt O, Naegele T, Raygrotzki S, Weller M, Ernemann 539 Akins CW. The case for concomitant carotid and coronary U. Stenting of vertebrobasilar arteries in symptomatic artery surgery. Br Heart J 1995; 74(2):97–8. atherosclerotic disease and acute occlusion: case series and 540 Davenport RJ, Dennis MS, Sandercock PA, Warlow CP, review of the literature. J Vasc Surg 2006; 43(6):1145–54. Starkey IR, Ruckley CV et al. How should a patient 556 Cloud GC, Crawley F, Clifton A, McCabe DJ, Brown MM, presenting with unstable angina and a recent stroke be Markus HS. Vertebral artery origin angioplasty and managed? Br Med J 1995; 310(6992):1449–52. primary stenting: safety and restenosis rates in a 541 Hertzer NR, Loop FD, Beven EG, O’Hara PJ, Krajewski LP. prospective series. J Neurol Neurosurg Psychiatry 2003; Surgical staging for simultaneous coronary and carotid 74(5):586–90. disease: a study including prospective randomization. 557 Coward LJ, Featherstone RL, Brown MM. Percutaneous J Vasc Surg 1989; 9(3):455–63. transluminal angioplasty and stenting for vertebral artery 542 Borger MA, Fremes SE, Weisel RD, Cohen G, Rao V, stenosis. Cochrane Database Syst Rev 2005; (2):CD000516. Lindsay TF et al. Coronary bypass and carotid 558 Osterberg L, Blaschke T. Adherence to medication. N Engl J endarterectomy: does a combined approach increase risk? Med 2005; 353(5):487–97. a meta-analysis. Ann Thorac Surg 1999; 68(1):14–20. 559 Benner JS, Glynn RJ, Mogun H, Neumann PJ, Weinstein 543 Naylor AR, Mehta Z, Rothwell PM, Bell PR. Carotid artery MC, Avorn J. Long-term persistence in use of statin disease and stroke during coronary artery bypass: a critical therapy in elderly patients. J Am Med Assoc 2002; review of the literature. Eur J Vasc Endovasc Surg 2002; 288(4):455–61. 23(4):283–94. 560 Jackevicius CA, Mamdani M, Tu JV. Adherence with statin 544 Naylor AR, Cuffe RL, Rothwell PM, Bell PR. A systematic therapy in elderly patients with and without acute review of outcomes following staged and synchronous coronary syndromes. J Am Med Assoc 2002; 288(4):462–7. carotid endarterectomy and coronary artery bypass. Eur J 561 Trewby PN, Reddy AV, Trewby CS, Ashton VJ, Brennan G, Vasc Endovasc Surg 2003; 25(5):380–9. Inglis J. Are preventive drugs preventive enough? A study 545 Naylor AR. A critical review of the role of carotid disease of patients’ expectation of benefit from preventive drugs. and the outcomes of staged and synchronous carotid Clin Med 2002; 2(6):527–33. surgery. Semin Cardiothorac Vasc Anesth 2004; 8(1):37–42. 562 Hamman GF, Weimar C, Glahn J, Busse O, Diener HC. 546 Latchaw RE, Ausman JI, Lee MC. Superficial Adherence to secondary stroke prevention strategies: temporal-middle cerebral artery bypass. A detailed results from the German Stroke Data Bank. Cerebrovasc Dis analysis of multiple pre- and postoperative angiograms 2003; 15(4):282–8. in 40 consecutive patients. J Neurosurg 1979; 563 Haynes RB, Yao X, Degani A, Kripalani S, Garg A, 51(4):455–65. McDonald HP. Interventions to enhance medication 547 Hankey GJ, Warlow C. Prognosis of symptomatic carotid adherence. Cochrane Database Syst Rev 2005; artery occlusion: an overview. Cerebrovasc Dis 1991; (4):CD000011. 1:245–56. 564 Gigerenzer G, Edwards A. Simple tools for understanding 548 Karnik R, Valentin A, Ammerer HP, Donath P, Slany J. risks: from innumeracy to insight. Br Med J 2003; Evaluation of vasomotor reactivity by transcranial Doppler 327(7417):741–4. and acetazolamide test before and after extracranial- 565 Edwards A, Elwyn G, Mulley A. Explaining risks: turning intracranial bypass in patients with internal carotid artery numerical data into meaningful pictures. Br Med J 2002; occlusion. Stroke 1992; 23(6):812–17. 324(7341):827–30. 549 Klijn CJ, Kappelle LJ, Tulleken CA, van Gijn J. 566 Alaszewski A, Horlick-Jones T. How can doctors Symptomatic carotid artery occlusion. A reappraisal of communicate information about risk more effectively? hemodynamic factors. Stroke 1997; 28(10):2084–93. Br Med J 2003; 327(7417):728–31. .. ..

9781405127660_4_016.qxd 10/13/07 10:51 AM Page 901 References 901 567 Fuller R, Dudley N, Blacktop J. Avoidance hierarchies 570 Wald NJ, Law MR. A strategy to reduce cardiovascular and preferences for anticoagulation: semi-qualitative disease by more than 80%. Br Med J 2003; 326(7404):1419. analysis of older patients’ views about stroke prevention 571 Zarins CK, Gewertz BL. Atlas of Vascular Surgery. New York: and the use of warfarin. Age Ageing 2004; Churchill Livingstone, 1987. 33(6):608–11. 572 Schroeder T, Levi N. What steps can I take to minimise 568 Zhan C, Correa-de-Araujo R, Bierman AS, Sangl J, inadvertent cranial nerve injury. In: Naylor AR, Mackey Miller MR, Wickizer SW et al. Suboptimal prescribing in AE, eds. Carotid Artery Surgery: A Problem Based Approach. elderly outpatients: potentially harmful drug-drug and London: Saunders, 1999. drug-disease combinations. J Am Geriatr Soc 2005; 573 The European Carotid Surgery Trialists Collaborative 53(2):262–7. Group. Risk of stroke in the distribution of an 569 White C. ‘Polypill’ to fight cardiovascular disease. asymptomatic carotid artery. Lancet 1995; Summary of rapid responses. Br Med J 2003; 327:809. 345(8944):209–12. .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 903 17 The organization of stroke services 17.1 Introduction 903 17.2 Planning and developing a stroke service 906 17.3 Comprehensive stroke service 906 17.4 General principles when discussing a comprehensive stroke service 906 17.5 Interventions to improve access to early specialist assessment and treatment 910 17.6 Organized inpatient (stroke unit) care 914 17.7 Transfer from hospital to community 921 17.8 Continuing rehabilitation and reintegration back to normal life 923 17.9 Longer-term follow-up and chronic disease management 925 17.10 Generic issues in stroke service delivery 926 17.11 Planning, developing and maintaining a stroke service 927 17.12 Evaluating and monitoring stroke services 934 17.13 Stroke guidelines 943 17.14 Integrated care pathways 943 17.15 Impact of a comprehensive stroke service 944 17.16 Cost-effectiveness of stroke services 944 Relatively little attention was paid to stroke in the UK 17.1 Introduction until the publication of the King’s Fund Consensus Conference 13 in 1988. This highlighted the many defi- ciencies in the services provided for stroke patients, concluding that ‘services were often haphazard and 17.1.1 The impact of stroke poorly tailored to the patient’s needs’. Since then, in Governments, and in particular those responsible for the UK, stroke has moved up the political agenda, and providing healthcare, have become increasingly aware was identified as a key chapter in the National Service of the impact that stroke has on the health of the Framework for Older People in England and Wales population and the cost to the community. In the UK, (www.dh.gov.uk/PolicyAndGuidance/HealthAndSocial stroke patients account for about 6% of hospital costs CareTopics/OlderPeoplesServices) and the Coronary and almost 5% of National Health Service costs 1 Heart Disease and Stroke strategy in Scotland (www. (www.nao.org.uk/stroke). Studies from other countries scotland.gov.uk/library5/health/chds). There has been a (e.g. Sweden, US, Canada, Netherlands and Japan) sug- similar and continuing emphasis on stroke in many gest that the financial burden may be even greater than other countries. These changes have led to a tremendous in the UK, possibly because of greater expenditure on surge of interest in stroke in general, and in stroke ser- all health services. 2–11 Moreover, demographic changes vices in particular. Over the last few years, an increasing are likely to cause increasing mortality and morbidity amount of research has been done to determine the best in the developing world. 12 and most cost-effective ways of providing care for stroke patients. This chapter will cover the organization of services for Stroke: practical management, 3rd edition. C. Warlow, J. van Gijn, people who have had a transient ischaemic attack or M. Dennis, J. Wardlaw, J. Bamford, G. Hankey, P. Sandercock, G. Rinkel, P. Langhorne, C. Sudlow and P. Rothwell. Published stroke. Inevitably, the discussion will tend to reflect the 2008 Blackwell Publishing. ISBN 978-1-4051-2766-0. UK – and to some extent other North European and 903 ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 904 904 Chapter 17 The organization of stroke services Australian – models of care, but we hope it will have 17.1.3 Establishing the effectiveness of stroke relevance for services elsewhere. services In an ideal world our decisions about the delivery of 17.1.2 Aims of stroke services stroke services would always be informed by robust evid- The overall aim of stroke services is to deliver the care ence from randomized trials and priority given to those required by patients and their families in the most effi- aspects of care that have been proven to be effective. cient, effective, equitable and humane manner possible. However, we must recognize that carrying out random- Such services may not necessarily be stroke-specific; ized trials of complex interventions such as stroke they could be part of those for internal medicine, care services is challenging 14 and these trials are often few of the elderly, neurology, rehabilitation or continuing in number and difficult to interpret. There is no reason care. why most of the important methodological principles Good organization is probably the most important (www.cochrane.org/resources/handbook) of robust factor in determining service effectiveness. When con- clinical trials (random treatment allocation, conceal- sidering exactly how stroke services should best be organ- ment of treatment allocation, blinding of follow-up, ized, it is useful to consider the main objectives when intention-to-treat analyses) should not apply to those of caring for patients with stroke and transient ischaemic stroke services. However, randomized trials of complex attack (TIA) (Table 17.1 and Fig. 17.1). In addition, given interventions such as stroke services offer some unique the lack of evidence for many of our interventions, challenges. In particular it is often difficult to develop services should facilitate research and education. We and describe the intervention adequately, to blind the have not included primary prevention among the com- trial participants to their treatment and to rule out con- ponents of care, although this is potentially the most founding from other aspects of care. Therefore even if a effective method of reducing stroke-related death, dis- service is shown to work well in one setting, specific ability and handicap, at least in the long term (Chapter local factors may have influenced the results. For these 18). Primary stroke prevention has so much in common reasons we have approached the evaluation of stroke with the prevention of other vascular diseases that it makes services in the following way: more sense to link these preventive services together, • Has the service component been shown to work in a especially as their success is likely to depend more on specific circumstance (in one randomized trial)? political and social change than on health services. • Has it been shown to work in several settings (system- atic review of clinical trials)? • Do we know if the benefits justify the costs (economic analyses)? Table 17.1 Key components for managing stroke and transient The Stroke Review Group of the Cochrane Collabora- ischaemic attack. tion is an excellent source of this type of information, Public awareness to recognize warning signs and seek help providing an increasing number of systematic reviews promptly (section 17.5.1) (http://www.dcn.ed.ac.uk/csrg). Prompt and accurate assessment and diagnosis (including It is important that those who are responsible for transportation) (Chapter 3) planning services should be aware of all the available Specific acute medical and surgical treatment (Chapters 11, evidence concerning the effectiveness of interventions, 12, 13, 14 and 15) and the methods of delivering these interventions to the Identification and assessment of patients’ problems appropriate patients. However, often we will not have (chapter 11) any reliable evidence from randomized trials to guide Secondary prevention of further vascular events (Chapter 16) our decisions simply because of the methodological General care, including interventions to resolve problems challenges outlined above. Planners should be aware (includes many aspects of rehabilitation) (Chapter 11) Provision of information and advice (section 17.10.1) that lack of evidence of benefit is not the same as evid- Terminal care for patients who are unlikely to survive ence of lack of benefit. If there are reliable data concern- (section 10.4) ing the cost of interventions, and various other aspects Hospital discharge and reintegration into the community of the service, this may allow healthcare planners to (section 17.7) make more informed choices about which services Continuing or long-term care for severely disabled patients should be provided. At the same time, they should not (section 17.8) deny patients services which are generally accepted as Follow-up to detect and manage late-onset problems being effective just because of lack of randomized trial (section 17.9) evidence. .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 905 17.1 Introduction 905 Patient develops symptoms Seeks help Public education programme Assess urgency and appropriateness Specialist assessment/triage Persisting Transient symptoms and low symptoms and/or risk of recurrence high risk of recurrence Diagnosis/assessment, Diagnosis/assessment, secondary prevention emergency treatment Rapid access neurovascular clinic Acute stroke assessment area* Diagnosis/assessment Medical management Nursing management Multidisciplinary rehabilitation Prevention of complications Secondary prevention Discharge planning Comprehensive stroke unit Fig. 17.1 A diagram illustrating Discharge planning Continuing the components, functions and Multidisciplinary multidisciplinary interrelationships of a hospital-based rehabilitation team rehabilitation stroke service that is integrated with Early supported Rehabilitation community services. The upper box discharge stroke unit** indicates the objectives of the services and the lower (purple) boxes (in italics) Continuing rehabilitation/community reintegration a proposed solution. We believe that elements of the rehabilitation process Therapy-based rehabilitation/stroke liaison worker services are important even on the day of the stroke onset. *Could be in stroke unit Re-assessment of needs/further rehabilitation or stroke centre. **Could continue in Primary care/chronic disease management comprehensive stroke unit. are delivered. Developing a stroke service in the US or Clinical trials of services can be complex and Germany would present different challenges from challenging – remember that lack of evidence of doing so in Scandinavia, the Netherlands or the UK. benefit is not the same as evidence of lack of benefit. There are likely to be particular challenges in the for- mer Soviet bloc countries and the developing world 17.1.4 Pressures that may shape stroke services where many basic assumptions about the availability Several factors other than evidence of effectiveness or of healthcare resources may simply not apply. cost-effectiveness may shape stroke service delivery and • Needs of different patient groups: stroke presents a com- constrain the options available to clinicians and service plex challenge to service planners in that most patients planners. These can include: require a similar general service but a small number • Local healthcare culture and economy: the traditional (for instance those with subarachnoid haemorrhage or approach to providing healthcare, and the way it is those eligible for intravenous thrombolysis) may require funded or reimbursed, will influence the way services more specialist services. There is frequently a tension .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 906 906 Chapter 17 The organization of stroke services between the need to centralize specialist services for can the needs of the population realistically be met? the few and the need to devolve more general services How should these resources best be organized? for the majority to local hospitals and communities. • How do we know if we have succeeded (section 17.12)? • Views of patients and families: one of the common How can the performance of the stroke services be complaints from patients and carers is the discon- monitored and maintained? tinuity of services they receive (http://www.chss.org.uk/ In this chapter we will begin (sections 17.3–17.10) pdf/research/Improving_services_patients_and_carers_ with a review of the effectiveness (and where possible views.pdf), resulting in fragmentation of care and dis- cost-effectiveness) of components of stroke services. In satisfaction with services. the second part (sections 17.11–17.16) we will discuss • Resources available: it is relatively easier to organize or practical approaches to planning, developing, monitor- reorganize a stroke service if the basic staffing levels ing and maintaining a stroke service. and investigation services are already available, albeit distributed around the hospital or community. For this reason much of our discussion will be relevant mainly to well-resourced services in developed countries. 17.3 Comprehensive stroke service Local stroke services must be tailored to local conditions; there is no perfect blueprint that can be applied everywhere. We have used the term ‘comprehensive stroke service’ to mean a stroke service that covers most of the needs of patients with stroke or transient ischaemic attack (TIA) and which is integrated in a way that provides a continu- ous patient journey – ‘a seamless service’ (Table 17.2 and 17.2 Planning and developing a stroke service Fig. 17.1). In doing so it should provide all of the func- tions outlined in Table 17.1. Figure 17.1 outlines what we believe to be important components of a comprehen- The way in which a service is best provided will depend sive stroke service, and in this chapter we will discuss the on local history, geography, needs, resources, people and evidence and rationale for some of those suggestions. politics. Any stroke service must therefore be tailored to the local conditions to achieve maximum effectiveness. A comprehensive stroke service should provide for For this reason, it is difficult to be dogmatic about exactly most of the needs of most patients and do so within how services should be organized. In this chapter we a ‘seamless service’. will attempt to provide general guidance about the prin- ciples that should be of use to the clinician, public health physician or health service manager (administrator) in planning a service. When planning or reviewing stroke services, it is use- 17.4 General principles when discussing ful to start by addressing several questions: a comprehensive stroke service • Where do we want to be (sections 17.5–17.9)? What is the evidence for the effectiveness (and cost- effectiveness) of the components of both the existing When proposing various approaches to service delivery and planned stroke service? – and suggesting particular components of a compre- • Where are we now (sections 17.11)? What are the hensive stroke service – we have tried to employ some needs of the population to be served by the stroke general principles: service and what are the current resources, people and • Basic needs first – stroke services should first ensure facilities committed to the management of patients they provide basic care (from medical, nursing and with TIAs and stroke? therapy staff) for all stroke patients and their families. • How will we get to where we want to be (section 17.12)? • Evidence-based options – priority should be given to What are the major gaps in the present provision of those aspects of care that are generally accepted as services (i.e. unmet needs and failure to provide effect- being, or have been proven to be, effective. The Stroke ive interventions)? What resources, people and facilities Review Group of the Cochrane Collaboration is an will be required to meet the needs of the population? excellent source of this type of information (http:// Assuming that resources are limited, to what extent www.dcn.ed.ac.uk/csrg). .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 907 17.4 General principles when discussing a comprehensive stroke service 907 Table 17.2 Key objectives and service solutions for managing patients with stroke and transient ischaemic attack. Objectives Proposed service options Cross-reference Public awareness to recognize warning Public education campaigns section 17.5.1 symptoms and seek help promptly Prompt and accurate assessment and Transport (ambulance and helicopter) protocols sections 17.5.2, 17.5.3, 17.5.4 diagnosis Emergency department and hospital protocols section 17.5.6, 17.5.7 Designated stroke centres Telemedicine networks Rapid access TIA clinics Specific acute medical and surgical Designated stroke centres sections 17.5.5, 17.5.6, 17.6 treatment Telemedicine networks chapters 12, 13, 14 Stroke units Identification and assessment of patients’ Designated stroke centres sections 17.5.5, 17.5.6, 17.6 problems Telemedicine networks Stroke units Secondary prevention of further vascular Stroke units sections 17.5.5, 17.5.6, 17.6 events Designated stroke centres chapter 16 Rapid access TIA clinics General care, including interventions to Stroke units section 17.6 resolve problems (includes many aspects chapter 10 of rehabilitation) Terminal care for patients who are unlikely Stroke units section 17.6 to survive section 10.4 Hospital discharge and reintegration into Early supported discharge services section 17.7 the community Discharge planning Continuing or long-term care for severely Therapy-based rehabilitation services section 17.8 disabled patients Stroke liaison worker services Follow-up to detect and manage late-onset Chronic disease management section 17.9 problems Outpatient clinics • Patient and carer views – surveys of the views of • Level of development – most of our discussion reflects patients and carers have frequently highlighted a experience from developed western economies. It is wish that care is coordinated and provided by expert at present difficult to give specific advice for other staff. A survey of user views in Scotland (http://www. healthcare settings although many of the general prin- chss.org.uk/pdf/research/Improving_services_patients ciples above will apply. _and_carers_views.pdf) highlighted five main themes: – care should be provided in designated stroke units; 17.4.1 Why refer stroke and transient ischaemic – staff should be trained in the physical and emo- attack patients to hospital? tional needs of stroke patients; – comprehensive information should be available to Before discussing each service component in turn we all patients and carers; also need to consider why we have emphasized a system – trained liaison staff should help with patients of hospital-based rather than community-based services returning home; (at least during the earlier phase of the illness). – there should be access to ‘someone to talk to’ (some- The vast majority of patients who have a stroke or one who understands the challenges that patients TIA have it in the community and not in hospital. and their carers face). Community-based incidence studies have shown wide • Awareness of alternatives – where possible we have variations in hospital admission rates for stroke, varying tried to acknowledge that there may be valid altern- from 55% in Oxfordshire, UK, in the 1980s to over 95% ative approaches to achieving the same objective. in Sweden and Germany in the 1990s. 15–17 The main However, frequently there is no research evidence reason for admission to hospital in the past was for nurs- to help make such decisions about the best service ing care rather than diagnosis and treatment. 15 However, option. the need for early imaging to establish whether a stroke .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 908 908 Chapter 17 The organization of stroke services is ischaemic or haemorrhagic has now become accepted are less reliably diagnosed than strokes and more often and immediate imaging appears to be the most cost- require input from a specialist, especially when the effective option. 18 The potential for early intervention to diagnosis is crucial to future decision-making – for ex- limit brain damage and prevent stroke recurrence 19–21 is ample, in distinguishing between carotid and posterior increasingly being recognized (Chapters 12, 16). Finally circulation TIAs in a patient with severe carotid stenosis there are a small number of comparisons (in clinical (section 4.3.1) (Table 4.3). Thus, referral to hospital for trials) of a conventional policy of admitting most patients assessment, even if only as an outpatient, is the accepted to hospital with a policy of avoiding admission by pro- norm for such patients. viding a rapid-response service in the community to sup- port patients in their own homes. Although the evidence 17.4.2 Who may not need hospital admission? is limited, 22,23 major practical problems were identified with the policy of hospital avoidance 23,24 and clinical out- In view of the above discussions, can we identify a group comes were best in those admitted directly to a stroke unit. of individuals who do not need hospital admission after Transient events, transient ischaemic attacks (TIAs) and stroke or TIA? In the absence of direct evidence from episodes that may be confused with them (Table 17.3) randomized trials comparing different policies of care, we can only base our decisions on a logical rationale. Table 17.3 Some of the non-cerebrovascular problems referred Hospital admission appears to be valuable for rapid to one of our neurovascular clinics by general practitioners over diagnosis and assessment, prompt acute treatment and a period of 5 years. There were, of course, countless patients secondary prevention plus effective early multidisciplin- with transient symptoms in whom no definite diagnosis could 25 ary rehabilitation. We would therefore suggest that be made. the following groups may be able to be managed without hospital admission: General medical problems Cardiac syncope (dysrhythmias, aortic stenosis) • For people with a TIA who are at low risk of early recur- Vasovagal syncope rence, early assessment at a one-stop neurovascular Cough syncope outpatient clinic 61,62 could be justified providing we Postural hypotension can reliably identify them (section 16.2). Hyperventilation • People with a stroke who are at low risk of early Sleep apnoea recurrence or other complications such as neurological Hypoglycaemia deterioration (section 16.2); once again this requires Neurological problems accurate identification of low-risk individuals. In Migraine (both with and without headache) practice, predicting neurological deterioration and Epilepsy other complications is difficult so this group is likely Transient global amnesia to include only those individuals who have delayed Glioma Meningioma referral for several days and so are past the higher-risk Cerebral metastases period for developing problems. Subdural haematoma • People who refuse hospital admission; rapid assess- Lymphocytic meningitis ment at a one-stop neurovascular clinic may be very Peripheral neuropathy appropriate for this group. Guillain–Barré syndrome Cervical myelopathy Brachial neuritis 17.4.3 Why is rapid specialist assessment and Mononeuropathies treatment necessary? Herpes zoster neuropathy Rapid specialist assessment of patients experiencing Bell’s palsy symptoms of acute stroke or TIA can only really be Syringobulbia Myasthenia gravis justified if that assessment improves patient outcomes Multiple sclerosis (including ‘softer’ outcomes such as reassurance and Motor neurone disease advice). The arguments in favour of rapid assessment Creutzfelt–Jakob disease have been made for both acute stroke and TIA patients: Psychiatric problems • Thrombolysis for acute stroke has been licensed for Somatization disorder several years in US and has had a restricted licence Ophthalmic problems in Europe since 2003. However, thrombolysis has not Retinal vein occlusion had the impact many of its supporters would have Glaucoma expected. First, it is a complex intervention requiring .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 909 17.4 General principles when discussing a comprehensive stroke service 909 intensive use of resources and personnel with a narrow • Recent epidemiological studies suggest that similar therapeutic window. Current product licences restrict issues may arise with the rapid diagnosis and treat- its use to within 3 h of symptom onset. This requires ment of patients with TIA or mild stroke. 29,30 The very rapid specialist assessment ensuring accurate risk of very early recurrence of cerebrovascular events diagnosis of ischaemic stroke prior to the initiation appears to be much greater than was previously of treatment. In doing so, one needs to demonstrate thought (approximately 10% in the first week and up the absence of haemorrhage on brain scanning, good to 20% in the first month; section 16.2.1) emphasizing blood pressure control, National Institute of Health the importance of very early diagnostic assessment stroke scale (NIHSS) score of greater than 4, and an and prevention. More recently, systems of identifying exclusion of patients at high risk of bleeding with very high-risk individuals have been reported based on thrombolytic therapy. Treatment must be given the presenting clinical features 21,31 or MRI findings. 32 within 3 h of symptom onset and the earlier that treat- The underlying assumption is that identifying indi- ment is given the better the outcome (section 12.5). viduals at high risk of recurrent stroke or TIA will Initial enthusiasm in the US resulted in a large number lead to effective prevention of these recurrent events. of centres commencing acute treatment with recom- Although this remains uncertain for drug treatments, binant tissue plasminogen activator (rt-PA). However, there is good evidence that very early carotid end- this demonstrated the very significant hazards of arterectomy (within 2 weeks) is not associated with an thrombolysis when implementation is poorly con- increased operative risk and results in improved stroke trolled. Haemorrhage rates as high as 16% were experi- prevention (section 16.11.8). 33,34 Furthermore there is enced and this high risk was closely correlated with no convincing argument for delaying the implementa- violations of treatment protocols. 26 The implementa- tion of other secondary prevention treatments beyond tion of a stroke quality improvement programme was the first few days post-stroke. Randomized trials of associated with a reduced rate of protocol deviation very early secondary prevention are currently under and symptomatic intracranial haemorrhage. As a way. result of these experiences, pressure groups such as the Brain Attack Coalition have proposed criteria for desig- 17.4.4 Barriers to rapid assessment and treatment nation of stroke centres that are permitted to provide treatment with rt-PA. 27 If we accept that rapid specialist assessment and treat- • Patients with intracerebral haemorrhage may benefit ment is worthwhile – at least for a proportion of stroke from early treatment with recombinant factor VII, 28 and TIA patients – we then need to consider how best although, at the time of writing, the treatment is not to achieve this (Table 17.4). A recent systematic review licensed for this use (section 13.3.1). of 54 observational studies examining the barriers to Table 17.4 Potential barriers to rapid specialist assessment and treatment, and interventions aiming to overcome them. Potential barriers Proposed solutions which have been associated with some improvements in observational studies recognition Failure of the patient or family to Public education programmes (section 17.5.1) Centralized Brain recognize symptoms of stroke or Attack Centre to seek urgent help initiatives react Failure of public and primary care Education programmes for the public and primary (designated medical services to recognize care services (section 17.5.1) stroke centres) stroke as an emergency (section 17.5.5) response Failure of ambulance and emergency Training of paramedics and emergency medical medical services to treat stroke as a services; rapid transport systems (section 17.5.2) medical emergency Training of emergency Telemedicine reveal Delays in diagnostic assessment department staff and networks (to (particularly brain imaging) redesign of hospital facilitate remote systems (section 17.5.3) specialist reperfusion Delays in delivery of acute drug Centralization of expertise assessment and therapy (where appropriate) (designated stroke centre) management) including physician’s uncertainty (section 17.5.5) (section 17.5.6) .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 910 910 Chapter 17 The organization of stroke services delivery of thrombolysis in acute stroke is also relevant urgent medical help. 38 Even with adequate knowledge when considering the barriers to more comprehensive the public has to be convinced that it is appropriate to stroke assessment and treatment, including transient call emergency services, not primary care services which ischaemic attacks. 35 so often leads to delays in management (see below). A Barriers included: number of suggestions have been made for public educa- • failure of the patient or family to recognize symptoms tion programmes to increase the knowledge of stroke of stroke or to seek urgent help; symptoms 39,40 in the hope that they would reduce delays • initially seeking help from the general practitioner and so improve treatment rates with thrombolysis, and rather than calling an ambulance; any other new emerging acute stroke treatments. How- • paramedical and emergency department staff triage ever, experience in acute myocardial infarction has stroke as non-urgent; raised concerns about the effectiveness of media cam- • delays in obtaining brain imaging; paigns. 41 Furthermore, even where people have appro- • inefficient processes of in-hospital emergency stroke priate knowledge this may not translate into action; care; patients in the Asymptomatic Carotid Atherosclerosis • difficulties in obtaining consent for treatment Study 42 showed delays in reporting stroke symptoms (thrombolysis); despite receiving targeted education. Finally, the sym- • physician uncertainty of administering treatment ptoms transmitted to the public in education campaigns (thrombolysis). are non-specific and identification and immediate action These potential barriers operate right across the initial by the public could result in a large increase in the patient pathway and a range of solutions is likely to be number of non-stroke patients reporting to emergency required if acute treatments are to be provided efficiently departments. and equitably. Only one controlled clinical trial 43 has evaluated a community education programme by comparing two communities. In the ‘intervention community’ the programme targeted ‘at risk’ members of the public, emergency medical services, emergency department 17.5 Interventions to improve access to physicians, neurologists and primary care providers, a early specialist assessment and treatment higher priority for the transport of acute stroke patients by the emergency medical services. The control com- munity did not receive these interventions. The propor- When considering the necessary components to over- tion of eligible patients treated with rt-PA appeared to come the barriers to early assessment and treatment, be greater in the intervention community and there was researchers have developed a checklist (based on the ‘five also a non-significant increase in the proportion of Rs’) to describe the key stages: patients presenting within 2 h of symptom onset (Table • recognition of symptoms; 17.5). There were also reductions in within-hospital • reaction to the acute illness; delays and in the reluctance of physicians to give rt-PA. • response of emergency services; The observed gains were relatively modest and we still • revealing the diagnosis; need more robust evidence from properly randomized • reperfusion therapy. trials. Although public education programmes are likely We can use this scheme to consider the potential solu- to be important, they may have only a relatively modest tions to these barriers. When you take this approach impact on speedy referral rates. (Table 17.4), it is striking that there is often no really reli- able evidence for particular components of the early 17.5.2 Transport protocols patient journey, and that many of the interventions that (prehospital assessment and transport) have been tested have overlapping roles. The next stage in the chain of referral is the rapid response of ambulance and emergency medical services 17.5.1 Public education programmes to treat stroke as a medical emergency. There are no ran- One of the most consistently reported prehospital domized trials of this approach but observational studies barriers to acute stroke assessment and treatment is that have described the use of rapid recognition instru- patients or families have a poor knowledge of stroke. ments 44,45 to allow paramedical staff to triage patients Poor awareness of stroke symptoms 36 and a failure to with suspected stroke. These instruments appear to have recognize their seriousness 37 often delays the request for acceptable diagnostic accuracy but the their impact on .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 911 17.5 Interventions to improve access to early specialist assessment and treatment 911 Table 17.5 Interventions aiming to facilitate rapid access to specialist assessment and treatment in acute stroke. Intervention: Education Training of Training of Designated stroke Telemedicine programmes for the paramedics and emergency centres (public networks (networks public and primary emergency medical department staff education, rapid between sites care services services; rapid and redesign of transport, including local transport systems hospital systems emergency staff training) department training, hospital system redesign) Country: USA UK, USA USA, Germany USA, Canada, USA, Germany Germany Main Increase in Acceptable diagnostic Reduction in Reduced delays Telemedicine observations: proportion of accuracy of hospital delays both prehospital approach feasible people presenting ambulance staff in and door-to- and within and acceptable early with identifying stroke needle times hospital Modest increases in symptoms of patients consultation time suspected stroke (of 6–14 min) Absolute 5% NA 9% 4–17% 2–20% increase in rt-PA use*: Post- 6% 21% (in selected 11% 4–22% 2–25% intervention helicopter rt-PA use transport patients) References: 43 45,183,184 185 185,186 55,56 Summary of the findings of a range of studies describing changes in the chain of referral and treatment of acute stroke patients. None of these were randomized controlled trials. *% delivery after intervention minus % delivery before. rt-PA recombinant tissue plasminogen activator; NA, no data available. patient outcomes is unknown. Furthermore, the imple- these are complex systems and frequently other changes mentation and evaluation of ambulance protocols has were also made (e.g. moving CT scanning and near- often been included in an assessment of wider service patient biochemistry testing to within the emergency changes (e.g. implementation of public education pro- department) which may have influenced the outcome. grammes and establishment of stroke centres). Helicopter transport appears to be a feasible transport 17.5.4 Coordination of services option (Table 17.5) and there are claims that it can be 46 cost-effective. However, it is notable that even in reports One of the major challenges with any of the initiatives of helicopter transport of highly selected patients only a outlined above is how to coordinate them within the minority transported this way actually received rt-PA. broader picture of delivering stroke care, in particular, the need to be able to access expert advice and diagnostic imaging within a very short time. In general, two 17.5.3 Emergency department protocols approaches appear to have been applied to try and tackle Many enthusiasts have advocated hospital acute ‘stroke this problem: teams’ to increase the proportion of patients who are • designated stroke centres, a centralized approach eligible to receive thrombolytic therapy. Once again, where patients are transported to a single centre (‘hub’ there are no randomized trials of such strategies, al- model); though several observational studies have been described • telemedicine networks emphasizing a network of (Table 17.5). In general, these studies indicate that the expert advice usually delivered from some form of protocols were associated with reduced time delays specialized centre using teleconferencing technology within the emergency department, particularly if backed (‘hub and spoke’ model). up with educational programmes and rapid access to We shall consider in turn the rationale and evidence appropriate expert advice. However, it is also clear that for these different approaches. .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 912 912 Chapter 17 The organization of stroke services No data are available on whether patient outcomes 17.5.5 Designated stroke centres were improved by such changes but across these studies Following reports from the USA that without careful there was a reduction in symptom onset to treatment control of thrombolytic management, there is a high risk time and a modest increase in the delivery of rt-PA (Table of protocol violations and treatment complications, 26 17.5). There do not appear to be any analysis of the stroke quality improvement programmes have been impact of such changes on other aspects of stroke care developed. These were associated with a fall in protocol such as access to rehabilitation services and number of deviations from 50% to 19% and in symptomatic hae- patient transfers. morrhage from 16% to 6%. As a result of this and other In summary, designated stroke centres have gained experience, the Brain Attack Coalition (a group with ground in North America and Germany where the health- representatives from major professional and advocacy care system appears to be conducive to such systems of organizations involved in stroke care in the US) sug- care. In particular, stroke centres appear to have become gested criteria for the designation of stroke centres: 27,47 more established where there are more fragmented care • acute stroke teams; pathways (different providers delivering acute care and • stroke units or teams; rehabilitation) with relatively short lengths of stay. • written care protocols; There should be concern about the unnecessary transfer 54 • integrated emergency response systems; of unstable patients to stroke centres, since the majority • support services (in particular, the availability and inter- will not receive the intended treatment (thrombolysis) pretation of CT scans and rapid laboratory testing). and the stroke centre model (at least in the US) does not However, critics 48 have commented that the establish- emphasize the best evidence for service delivery through ment of stroke units was not given emphasis in the stroke units. 48 This, among other pressures, has raised Brain Attack Coalition recommendations. Stroke units, interest in alternative ways of delivering expert assess- for which there exists substantial randomized trial evid- ment and diagnosis. The most prevalent of these appears ence and which are seen as central to the Helsingborg to be telemedicine services. Declaration (with the target that organized stroke unit care should be available for all patients by the year Because stroke unit care is applicable to all patients, the 2005), were not considered an absolute requirement for overall impact on outcome is likely to be considerably primary stroke centres. By contrast, acute stroke teams, greater than that of tissue plasminogen activator – there which lack evidence of effectiveness, 49 were included is much more to acute stroke care than thrombolysis. among the key elements. In hospital surveys in the US, 50,51 stroke units were established in only a minority 17.5.6 Telemedicine services (networks) of centres. Similarly, limited access to stroke units was also seen in Canada and Australia. 52,53 Because stroke Telemedicine has been broadly defined as ‘the use of unit care is applicable to all patients, the overall impact telecommunication technology to provide medical on outcome is likely to be considerably greater than that information and services’. 55,56 In relation to stroke it can of tissue plasminogen activator – there is much more to be defined as ‘the process by which electronic visual and acute stroke care than thrombolysis. audio communications are used to provide diagnostic There are no randomized or controlled clinical trials and consultation support to practitioners at distant sites, evaluating designated stroke centres. However, four to assist in or directly deliver medical care to patients, prospective observational studies (Table 17.5) have and to enhance the skills and knowledge of distant reported improvements associated with establishing a medical care providers’. Telemedicine consultation has stroke centre. The establishment of such centres was been incorporated into a variety of aspects of modern complex but included: healthcare but only relatively recently into acute stroke • forming a ‘brain attack’ team; services. Some of the stated advantages 57 of such systems • establishing a care pathway (based on the five ‘Rs’ over traditional telephone consultations are: recognition, reaction, response, reveal and reperfu- • assisting in neurological assessment; sion; Table 17.4); • identification of neurological deterioration and provi- • establishing a protected bed in a stroke unit; sion of immediate feedback on treatment; • establishing public education campaigns; • identification of patients who do not require rt-PA; • raising the priority of stroke with emergency medical • to assess risk and benefit questions directly with the services; family without time delay; • ambulances bypassing other hospitals to get to a stroke • direct supervision of specialist procedures such as centre. delivery of rt-PA; .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 913 17.5 Interventions to improve access to early specialist assessment and treatment 913 • to monitor haemodynamic status and neurological Telemedicine (teleconsultation) services appear to examination prior to transport to a stroke centre; offer a promising approach to improve access to acute • ability of the stroke specialist to direct advice to the assessment and treatment, particularly where the health- transport team on the therapeutic needs of the patient. care economy does not favour stroke centres, where Proponents of telemedicine for acute stroke have populations are dispersed, and where local hospital pointed out that over half of hospitals in the US are stroke services can provide good post-acute stroke care located in rural areas and there is currently a shortage of (but not rapid assessment and treatment). A final point is neurologists and radiologists to provide expert advice that telemedicine services may allow workable medical across the whole country. 58 In addition, surveys of on-call rotas to be developed by allowing a large group neurologists in the US have shown less than half felt of experienced stroke specialists – who may be based in comfortable about giving intravenous rt-PA. 58 Tele- different hospitals – to cover a region. medicine also fits well within the concept of establishing primary and comprehensive stroke centres that provide 17.5.7 Transient ischaemic attack services a network of coverage to related sites. Lastly, tele- medicine may allow other sites within a network to par- Most of the debate has focused on services for patients ticipate in clinical trials; from consenting to follow-up, with stroke. However, substantial numbers of patients all the components for a quality stroke trial can poten- have a transient ischaemic attack (TIA) and have made a tially be achieved through telemedicine. Many of the full clinical recovery before they seek help. The risk of reported barriers are essentially man-made (for example, stroke after a TIA, or of recurrent stroke after a first reimbursement for services and the availability of suit- stroke, is very high in the early period and tails off later able equipment). (section 16.2.1). Thus, one has most to gain from starting There are no completed randomized trials of tele- secondary prevention as early as possible. Also, since the medicine services in acute stroke to provide any reliable accurate diagnosis of TIA depends on a good history, evidence about their impact on patient outcomes. How- it makes sense to assess patients as soon after the event ever, several key questions have been addressed using as possible. Traditionally, services have varied between other study designs. 55,56 These indicate (Table 17.5) that countries. telemedicine services in acute stroke appear to be: In the US it is relatively common for TIA patients to • Practical; experience from the US and Germany 55,56 be seen in the emergency room. In Scandinavia and has demonstrated the feasibility of establishing tele- Germany they are often admitted to hospital. In the UK medicine networks. These systems usually used some the move has been towards early specialist assessment form of teleconferencing facility linked to a stroke in outpatient clinics, so-called ‘one-stop’ neurovascular, centre. Technical failures were uncommon (0–4% of TIA or stroke clinics which are characterized by: consultations). • rapid access to a specialist opinion on TIAs, minor • Reliable; if telemedicine services are to be useful they strokes and episodes that may mimic them; must offer reliable diagnostic evaluation of both • streamlined access to the necessary investigations (which clinical features and brain imaging. The interrater requires close liaison with the radiology department); reliability of assessing neurological status (using the • close links with surgeons who can offer timely carotid NIH Stroke Scale, the Scandinavian Stroke Scale, or the endarterectomy. European Stroke Scale) was as good as with face-to-face These clinics ought to minimize unnecessary hospital assessment. 55 Two studies 59,60 reported complete agree- admission and delay in accessing specialist opinion, ment between telemedicine assessment of CT scan and investigation and treatment. However, they attract pa- conventional neuroradiology from the point of view tients with a wide range of other neurological conditions of eligibility for thrombolysis and major exclusions. (Table 17.3), so the clinician must have neurological • Feasible and acceptable; on average telemedicine training, or at least easy access to sound neurological consultations were about 10 min longer than conven- advice. 61,62 tional bedside consultations. Although this appears to be more than offset by reductions in the need for Rapid-access ‘one-stop’ neurovascular clinics should transfers to other hospitals, there were no explicit provide rapid clinical assessment of patients who may comparisons. Patient and clinician satisfaction with the telemedicine service was reported to be good. 55,56 have had a transient ischaemic attack or minor stroke, with streamlined and cost-effective investigations and • Associated with improved delivery of rt-PA; the propor- early intervention to reduce the risk of a serious tion of patients receiving rt-PA increased in association vascular event. with telemedicine services. .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 914 914 Chapter 17 The organization of stroke services However, the recognition that TIA patients may have there are some uncertainties about whether current a 20% risk of early stroke occurrence within the first treatments prevent early recurrence, most would argue 2 weeks, with the majority in the first day, 30 has raised that a very high early risk of recurrence (e.g. greater concerns about delays in assessment and diagnosis even than 20% in the first week) would warrant hospital where rapid-access clinics are available. In addition, the admission. Indeed, to deny admission to similar high- effectiveness of carotid endarterectomy is greatest if risk individuals with an acute coronary syndrome carried out within 2 weeks of symptom onset. 34 These would be considered unacceptable. developments challenge us to diagnose and treat TIA and The next few years are likely to see a greatly increased minor stroke far more quickly. In general, three appro- move towards rapid assessment and treatment of aches appear feasible (although none have been tested TIA and minor stroke patients and the solutions adopted in clinical trials). will vary depending on the region. However, it is crucial • Outpatient management: with a daily stroke clinic, it is that we have good evidence about what forms of possible to see most TIA and minor stroke patients early assessment and treatment effectively reduce early within 48 h of symptom onset. 34 However, this is recurrence. resource-intensive 63 and presents challenges for the organization of investigations. • Assessment within emergency departments: if the emergency departments have rapid access to investiga- tions, including ultrasound and CT scanning, then 17.6 Organized inpatient (stroke unit) care it may be possible to provide a rapid turnaround of investigation and early secondary prevention. This approach, which appears to be common in the US, 64 The types of stroke services that hospitals traditionally carries the risk of inappropriate investigation and treat- provided have varied from place to place reflecting local ment if non-specialist staff are planning patient care. interests and politics. However, there is now widespread • Hospital admission: in some countries (e.g. Scand- acceptance that hospital-based stroke services should be inavia) hospital admission for TIA is an established organized within stroke units. 66,67 Much of the evidence approach, however it is not common elsewhere for this comes from a systematic review of clinical trials (including the UK). Recently a scoring system has been that compared the outcome for stroke patients cared developed which may identify individuals at highest for in a specialist stroke unit with those cared for in risk of early recurrence. 31,65 If a subgroup of TIA and general wards. Patients managed in stroke units are more minor stroke patients really are at extremely high risk likely to survive, return home and regain independence of early recurrence, then a short period of hospital (Table 17.6). Stroke units may also improve patients’ admission would be justified provided we can provide quality of life, and improvements in outcome may per- rapid assessment and effective treatment. Although sist for several years. 68 Table 17.6 Summary of patient outcomes in the stroke unit trials. Stroke Conventional Odds ratio b Absolute risk Difference in unit a care a (95% CI) difference c number of (95% CI) outcomes per 1000 admitted d Home (independent) 45% 40% 1.25 (1.12, 1.40) 0.05 (0.02, 0.07) 50 Home (dependent) 17% 15% 1.16 (0.88, 1.53) 0.01 (–0.01, 0.03) 10 Institutional care 14% 16% 0.84 (0.72, 0.98) –0.02 (0.00, 0.04) –20 Dead 23% 28% 0.80 (0.70, 0.91) –0.04 (–0.02, –0.06) –40 a The proportion (%) of patients with various outcomes at the end of scheduled follow-up (median 1 year) in the randomized trials of stroke unit care vs conventional care. b The odds ratio for that outcome (95% confidence interval). c The proportion of outcomes achieved (+) or avoided (–) with stroke unit care. d The number of outcomes achieved (+) or avoided (–) for every 1000 patients cared for in a stroke unit, assuming the absolute risk of an outcome in the population is similar to that in the trials. Figures based on data from 31 trials (6900 patients). 68 .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 915 17.6 Organized inpatient (stroke unit) care 915 • Ward base: effective stroke units have usually been Stroke patients managed in a stroke unit are more based in a discrete ward with dedicated nursing staff. likely to survive, return home and regain Mobile stroke teams do not appear to improve patient independence than those managed in a general outcomes. 70 medical or even in a neurology ward. • Specialist staffing: they have been staffed by medical, These trials were testing much more heterogeneous nursing and therapy staff with a specialist interest and interventions than is usual in drug trials, where the inter- expertise in stroke and/or rehabilitation. vention is precisely defined in terms of the chemical, the • Multidisciplinary team working: they have always dose and the timing. Because of this heterogeneity of the included good multidisciplinary communication intervention – but not of the results – it is sometimes (defined as a formal meeting of all staff once per week difficult to generalize from the stroke unit overview, to plan the management of individual patients). and some important questions remain when applying • Education and training: they have incorporated pro- the results to everyday clinical practice. grammes of education and training for staff and provision of information for patients and carers. Although many stroke units were not described in 17.6.1 What is stroke unit care? detail, several consistent features of the process of care Although comprehensive stroke unit care is a complex have been described 66,69 (Table 17.7). This typically did and multi-faceted intervention, the key components are not depend on high-tech facilities but did implement a reasonably well described 69 (Table 17.7): systematic approach to care that incorporated: Table 17.7 Outline of comprehensive Structure stroke unit care. Geographically discrete ward Medical staff with specialist interest in stroke and rehabilitation Nursing staff with specialist interest in stroke and rehabilitation Multidisciplinary staffing (nursing, medical, physiotherapy, occupational therapy, speech therapy, social work) Coordination of care Regular multidisciplinary team meetings (formal meeting of all staff once weekly, informal meetings 2–3× per week) Close linking of nursing and multidisciplinary team care Educational programmes for staff Assessment and monitoring Rapid admission to stroke unit Medical history and examination Routine investigations (biochemistry, haematology, ECG, CT scanning) Further selective investigations (carotid ultrasound, echocardiogram, MR scanning) Nursing assessments (vital signs, general care needs, swallow test, fluid balance, pressure areas, neurological monitoring) Therapy assessments of impairments and disability Early management Careful management of food and fluids Pyrexia management, paracetamol for pyrexia, antibiotic for suspected infection Hypoxia management, oxygen if hypoxia, drowsiness or cardiorespiratory disease Glycaemic management, insulin for hyperglycaemia Early mobilization, up to sit, stand and walk as soon as possible Careful positioning and handling Pressure area care Avoid urinary catheterization if possible Ongoing multidisciplinary rehabilitation Early goal setting Early involvement of carers in rehabilitation Provision of information to patients and carers Discharge planning Early assessment of discharge needs Discharge plan involving patient and carers .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 916 916 Chapter 17 The organization of stroke services • Careful assessment and monitoring – of medical, nurs- and a major source of dissatisfaction in non-specialized ing and therapy needs. wards, may in part explain the success of stroke units, • Early active management – incorporating management but this would be difficult, if not impossible, to prove. 72,73 of food and fluids, control of pyrexia, hypoxia, hyper- glycaemia, early mobilization, careful positioning and 17.6.2 Types of stroke unit handling, and avoidance of urinary catheterization. • Ongoing multidisciplinary rehabilitation – with early Although the basic principles of stroke unit care are goal setting, early involvement of carers in rehabilita- reasonably well described they have been delivered in a tion, and provision of information to patients and carers. variety of ways, and the term ‘stroke unit’ means differ- This also included early planning of discharge needs. ent things to different people. So it is important to define Many of these processes of care will come as no sur- our terms (Table 17.8). Inevitably, the question arises prise to those experienced in stroke care but recent audits as to which type of unit is most effective. This is very support concerns that many are not routinely provided. 71 difficult to answer with confidence and attempts to do Beyond describing the basic components of stroke so 74 are prone to bias. They do indicate (Table 17.9) that unit care, it is difficult to determine whether the effect- effective stroke units have usually provided organized iveness of the stroke units is due to the total package of multidisciplinary care for a reasonable period of time. care, or to particular components. Some of the individ- We shall discuss this in more detail. ual components can be evaluated in randomized trials (e.g. guidelines for prevention of deep venous thrombo- Effective models of organized inpatient (stroke unit) sis, early mobilization, intensive physiotherapy), and care are able to provide specialist multidisciplinary these trials can be reviewed systematically to provide care for most of the inpatient pathway of care. reliable data, but some of the less well-defined compon- ents and any possible synergy between them is much Acute stroke units less amenable to such assessment. For example, better communication between health professionals, stroke ‘Acute’ refers to the policy of rapid admission of the patients and their carers, which is so often inadequate stroke patient to the stroke unit. Admitting all acute Table 17.8 Classification of organized inpatient (stroke unit) care. Type Philosophy Patient MDT base Timing of Timing of discharge Type of care of care group admission from a stroke unit Acute Acute care; life Stroke Ward Acute (hours) Early (3–7 days) Acute medical & nursing care (intensive) support (with high staffing levels) stroke unit Acute (semi- Acute care; close Stroke Ward Acute (hours) Early (3–7 days) Acute medical & nursing care intensive) monitoring Monitoring & management of stroke unit physiological variables Comprehensive Acute care Stroke Ward Acute (hours) Later (days – weeks) Acute medical & nursing care stroke unit and multi- some referral to Non-intensive management of disciplinary specialist physiological variables rehabilitation rehabilitation Early active multidisciplinary rehabilitation Rehabilitation Multidisciplinary Stroke Ward Delayed Later (weeks) Multidisciplinary stroke unit rehabilitation (days) rehabilitation Mixed Multidisciplinary Stroke & Ward Early (hours– Later (weeks) Multidisciplinary rehabilitation rehabilitation other days) rehabilitation unit disabling illness Mobile stroke Acute care and/or Stroke Mobile Early (hours– Later (weeks) Acute medical care and/or team multidisciplinary (no ward) days) multidisciplinary rehabilitation base rehabilitation; no specialist nursing input This table summarizes, in broad terms, the characteristics of different types of stroke unit. MDT multidisciplinary team. .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 917 17.6 Organized inpatient (stroke unit) care 917 Table 17.9 Effectiveness of different types of stroke unit: summary of direct and indirect comparisons of different models of stroke unit care. Type of stroke Number of Control service comparator Estimated odds ratio for Statistical unit evaluated comparisons in trials death or dependency significance (subjects) (95% CI) compared with a general ward Acute (intensive) stroke unit None No data No data No data Acute (semi-intensive) stroke unit 3 (530) Mixed rehabilitation unit, 0.77 (0.53–1.12) P = 0.17 comprehensive stroke unit Comprehensive stroke unit 10 (3010) General ward, mobile 0.89 (0.79–0.99) P = 0.03 stroke team Rehabilitation stroke unit 7 (870) General ward, mixed 0.85 (0.77–0.93) P = 0.0007 rehabilitation unit Mixed rehabilitation unit 9 (1170) General ward, rehabilitation 0.87 (0.79–0.96) P = 0.007 stroke unit Mobile stroke team 5 (1000) General ward, comprehensive 0.98 (0.91–1.10) P = 0.5 stroke unit The results of a series of comparisons of different types of stroke unit care. 74 This approach attempts to provide a standard measure of stroke unit effect where trials have used different control groups. Data analysis used methods 185 of combining direct comparisons (trials that directly compare a stroke unit vs general ward) with indirect comparisons (e.g. the effect of a stroke unit vs general ward is inferred from trials of stroke unit vs mobile team and mobile team vs general ward). Data are presented as the summary estimate of the odds ratio (95% confidence interval) for death or dependency of stroke unit vs general ward. Please note that these estimates are based in part on indirect comparisons of treatments and are potentially subject to bias (e.g. from different patient groups being recruited into different trials). stroke patients directly into a unit makes the introduc- improve patient outcome. The approach has come tion of assessment protocols easier, allows expertise to be under more scrutiny recently, presumably because it focused, and will certainly facilitate the large random- inevitably requires extra resources due to the high ized trials of acute interventions that are needed to staffing levels and expensive equipment. Also, there identify effective treatments. 75 It also facilitates a policy is currently little evidence from randomized trials of active early mobilization (section 11.11), hydration that the various individual interventions employed (section 11.18), control of temperature (section 11.12), are effective. Thus, although stroke intensive care avoidance of hypoxia (section 11.2) and large changes units may help, we need randomized trials to evaluate in blood pressure (section 11.7). Although there are no them. reliable data from randomized trials demonstrating that • Semi-intensive units are similar to coronary care units any one of these interventions improves outcomes, they where monitoring and intervention focus on physio- are supported by a reasonable theoretical rationale and logical variables but not life support. There have been some observational data. 76–79 three small clinical trials of semi-intensive units that In some centres, particularly in North America and have rather inconclusive findings. One 85 indicated Germany, there has been a vogue for admitting stroke no benefit over care in a less intensive setting while patients to ward areas with facilities for intensive mon- two others 86 have indicated a potential benefit. As itoring of physiological functions (cardiac, respiratory mentioned above, there is limited evidence for the and neurological). Interventions are introduced to cor- individual elements of monitoring and intervention rect these abnormalities (e.g. raised intracranial pressure, although one trial 79 has indicated that intensive moni- systemic hypertension), in the belief that this will toring can reduce early neurological deterioration after improve outcome. 80 Broadly speaking two approaches stroke. have been described: • Intensive care units which can offer all monitoring Comprehensive stroke units (including intracranial monitoring) and life support options (e.g. respiratory support). There have been Perhaps the most successfully implemented model has several non-randomized studies of stroke intensive been the comprehensive stroke unit, which admits care units, 81–84 but there is no good evidence that these patients acutely and then provides at least a few weeks .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 918 918 Chapter 17 The organization of stroke services Study or sub-category Odds ratio Odds ratio 95% Cl 95% Cl Comprehensive stroke ward vs general medical ward Akershus Athens Beijing Edinburgh Goteborg-Sahlgren Joinville Perth Trondheim Umea Subtotal (95% CI) 0.83 (0.71–0.97) Test for overall effect: Z = 2.39 (P = 0.02) Rehabilitation stroke ward vs general medical ward Dover (GMVV) Nottingham (GMVV) Orpington-1993 (GMVV) Orpington-1995 Subtotal (95% Cl) 0.83 (0.57–1.23) Test for overall effect: Z = 0.92 (P = 0.36) Mobile stroke team vs general medical ward Cape Town Fig. 17.2 The results of a systematic Manchester review of randomized trials testing the Montreal effectiveness of stroke unit care compared Uppsala with a general medical ward. The result of Subtotal (95% Cl) 0.96 (0.69–1.34) Test for overall effect: Z = 0.22 (P = 0.82) each individual trial, expressed as the odds ratio, is represented by a purple Mixed rehabilitation ward vs general medical ward box with a horizontal line indicating the Birmingham 95% confidence interval. The block size Helsinki IIIinois is proportional to the amount of Kuopio information in the trial. An odds ratio to New York the left of the vertical line indicates that Newcastle the odds of the outcome (in this case Subtotal (95% Cl) 0.65 (0.47–0.90) events refer to death or dependency Test for overall effect: Z = 2.57 (P = 0.01) between 6 and 12 months after Total (95% Cl) 0.82 (0.73–0.92) randomization) is less with stroke unit Total events: 1266 (Treatment), 1341 (Control) care than care in general wards. Estimates Test for overall effect: Z = 3.22 (P = 0.001) based on an overview of all the trials 0.1 0.2 0.5 1 2 5 10 are represented by solid diamonds. Favours Favours Reproduced with permission from the stroke unit general ward Stroke Unit Trialists’ Collaboration. 88 of rehabilitation. Such a model, which is widespread in medium-sized hospitals where one team can manage most Norway and Sweden, is supported by the largest group stroke patients within one unit. In practice, although of clinical trials included in the systematic review these units provide most care for most patients, referral (Fig. 17.2) and results from a national stroke register of some patients with ongoing complex rehabilitation in Sweden. 87 Although we believe that rehabilitation needs to other rehabilitation services is common. should start on the day of the stroke, such units do pre- sent some practical challenges. For example, very sick Rehabilitation stroke units stroke patients might require care that would disrupt a rehabilitation unit, at a time when they are unlikely to Several trials have indicated benefit from rehabilitation benefit from a rehabilitation environment. However, units that admit patients a few days after stroke onset models that separate acute assessment and rehabilita- and continue rehabilitation for several weeks. These tion areas may disorientate some patients (and their trials have inevitably examined a more selected patient families) and can compromise continuity of care. The group who are stable enough for that environment and comprehensive unit approach appears to be well suited to have ongoing rehabilitation needs. .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 919 17.6 Organized inpatient (stroke unit) care 919 main role for stroke teams is probably to provide out- Mixed rehabilitation units reach care for those unable to get into the stroke unit, The meta-analysis of organized inpatient (stroke unit) and to prioritize admission to the stroke unit. care predominantly included trials comparing organized care within stroke-specific units with care in general The main advantage of caring for stroke patients in wards (general medicine or general neurology). However, one place is that nurses can play a major role in the some trials explored the impact of organizing stroke rehabilitation process. care within generic rehabilitation services (e.g. geriatric medicine or neurological rehabilitation services); patients 17.6.3 Who should be admitted to the achieve better outcomes in mixed rehabilitation units stroke unit? than in general wards without multidisciplinary team care. 88 Comparisons with stroke-specific units indicate Stratification of patients within the meta-analysis by a trend towards better outcomes in stroke-specific units stroke severity showed that patients with mild, moderate (Table 17.9) but there are insufficient data to determine and severe strokes are all likely to benefit from stroke whether stroke-specific units are clearly more effective. unit care. 88,89 In terms of absolute outcomes more severe Mixed rehabilitation units may have a role in smaller patients have the greatest survival advantage from stroke hospitals or very specialized services (e.g. young adult unit care but milder patients gained more in terms of rehabilitation) where there may be too few stroke regaining independence. Data from non-randomized, patients to make stroke-specific services viable. How- and thus potentially biased, comparisons of outcomes ever, stroke-specific services certainly allow more spe- following admission to a stroke unit or general care sup- cialization among the team members, which enhances port the notion that stroke unit care benefits unselected the educational and research potential of the service. stroke patients. 87,90–92 Of course, bed shortages may force Training of junior doctors and other staff, which might staff to make triage decisions. Many of the serious early suffer if all the stroke patients were managed by a single complications of stroke (including early neurological team, can be protected and probably improved by deterioration) are more common with severe stroke and organizing rotation of staff through the unit. Moreover, may respond to stroke unit care. Milder stroke patients stroke specialists are more likely to be enthusiastic about (e.g. those who are mobile) probably have less to lose teaching students and staff about strokes than general- from not being admitted to a stroke unit. ists or those with another specialist interest. There is little evidence that patients of particular ages gain more or less from care in a stroke unit. 88,93 There are good reasons to believe that older stroke patients may be Mobile stroke teams at higher risk of some complications, and therefore have Overall the trials in the meta-analysis indicated that a more to gain from admission to a stroke unit. Although stroke team working across several general wards may we think that needs rather than age should dictate where improve aspects of the processes of care (e.g. access to and by whom patients are managed, local conditions specialist assessments) but cannot achieve patient out- will often dictate which service is the best option. For comes as good as those of a team based in a stroke example, where an age-related geriatric service (e.g. unit. 24,49,70 Probably the most important advantage of one that admits any patient who is over 75 years old, having the patients in one location is that the nursing whatever the problem) already provides effective stroke staff can play a greater role in the rehabilitation process. rehabilitation, there may be a case for adding a new When patients are scattered, it is more difficult to incor- rehabilitation service for younger stroke patients rather porate the essential role of nurses. Also, stroke patients than dismantling the current service. Professionals and managed in acute general areas have to compete for patients’ families are often concerned that younger nursing time with patients who may be perceived to patients’ morale will suffer if they are treated in a ward have more urgent needs (e.g. chest pain). Stroke patients with mainly older patients, although this concern is by may, for example, need regular toileting to maintain no means universal. continence and thus dignity. These aspects of care are very important, but can be seen as less urgent. A geo- 17.6.4 How long should patients remain graphically defined stroke unit removes this competition in the unit? for nursing time, and allows the nurses to take on a new role – not just as carers, but also as facilitators of patient Some units, particularly those that admit patients acutely, independence – and to continue therapy (directed by define maximum lengths of stay. It seems to us that the specialist therapists) throughout the 24-h period. The only reason to do this is to allow admission of new cases. .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 920 920 Chapter 17 The organization of stroke services If the unit is of sufficient size for the population’s needs, A survey in one of our own medical units, which works flexibly and is efficient in discharging patients, admits between 200 and 250 stroke patients each year, then a defined maximum length of stay should not be demonstrated that the number of stroke inpatients needed. If one does insist on a maximum length of on any one day varied between 9 and 35 over a year. stay, one must ensure that there are facilities and staff Therefore, stroke services must be flexible enough to elsewhere to deliver appropriate continuing care (e.g. cope with large fluctuations in the numbers and types ‘slower-stream’ rehabilitation facilities; 93 and that pa- of patients referred. tients are not left to languish on an acute medical ward. One might argue that patients who are no longer improving, but are having to wait for placement in the 17.6.6 Who should staff and run the stroke unit? community or an institution, should not be kept in a stroke unit. However, for some individuals the unit may The units included in the systematic review were run by offer the best environment to maintain any functional geriatricians, neurologists, general (internal) physicians improvement already gained. Moves under these cir- and rehabilitationists. Indirect comparisons of the bene- cumstances should only be considered where beds are fits of units run by different specialist groups did not limited and patients who are judged likely to gain more show any significant differences. 66,88 We believe that from the unit environment are waiting to be admitted. whoever is responsible should have the necessary know- ledge, training and above all enthusiasm to take on the task. The most appropriate professional group will vary 17.6.5 How large should a stroke unit be? from place to place. For example, in the Netherlands, Age-specific and sex-specific stroke incidence data and practically all acute stroke patients are managed by details of the hospital catchment population, along with neurologists, while in the UK most are managed, at least hospital activity data, should allow an estimate of the initially, by general physicians and geriatricians. 71,96 number of patients who are likely to require admission to British neurologists may have the knowledge and training hospital each year (section 17.11.3). Unfortunately, there to diagnose and investigate stroke patients, but unfort- may be variations due simply to chance or the season unately most do not have the time or interest, support of the year. Although consistent seasonal differences in staff, access to beds, or training in rehabilitation to run a the incidence of stroke have not been demonstrated in comprehensive stroke service without help from other community-based studies, at least in temperate regions, specialists. In the UK, geriatricians are often in the best there is an excess of hospital admissions and stroke position to take a leading role, although most need extra deaths during the winter. 94,95 Of course, this may simply training in neurology and the active participation of reflect referral bias and a higher case fatality during cold their local neurologist – who can very usefully contribute weather. Whatever the explanation for any seasonal to diagnosis and management of patients, especially variation, it does cause difficulties when planning stroke those with unusual causes of stroke, patients with ‘funny services. Prior to the development of a stroke unit, a turns’ and the many and varied neurological problems survey in one of our own medical units, which admitted that arise in inpatient and outpatient stroke care between 200 and 250 stroke patients each year, demon- (Table 17.3). The advent of thrombolysis requires addi- strated that the number of stroke inpatients on any tional training, expertise and support in assessment one day varied between 9 and 35 over a year. Therefore, of the ‘brain attack’ patient and interpreting acute whatever organization one sets up to manage these imaging. In the UK training for stroke specialists is patients, it must be flexible enough to cope with large being implemented and an increasing number of hospi- fluctuations in their numbers. The unit should be able to tals are appointing specialist stroke physicians (from a accommodate different proportions of men and women, variety of specialist backgrounds) to coordinate stroke as the proportions are bound to fluctuate. The inevitable services. limit on the number of beds may be managed by ensur- ing that the stroke unit is part of a larger area, into which 17.6.7 Planning a stroke unit development it can expand with demand and then contract again. Such arrangements also mean that, at times, non-stroke The good evidence for the effectiveness of stroke units patients are cared for in the stroke unit. Inevitably, there means that in many countries their development is are times when resources are not adequate to meet all the supported by national initiatives. However, the devel- needs of the patients, and it is then important for the opment of such units is often resisted by those who team to support its members in the difficult task of prior- perceive them as a threat. Some points are worth con- itizing – in other words rationing – care. sidering in local discussions. .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 921 17.7 Transfer from hospital to community 921 • Stroke units generally make more efficient use of of patients following thrombolysis in an emergency existing staff and beds, and so may eventually even department setting. This option has a good evidence increase the resources available to other specialties. base to support it and would be most appropriate for • De-skilling of their junior medical, nursing and para- small to medium-sized hospitals (e.g. 100–300 stroke medical staff can be overcome by rotating staff and patients per year). students through the unit. • Acute stroke unit and rehabilitation stroke unit: in larger • Adopting an evolutionary approach to developing the hospitals, it may be difficult to provide all stroke unit service may be encounter less resistance – for example, beds in one place. In this circumstance, a combination one might introduce a stroke assessment protocol of acute stroke unit and rehabilitation stroke unit has before trying to set up an acute assessment area, or a often been adopted. In theory this should provide stroke team working on the general medical wards comprehensive stroke unit care but is largely untested before trying to set up a geographically defined stroke in clinical trials. If this model is adopted, we suggest unit. Stroke units are not just a research intervention, that the components of comprehensive stroke unit they have been implemented in a range of settings care must be provided throughout the patient journey with improvements in outcome. 71,87,91,92 (e.g. early mobilization and early multidisciplinary • One can try to influence the local organizations that planning applies in the acute stroke unit, patients in fund healthcare (i.e. health authorities and general or the rehabilitation stroke unit should have access to family practitioners in the UK; health insurers in other acute medical care). In addition, the patient and family countries) to exert pressure for change, since they are should experience a continuous process of care rather generally keen to fund services for which there is scien- than discontinuity between components of the service. tific evidence of efficacy (curiously, and irritatingly for • Mixed rehabilitation unit: in smaller hospitals with dis- those of us concerned with stroke, neither coronary care persed populations (e.g. less than 100 stroke admissions units nor regional oncology services have been nearly per year), it may be appropriate to have protocols of as well evaluated as stroke units, and yet their utility is delivery of acute care in general wards and to provide said to be ‘obvious’ and they are widely encouraged). rehabilitation services in a mixed rehabilitation setting. Telemedicine services (section 17.5.6) may offer options Curiously, and irritatingly for those of us concerned for accessing expert advice from larger stroke centres. with stroke, neither coronary care units nor regional In a stroke unit development it is important to com- oncology services have been nearly as well evaluated pare the characteristics of your planned service with as stroke units, and yet their utility is said to be those in the stroke unit trials to ensure you are delivering ‘obvious’ and they are widely encouraged. care that is, as far as possible, evidence-based (Tables 17.8 and 17.9). There are likely to be significant pressures (other than evidence of effectiveness) that will influence local service delivery, since its structure must be tailored to local needs, resources, geography, people and politics. One needs to consider what is the best way to deliver compre- 17.7 Transfer from hospital to community hensive stroke care within the constraints of the local circumstances. We suggest three factors come into play: • evidence that a particular model of stroke unit is effect- One of the main areas of concern to patients, and more ive (Table 17.9); particularly to carers, is the organization (or rather the • the ability of the stroke unit model to deliver all lack of organization) of hospital discharge. 72,97 One can aspects of care required; well understand their concern. One day patients are • the ability of the stroke unit to provide for the broadest being cared for in hospital by a team of professionals, and group of stroke patients (i.e. meet the needs of the the next they are at home and the responsibility of the stroke population). carers. A number of things can be attempted to try and The following list of options outline some common reduce the stress of the transition from hospital to home: solutions: • Provide adequate information (section 17.10.1) and • Comprehensive stroke unit: where possible, the com- train the carers while the patient is in hospital, for ex- prehensive stroke unit, which combines both acute ample invite the carers to therapy sessions and involve care and rehabilitation for the majority of patients, them in the patient’s care on the unit. Unfortunately, is probably the preferred option. This could include trials of information provision and patient education the delivery of thrombolysis or take over the care do not provide clear evidence to guide practice. .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 922 922 Chapter 17 The organization of stroke services • Predischarge home visits with the patient and one or cost-effective, reducing the overall cost of caring 99 and more members of the team to ensure that the home is the subject of further research. environment is tailored to the patient’s needs. Also, In some circumstances, it may be appropriate to further informal visits home, initially for a day and then break down the boundary between hospital and home graduating to overnight and weekend stays, allow the by organizing for the patient to attend a day hospital or patient and carer to gain confidence, identify potential outpatient department regularly to be reviewed medic- problems and help maintain morale. Although this is a ally, or to receive further rehabilitation input. well-established approach we could not identify any clinical trials of such policies. 17.7.1 Early supported discharge services • Predischarge case conferences allow the patient and carer to meet with the hospital-based team and any Partly in response to the well-reported limitations of professionals who are to be involved in their care in conventional hospital discharge arrangements, a num- the community. Once again this is a well-established ber of services have been developed to try and improve approach for which we could not identify any clinical the transition between hospital and community, in trials. particular, early supported discharge services, which aim • Clear guidelines about who to contact in the event to accelerate discharge home from hospital but provide of problems. General (family) practitioners, or one of more continuity of rehabilitation in the home setting. their team, are the ideal points of contact, but they To date, 12 randomized trials have tested this approach can only fulfill this role if adequately briefed before to care in a variety of settings around the world. 70,100 the patient’s discharge. It follows that detailed records Most were based around a small multidisciplinary team of the patient’s problems, and the plans for support in of physiotherapy, occupational therapy, nursing and the community, must be relayed to those expected to assistant staff – with input from medical, speech and monitor the home situation. language therapy and social work staff. The teams were • A programme of training carers to manage their new either hospital-based (and went out to the patient’s role has been tested in one moderately large random- home) or community-based (and came into hospital ized trial. 98 This involved training by stroke unit staff to recruit patients). All incorporated regular multi- about stroke plus practical caring skills. The brief pro- disciplinary team meetings to plan patient care. A typical gramme also continued after discharge. This approach pathway of care is shown in Table 17.10. was surprisingly effective – trained carers reported less Typically these services input for up to 3 months but caregiver burden, anxiety or depression and had a in some cases this might be shorter with handover to higher quality of life, while patients in the carer train- other community services. 70 ing group experienced less anxiety and depression Even compared with high-quality care from a hospital- and a better quality of life. The intervention was also based stroke unit, an early supported discharge team Table 17.10 Illustrative pathway of care Hospital admission Early identification of eligible patients in hospital with early supported discharge service. Early assessment by a ‘key worker’ from the early supported discharge team (one individual who supervised that patient’s care) Assessment of home needs through a home visit (with or without the patient present) Identification of recovery goals with patient and carer Plan discharge home Discharge home Discharge home with very early input (within 24 h) by members of the early supported discharge team Implement rehabilitation plan Continuing rehabilitation in the home setting (up to 4–5 days per week if necessary) Access relevant services Multidisciplinary team review of progress Negotiated withdrawal of the team as recovery goals are achieved Planned discharge from the service with later follow-up and review Discharge from early supported discharge team .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 923 17.8 Continuing rehabilitation and reintegration back to normal life 923 Study or sub-category Odds ratio Odds ratio 95% Cl 95% Cl ESD team coordination and delivery Adelaide Belfast London Fig. 17.3 The results of a systematic Manchester review of randomized trials testing Montreal the effectiveness of early supported Newcastle discharge (ESD) services compared with Stockholm conventional discharge services. The Subtotal (95% Cl) 0.71 (0.53–0.94) result of each individual trial, expressed Test for overall effect: Z = 2.40 (P = 0.02) as the odds ratio, is represented by a ESD team coordination purple box with a horizontal line Oslo indicating the 95% confidence interval. Trondheim Subtotal (95% Cl) The box size is proportional to the 0.68 (0.46–1.01) Test for overall effect: Z = 1.89 (P = 0.06) amount of information in the trial. An odds ratio to the left of the vertical line No ESD team Akershus indicates that the odds of the outcome Bangkok (in this case events refer to death or Subtotal (95% Cl) 1.23 (0.79–1.91) dependency between 6 and 12 months Test for overall effect: Z = 0.92 (P = 0.36) after randomization) is less with the ESD Total (95% Cl) 0.79 (0.64–0.97) service than with conventional care. Total events: 359 (Treatment), 398 (Control) The estimates based on an overview of all Test for overall effect: Z = 2.28 (P = 0.02) the trials are represented by diamonds. Reproduced with permission from the 0.1 0.2 0.5 1 2 5 10 Early Supported Discharge Trialists. 100 Favours ESD service Favours control could not only accelerate discharge home (with an aver- discharge services appear to be an essential component age reduction in length of stay of 7 days), but could also of a truly comprehensive stroke service and should par- result in the patient having a greater chance of remain- ticularly target patients with mild to moderately severe ing at home and regaining independence (Fig. 17.3). strokes. Overall, for every 100 patients randomized to early supported discharge services, an extra five (95% CI 1–9) Early supported discharge services – based on a specialist remained at home and an extra five (95% CI 1–10) were multidisciplinary team – can help accelerate discharge independent at 6–12 months after the stroke. Good home and improve the longer-term recovery of selected results were most likely with a well-resourced, coordin- stroke patients without incurring excessive costs. ated multidisciplinary supported discharge team and if patients were recruited with mild to moderate stroke severity (i.e. those who achieve a Barthel Index of at least 10/20 in the first week – equivalent to rapidly regaining standing balance). There is a suggestion that such serv- 17.8 Continuing rehabilitation and ices may not work as well in more dispersed rural popula- reintegration back to normal life tions 101 but this requires confirmation. Economic analyses 100 indicate that the potential saving in hospital costs (reduced hospital bed-days) is greater Even where stroke patients have received good care in than the additional costs of community rehabilitation. hospital and around the discharge period, they may still In practice it is often difficult to release such costs, but have difficulty maintaining independence and reinte- at the very least early supported discharge services gration back to normal life (Table 17.11). At this stage of appear to offer a way of both improving patient care the patient journey, services are often very variable and and optimizing the use of a limited number of hospital may be completely non-existent. This probably reflects beds. In addition to the ‘harder’ outcomes above, it the diversity of approaches in different countries but also is also noteworthy that patient and carers allocated to a limited evidence base to suggest that effective inter- early supported discharge services were more likely to ventions really can improve recovery. In general terms report satisfaction with their services. 70 Early supported two broad approaches have been tested at this late stage: .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 924 924 Chapter 17 The organization of stroke services Table 17.11 Common problems that arise late after a stroke, • Stroke liaison worker services (provided by stroke often when the patient is no longer in hospital. nurses, family support workers or specialist social workers and primarily aiming to improve participa- Patient tion in normal living and quality of life). In practice Deteriorating function due to: these have included a mix of interventions which – inactivity 102 could deliver a programme of rehabilitation (e.g. ) or – progressing comorbidities respond to identified problems. 103 – lack of continuing rehabilitation input? There is considerable variation within these broad – depression or anxiety, even agoraphobia (section 11.31) – overprotection from the carer approaches but they do provide a framework for consid- Social isolation ering the evidence. Financial difficulties (section 11.33.6) Sexual dysfunction (section 11.33.5) 17.8.1 Therapy-based rehabilitation services Undetected rise in blood pressure Central post-stroke pain (section 11.23) In a systematic review 104 of therapy-based rehabilitation Carer (section 11.34) services, most of the trials assessed input in the patient’s Physical ill-health due to the strain of caring home by an occupational therapist although the review Depression or anxiety also included physiotherapy or multidisciplinary team Poor relationship with patient because of personality change services (such as community teams and day hospital Social isolation because unable to get out to meet people services). Therapy-based rehabilitation services (when Financial difficulties compared with no routine intervention) helped prevent stroke patients deteriorating in their ability to carry out • Therapy-based rehabilitation services (provided by phy- activities of daily living (ADL) (Fig. 17.4) and improved siotherapy, occupational therapy or multidisciplinary ADL scores. Therefore, even relatively late after stroke staff and primarily aiming to increase activities in daily onset (several months) patients may gain from input living). In practice this might include a range of task- from a therapist. What is less clear is the absolute bene- related interventions aiming to improve mobility, act- fit likely to be achieved and the cost-effectiveness of ivities of daily living or specific tasks such as dressing. these services. However they do indicate that within a Study or sub-category Odds ratio Odds ratio 95% Cl 95% Cl Mixed service Hong Kong Fig 17.4 The results of a systematic London review of randomized trials testing the Philadelphia South London effectiveness of therapy-based Subtotal (95% Cl) 0.73 (0.42–1.27) rehabilitation (provided to patients living Test for overall effect: Z = 1.12 (P = 0.26) at home) compared with conventional services (usually no routine input). The Physiotherapy Copenhagen result of each individual trial, expressed Kansas as the odds ratio, is represented by a Subtotal (95% Cl) 0.66 (0.23–1.90) purple box with a horizontal line Test for overall effect: Z = 0.77 (P = 0.44) indicating the 95% confidence interval. Occupational therapy The block size is proportional to the Cardiff amount of information in the trial. An Glasgow odds ratio to the left of the vertical line Nottingham 1995 indicates that the odds of the outcome (in Nottingham 1997 this case events refer to poor outcomes – Nottingham 1999 TOTAl death, deterioration in activities of daily Subtotal (95% Cl) 0.73 (0.55–0.96) living or dependency between 6 and Test for overall effect: Z = 2.23 (P = 0.03) 12 months after randomization) is less with the therapy service than with Total (95% Cl) 0.72 (0.57–0.92) Total events: 245 (Treatment), 218 (Control) conventional care. Estimate based on an Test for overall effect: Z = 2.61 (P = 0.009) overview of all the trials are represented by diamonds at the bottom of the figure. 0.1 0.2 0.5 1 2 5 10 Reproduced with permission from the Favours therapy Favours control Outpatient Therapy Trialists. 104 .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 925 17.9 Longer-term follow-up and chronic disease management 925 comprehensive stroke service, some system of review disease management. The rationale is to provide a and further intervention by therapists are valuable. system for managing the various needs, in particular prevention of disease and management of disability. A comprehensive stroke service should include The best setting in which to provide such chronic systems that can review the stroke patient’s progress disease management is likely to depend on the local and provide further intervention by therapists if healthcare system. In countries such as the UK and the required. Netherlands with a well-developed primary care service, it may be appropriate for much of this continuing care The discussion above does not indicate how such to be provided by the patient’s own general family prac- therapy services should be delivered (e.g. in the patient’s titioner. In countries without developed primary care home, outpatient clinic, day hospital). The effectiveness services this may need to be done through a hospital out- of domiciliary physiotherapy and of physiotherapy pro- patient clinic. Whatever system is in place, it is import- vided in a day hospital have been compared in random- ant that it provides not only secondary prevention but ized trials. 105–109 Together they demonstrate only small an opportunity for reassessment, and identification and differences in outcome. 110 The relative costs of providing treatment of problems that may become apparent only care in these settings varied between the studies, with after several weeks or months (Table 17.11). A follow-up no clear conclusion. checklist should ensure that late problems are not over- looked (Table 17.12). Some patients make little progress in the first few 17.8.2 Stroke liaison worker services months after the stroke (perhaps due to intercurrent ill- The generic title of ‘stroke liaison worker’ can be defined ness) and are discharged to a supported environment, as ‘someone who provides emotional and social support but then unexpectedly begin to improve. Ideally, such and information to stroke patients and their families patients should be identified and re-enter a rehabilita- and liaises with services with the aim of reducing aspects tion programme, but few services are sufficiently well of handicap and improving quality of life for patients organized or adequately resourced to offer this. with stroke and/or their carers’. 111 These services (which have also been termed ‘specialist nurse support’, ‘stroke Secondary prevention, which will be lifelong, can be family care worker’ and ‘stroke family support organ- provided in a hospital-based clinic, but is managed izer’) often involve approaching patients and families more conveniently (for the patient) in the primary during hospital admission, when they can provide healthcare sector. information and education about stroke. They are also available for input after discharge home, particularly to identify problems or unmet needs and to develop cus- tomized solutions. 112 At least 15 randomized trials have Table 17.12 A stroke follow-up checklist. tested out this type of service in the UK, Australia, the US and the Netherlands. Their impact on patient outcomes Impairments Ask about weakness, balance, speech, is unclear 102,111,113,114 but they do appear to be valued pain by patients and carers. Disabilities Do you need help with any everyday In summary, in a comprehensive stroke service there activities? should be systems to allow review of stroke patients’ Aids and Do you need any aids? adaptations Have they been delivered yet? progress and to provide therapy input if required. There Are you using them? may be a number of ways to deliver such services (dom- Are they in good working order? iciliary, outpatient clinic, day hospital). Support services Are they in place? Are they appropriate and adequate? New problems Any new problems since last seen? Aspirations Anything you want to do but cannot do? Work Are you back to work? 17.9 Longer-term follow-up and chronic Driving Are you driving a car? disease management Carer Do you have a carer? How is your carer coping? Prevention Check blood pressure, cholesterol, diabetic control, smoking, diet, Stroke has now been recognized as one of the diseases exercise, adherence with medication which requires ongoing management – so-called chronic .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 926 926 Chapter 17 The organization of stroke services It is therefore important to tailor the provision of 17.10 Generic issues in stroke service delivery information to the individual’s needs and wishes. These complexities may in part explain the mixed results from clinical trials of information and education inter- ventions. 109 There is some evidence that information 17.10.1 Information for patients and carers combined with educational sessions improves knowl- One generic theme in the management of stroke – which edge and is more effective than providing information is relevant throughout the patient pathway – is com- only. Information provision alone seems to have no munication of information. Studies of the attitudes of measurable effect on mood, perceived health status or patients and their carers to medical services in general, quality of life for patients or carers. The impact on satis- and stroke services in particular, have demonstrated faction is inconsistent. that one of the greatest sources of dissatisfaction is with In the absence of good evidence to guide us we aim to communication. 72,97 Patients and carers may receive provide information using a number of different media, very little information about the nature of stroke, its including: cause, management and likely prognosis. 25,115 Even where • a notice board on the stroke unit (Fig. 17.5); information is provided, it may be in a form that is • an information pack containing appropriate leaflets; difficult to understand or retain. • audio and video tapes; Patients’ and carers’ perceptions of the stroke service • individual interviews with patients and carers by are likely to depend not just on the degree of recovery, members of the team; but also on the quality of communication. Although it is • patient and carer groups. easy to show that many patients receive little informa- However, there is probably no substitute for one of the tion, one must also remember that for some it may be team sitting down with the patient and family on one enough. Some patients do not want a lot of information, or more occasions to explain the situation and answer preferring to trust in the professionals’ judgement. 115,116 any specific questions. This can then be backed up with Fig. 17.5 A notice board at the entrance to a stroke unit. Typically it introduces the members of the stroke team, indicating how they might be contacted. It presents, in simple terms, what a stroke is and how it can affect the patient. We also include useful pamphlets and information relating to patient and carer groups. .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 927 17.11 Planning, developing and maintaining a stroke service 927 Table 17.13 Tips on improving communication with carers of occasionally be life-saving and produce a good func- hospitalized stroke patients. tional outcome (sections 12.7.3, 13.3.5, 13.3.6). Neuro- radiological interventions are now a key component Provide a notice board at the entrance to the unit of subarachnoid haemorrhage management (section introducing staff and informing relatives about how to 14.6). Inevitably, many smaller hospitals do not have contact members of the team these facilities. However, for the 1–2% of stroke Hold ward rounds during visiting times patients who are at risk of deteriorating rapidly with a Hold a regular ‘open access’ clinic where carers can meet the surgically remediable complication (e.g. acute hydro- consultant Invite carers to participate in patient care (including therapy cephalus), a management plan should be available. sessions) This should include a policy to ensure the early Set up a carers’ group identification and safe transfer of suitable patients to a Document the content of any discussions with relatives in neurosurgical centre. the notes and report these at team meetings to ensure • Transfer for continuing rehabilitation; where patients consistency have been admitted to one service for acute care but Arrange predischarge case conferences or family meetings have to be referred to a separate institution for rehab- Back up verbal communication with written or audio ilitation, needless delays can result. In such circum- material stances, one will often read in the case notes ‘waiting for rehabilitation’ when, of course, rehabilitation should have started on the day of the stroke (section 11.11). written material. One approach, which has been used Thus, one needs to organize services so that the pa- in other areas (e.g. oncology), is to record the interview tient’s needs are matched by the care provided at all and give the recording to the patient or family so they stages of his or her illness. can review what has been said as and when they wish. 117 • Transfer from hospital to community: these challenges This might overcome the problem of patients and fam- were considered in section 17.7. ilies only taking in a small proportion of the information given to them. Like all other areas of stroke care, a service Many problems can arise when patients are needs to establish a system ensuring that input – in this transferred from one service to another. Remember case the provision of information – is tailored to the rehabilitation is a journey, not a destination. individual needs of patients and carers (Table 17.13). 17.10.3 The hazards of transfer 17.10.2 Integration of services Where patients have to be transferred between institu- Inevitably, most stroke patients require both community- tions (or teams) to receive the appropriate care, there is a based and hospital-based resources at some stage in their very real danger that continuity of care will suffer. A con- illness. They may even need more than one hospital- sistent approach to patients and their families will not be based service. It is therefore important to consider how achieved without excellent communication between the these can be integrated to ensure that patients are appro- professionals involved. Patients’ medical records should priately placed at each stage of their illness and that follow them through the system, and ideally at least one transfers between each part of the service are as seamless health professional should be involved in a patient’s care as possible. Figure 17.1 illustrates how the components from admission to discharge, and perhaps even beyond. of a stroke service might fit together. Many places have the necessary facilities and skills to provide an excellent service to patients with stroke. More often, the problem lies with the organization and inte- gration of these facilities. The key areas where problems 17.11 Planning, developing and maintaining may arise are: a stroke service • Transfer from community to hospital; section 17.4 summarizes many initiatives to improve such trans- fers. Agreed protocols between primary and secondary In the remaining sections of this chapter, we will focus care are essential. on the practical aspects of planning, developing, estab- • Transfer to specialist hospital services (e.g. neuro- lishing, monitoring and maintaining a stroke service. surgery and neuroradiology); these facilities should The exact manner in which the service is best provided be readily available, because urgent neurosurgery can will depend on local history, geography, needs, resources, .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 928 928 Chapter 17 The organization of stroke services people and politics. Any stroke service must therefore be Table 17.14 Criteria for an ‘ideal’ stroke and transient tailored to the local conditions to achieve maximum ischaemic attack (TIA) incidence study. 122,186 effectiveness. Therefore it is difficult to be dogmatic about exactly how services should be organized. A large, stable well-defined study population; the number and sex of the people in the population should be available in at least 10-year age intervals during the study. 17.11.1 Determining the needs of the population This usually requires a recent census. Complete ascertainment of all patients with either stroke or If the aim of the service is to provide care for all those in TIA occurring in that population, whether referred to the population who require it, not just those who can hospital or not. This requires multiple overlapping afford to pay for it, then the first factor to consider is the methods to detect cases, including contacting primary incidence of stroke and transient ischaemic attack in health teams, review of hospital admissions, imaging that population. This provides the basis for a ‘needs records and death certificates. assessment’ which is fundamental to determining how Accurate assessment of the cross-boundary flows in both much stroke service should be provided. directions. First-ever-in-a-lifetime strokes and TIAs should be distinguished from recurrent strokes and TIAs. Incidence of acute stroke and transient ischaemic attack Prospective assessment of all suspected cases so that standard diagnostic criteria (WHO definition) can be Despite the huge burden that cerebrovascular disease applied rigorously soon after the patient presents to places on communities throughout the world, there are medical attention (so-called ‘hot pursuit’). less reliable data on its burden than one might expect. Studies should register patients with ‘TIAs’ as well as strokes Although a large number of ‘incidence’ studies have to ensure that mild strokes, which may be misclassified as appeared in the literature over the past 40 years, most TIAs by referring doctors in routine clinical practice, are have had methodological weaknesses that make their not under-represented. results, at least in part, unreliable. The criteria for an ‘ideal’ Brain imaging to determine the pathological type of stroke. study are listed in Table 17.14. Table 17.5 gives the age- This should be performed early enough after stroke onset specific incidence of stroke provided by some more-or- to reliably distinguish ischaemic from haemorrhagic stroke. Case ascertainment over whole years to avoid bias due to less ‘ideal’ studies, most of which were based on white any seasonal fluctuations in incidence. populations because little reliable information is avail- Standard methods of data presentation – i.e. not more than able elsewhere. 5 years of data averaged together, incidence for men and women presented separately, incidence in those of There are quite good data on the incidence of stroke over 85 years old if possible, incidence presented as and TIA in many white populations, some data on mid-decade age bands (e.g. 55–64 years) but 5-year Oriental populations, but little reliable data for other bands available, 95% confidence intervals. parts of Asia, South America or Africa. 18.2.1). The accuracy of mortality statistics also depends 17.11.2 Possible approaches to assessing need on the accuracy of the population denominators used Because of the dearth of reliable data on stroke incidence and thus on the reliability and timing of the most recent (i.e. the number of first-ever-in-a-lifetime cases of stroke census. Furthermore, mortality statistics only include occurring in the population over a defined time period), the deaths attributed to stroke (and not the number of those planning stroke services may decide to base their stroke episodes), and are not in themselves of much value estimates of need on routinely collected mortality data in estimating the need for health services. Although they (i.e. the number of deaths attributed to stroke in the are bound to reflect, at least indirectly, the incidence population over a defined time period). They should, of stroke in the population, the case fatality may differ however, be aware of the potential problems of adopting from place to place, and so there will not be a uniform this policy. relationship between stroke mortality and incidence. There is some evidence that case fatality is falling, while incidence is not (section 18.2). Mortality statistics These depend on the collation of data from death Hospital admission or discharge statistics certificates. They are thus dependent on the accuracy of death certification, which even in countries with quite These are an alternative source of information which may high postmortem rates is known to be poor (section reflect the incidence of stroke. Outpatient attendances .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 929 17.11 Planning, developing and maintaining a stroke service 929 Table 17.15 Annual age-specific incidence of stroke per 100 000 population in the 1980s and 1990s. (Adapted from Feigin, 2005 12 and Sudlow & Warlow, 1997. 187 ) Age (years) Place and mid-year of study 0–44 45–54 55–64 65–74 75–84 ≥85 Northern Europe Norway, Innherred (1995) 12† 40 217 741 1820 3039 Sweden, Soderhamn (1990) 12 67 313 976 2056 2995 Denmark, Frederiksberg (1989) 4 104 306 712 1298 1599 Finland, Espoo-Kaunianen (1990) 125 306 618 UK, London (1996) 21 87 221 516 891 1892 UK, Oxfordshire (1984) 9 57 291 690 1428 2009 UK, Teesside (1996) 11 89 297 611 1247 2099 UK, Scottish Borders (1999) 13 131 255 659 1587 2400 Eastern Europe and Russia Poland, Warsaw‡ (1991) 14 76 268 408 901§ 1355¶ Ukraine, Uzhgorod (2000) 10 400 750 1500 2600 750 Georgia, Tbilisi (2002) 12 78 360 721 1029 1030 Russia, Novosibirsk (1992) 28 246 496 1060 1554 1513 Central and Southern Europe France, Dijon (1987) 10 62 119 410 979 1641 Germany, Erlangen (1995) 16 105 196 508 1226 2117 Greece, Arcadia (1994) 14 82 218 568 1220 2661 Italy, Aosta (1989) 13 82 255 707 1607 3237 Italy, Belluno (1993) 10 114 242 720 1317 3413 Italy, Umbria (1988) 5 115 280 541 1458 2180 Italy, Calabria (1996) 10 69 149 570 1454 2040 Portugal, Porto (1999) 162 337 681 1092 1685 Australia and New Zealand Australia, Perth (1989) 17 98 207 511 1679 2369 New Zealand, Auckland (1991) 18* 82 253 647 1267 1967 Australia, Melbourne (1996) 105 213 535 1290 2900 North America US, Rochester (1988) 9 62 269 642 1272 2111 South America and West Indies French West Indies, Martinique (1999) 124 272 630 1196 1820 Barbados, West Indies (2002) 11 94 219 578 963 2208 Chile, Iquiqie (2000) 8 100 308 462 1037 1089 Eastern Asia Japan, Oyabe (1989) 153 308 781 1940 4385 *Age group 15–44 years. †Age group 15–44 years. ‡Subarachnoid haemorrhage excluded. §Age group 75–79 years. ¶Age group ≥ 80 years. are seldom recorded. Again, inaccuracies in diagnostic population-based stroke register. 120 But the biggest prob- codes may limit their usefulness. 118,119 Also, data may be lem is that these data only include those patients who distorted by double counting, which frequently occurs are admitted to hospital with a stroke. They are therefore when patients are transferred from an acute centre to a a better measure of the hospital service that is currently rehabilitation or continuing care facility. Stroke incid- provided than of the population’s needs. The relation- ence estimated from hospital discharge data has been ship between stroke incidence and hospitalization can found to be much higher than that derived from a only be known with certainty where a reliable stroke .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 930 930 Chapter 17 The organization of stroke services incidence study has been performed to determine the proportion of stroke patients who attend hospital. Incidence Where hospital admission rates have been determined, they vary considerably between places – 55% to over incidence 95% in developed countries, while there is no informa- Prevalence ∞ case fatality tion from Asia and Africa. 15,16,34,121 There is also relat- ively little information about changes in admission rates with time. 121 The rate of hospital admission, or even the attendance at an outpatient clinic, depends on several factors that are independent of population need and Case fatality may be difficult to predict. Stroke prevalence is another measure of the frequency of Fig. 17.6 A bath with water running in (representing incident stroke that some suggest is useful in planning services. cases), the water level (representing prevalence) and water Prevalence is the number of people who have ever had going down the plug hole (representing deaths in prevalent a stroke living in the population at any one point in cases). The prevalence is directly proportional to the incidence time. Prevalence data could be useful in determining the and inversely proportional to the case fatality among prevalent needs for long-term support services in the community. cases, although the mathematical relationship is quite However, we would argue that the greater the time complex. that elapses after an acute stroke or transient ischaemic attack (TIA), the less important disease-specific services needs, which uses both local data and information from become. If one is interested in determining the need for incidence studies from other areas (Table 17.16). If one long-term support services, it is more useful to estimate looks at the estimates of incidence from the most reliable the prevalence of disability due to all causes, rather than studies (Table 17.14), it is surprising how little variation just that related to stroke. In addition, stroke prevalence there is in the incidence of stroke (Table 17.15). How- can never reflect the true burden of stroke, because the ever, these estimates are based mainly on white popula- patients who die soon after a stroke are not represented. tions, and the incidence may vary in different ethnic There are also a number of important methodological groups. For instance, in both New York and London, difficulties with measuring the prevalence of stroke the incidence of stroke is significantly higher in blacks and TIA, not least the need to make accurate diagnoses than in whites 124,125 – although this has been attributed sometimes years after the actual event and to survey to case-finding artifact. 126 Also, the incidence of stroke thousands of people. Where this has been done the among people of Pacific origin in New Zealand is higher prevalence usually ranges from 5 to 10 per 1000 popula- than in those of European origin. 123 However, the incid- tion. 122 As a quicker alternative, one can estimate the ence of stroke among ethnic groups may not be the same prevalence of stroke from its incidence and case fatality as the incidence in the same ethnic group living in the using the ‘bath principle’ (Fig. 17.6). Where this has original population from which they came, and it may been done for a predominantly white population in be that differences between ethnic groups in the same New Zealand, the prevalence was estimated to be 8 per country may be more to do with socioeconomic circum- 1000 population (9 in men, 7 in women) aged 15 years stances than race. and older. 123 Of the prevalent cases, around 55% had not One should start with the age-specific and sex-specific made a complete recovery from their stroke, and 21% stroke incidence in the population that is closest to the needed help with self-care activities. local one in terms of geography, ethnic composition and culture. These rates can then be applied to the local Stroke prevalence is difficult to measure directly, and population numbers to obtain an age-specific and sex- of limited use for planning stroke services. standardized incidence. A comparison of routinely col- lected data from the area in which the stroke incidence study was performed with equivalent data from one’s 17.11.3 A practical approach to needs assessment own vicinity may give further evidence of the relevance The vast majority of health service planners are not for- of the incidence data. The cause-specific mortality data tunate enough to have had a recent, methodologically are likely to be the most reliable and easily obtained. One sound, stroke incidence study in the population they can then judge whether the incidence locally is likely to be serve. Rather than carrying out their own incidence greater or less than that in the available incidence study. study, which is time-consuming and expensive, we recom- An alternative approach might be to use local hospital mend the following approach for determining local admission or discharge data and to adjust them to take .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 931 17.11 Planning, developing and maintaining a stroke service 931 Table 17.16 A step-by-step guide to Step 1 Obtain the most accurate census data for the population of interest for each estimating approximately the number of sex and age band. strokes in a local population of interest. Step 2 Identify the ‘ideal’ incidence study that is likely to have been done in a similar population to the one of interest (e.g. geography, race). Step 3 Multiply each age-specific and sex-specific incidence by the number of people of that age and sex in the population of interest – e.g. if the incidence in men 65–74 years old is 690 per 100 000 and there are 11 000 men in this age band in the population of interest, one would expect about 76 men (i.e. 690 × 11 000/100 000) between 65 and 74 years of age to have a stroke each year in the population of interest. Step 4 Sum the numbers of patients of each sex in each age band expected to have stroke to obtain the total number expected in the population. Step 5 Consider making an adjustment for any marked differences in, for example, the cause-specific mortality between the population of interest and the population in which the incidence study was done. Step 6 If one is interested in transient ischaemic attacks, then their incidence is usually about 30% that of stroke (section 18.4). Step 7 The number of recurrent strokes is of the order of 30% of first-ever-in-a- lifetime strokes, so that to estimate the total number of strokes likely to occur in the population of interest, the number of incident strokes should be inflated by 30%. If interested in the total numbers of patients who could be referred to your service with suspected stroke or TIA then add the additional number of referred patients who may be expected to have a non-stroke diagnosis (possibly 50% for TIAs and 30% for acute stroke). 44,188 account of the likely proportion of stroke patients admit- than others. Therefore, if one plans to use hospital dis- ted. This will be more reliable for stroke than for TIAs. charge data, one would be wise to be selective in one’s One could estimate this proportion admitted by survey- choice of discharge codes. ing the local primary health teams and asking them to The crude incidence of stroke tells one how common report the proportion of patients with acute stroke who the problem is, but in itself is not sufficient to determine they refer to hospital. This may be reliable enough if they the health service needs of the population. Before one report that they refer virtually all cases (as in Sweden), can plan a service, one needs to have estimates of the but if the proportion is smaller, their estimate may following. be misleading. 16 Also, hospital discharge data may be Age-and sex-specific incidence. Younger patients require inaccurate for a number of reasons: 118,119 different facilities from older ones, e.g. retraining for • inaccuracy of routine clinical diagnosis, especially of employment. Older women more often live alone, and TIAs; may require more formal support in the community. • lack of CT brain scans to confirm the stroke diagnosis Type-specific incidence (that is transient ischaemic and exclude other diagnoses; attack, ischaemic stroke, primary intracerebral haemor- • use of vague terms, e.g. ‘acute hemiparesis’, ‘cere- rhage and subarachnoid haemorrhage). These data may brovascular disease’ in medical records and discharge be useful in more detailed planning of the population’s summaries from which routine codes may be derived; needs. Patients with TIA and minor ischaemic stroke • coding errors; require prompt diagnosis, investigation and initiation • failure to distinguish acute stroke admissions from and supervision of secondary prevention but not pro- those due to complications of an earlier stroke; longed inpatient care, rehabilitation or community • failure to code strokes occurring in hospital, or in the support services. Patients with subarachnoid haemor- context of another diagnosis. rhage require emergency hospitalization and neuro- Table 17.17 gives the likelihood that a patient allocated radiological and neurosurgical facilities (Chapter 14). one of the cerebrovascular codes on discharge from one Haemorrhagic strokes have a higher early case fatality, of five Scottish hospitals had actually had an acute so although they may require more care in the very stroke. It was also reported from Norway 120 that certain early stages, the longer-term burden of severe disability diagnostic codes were more likely to identify true strokes may be less than for patients with ischaemic stroke. The .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 932 932 Chapter 17 The organization of stroke services Table 17.17 Positive predictive values of Code All hospital Emergency cerebrovascular disease codes (ICD 9 or admissions admissions only 10) for a diagnosis of stroke. Positive predictive value for acute stroke (%) Subarachnoid haemorrhage 85–88 88–93 Intracranial haemorrhage 87–91 92–94 Non-traumatic intracranial haemorrhage 58–75 63–75 Stenosis/occlusion of a precerebral artery 1 0 Ischaemic stroke 83–93 86–95 Transient ischaemic attack 5 6 Stroke unspecified 78–87 82–88 Cerebrovascular disease, unspecified 25–28 31 Cerebrovascular disease sequelae 0 0 The proportion of patients who were allocated a cerebrovascular code (ICD 9 or 10) on discharge from five Scottish hospitals and who actually had had an acute stroke. The values are higher for emergency admissions. This study did not attempt to identify patients who had a stroke but received a non-stroke code. estimates of the relative frequency of the pathological countries, older people account for an increasing pro- type of stroke are remarkably similar in most of the portion of the population. Therefore, for a disease published incidence studies, which come from predom- such as stroke in which the incidence is much higher in inantly white populations (Fig. 17.7). Although the pro- older people, the total number of strokes will increase portion of patients who are haemorrhagic is probably unless offset by a falling incidence. However, one has to higher in Oriental populations, it is difficult to be cer- remember that case fatality is higher in older patients tain, because of limited information. The proportion of and in those with prior disability, and therefore fewer of strokes due to subarachnoid haemorrhage and primary these older stroke patients will survive to require long- intracerebral haemorrhage is greater in the young than term care. Changes in clinical practice are likely to force the old. rapid and quite unpredictable alterations in the shape of The prognosis of stroke and its subtypes. From the type- clinical services for stroke. The changes in management specific incidence and case fatality, one can estimate of stroke have, until now, been small and gradual, but the likely requirements for assessment and diagnostic with recent increases in research efforts, major changes services, acute care, rehabilitation and terminal care are more likely and the needs of the population could services, long-term care, community support and sec- alter rapidly and unpredictably. The introduction of ondary prevention. The prognosis of stroke and its coiling rather than clipping of ruptured intracranial subtypes has been discussed elsewhere (section 10.2.3). aneurysms after the ISAT trial (section 14.4.2) has resulted Information on stroke severity and comorbidity would in large, sudden and difficult-to-manage changes in be invaluable, since these will be major determinants services; patients once admitted under neurosurgeons of patients’ health service use. However, such data are and operated on may now be admitted under neuro- unlikely to be routinely available unless a stroke register logists and treated by radiologists. People’s expectations has been kept. of healthcare, often driven by media reports of medical Changes over time. When one is planning a stroke successes and failures, may also force changes in services service, one has to take account of any changes that that go beyond the evidence of effectiveness. may occur in the future, because one’s service will need to alter to take these into account. Trends in stroke 17.11.4 Assuming that resources are limited, mortality and incidence are discussed later, and there is to what extent will the needs of the population evidence that the incidence, and maybe even the sever- be met? ity, of stroke is falling in many western populations. Apart from the changing incidence, one also has to take Having obtained an estimate of one’s local age-specific, into account changing demographics. In most developed sex-specific and type-specific incidence and outcome of .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 933 17.11 Planning, developing and maintaining a stroke service 933 100% 2 10 12 6 5 5 9 9 15 90% 2 5 9 3 1 5 10 11 11 8 2 80% 10 15 10 70% 60% 50% 86 83 40% 78 82 77 73 73 72 Fig. 17.7 Histogram showing the 30% proportions of patients aged between 20% 45 and 84 years with a first-ever-in-a- 10% lifetime stroke due to ischaemia, primary intracerebral haemorrhage, subarachnoid 0% Dijon, Umbria, Oxfordshire, Frederiksberg, haemorrhage and uncertain, in more or France Italy England Denmark Perth, Rochester, Aosta, Soderhamn, less ‘ideal’ incidence studies. Only studies Australia USA Italy Sweden where a computed tomography (CT) brain Undetermined The numbers by the scan was done in more than 70%, and Subarachnoid haemorrhage columns indicate the including patients with subarachnoid Primary intracerebral haemorrhage percentage estimates haemorrhage, are covered. Ischaemic stroke stroke and TIA, one is still a long way from determining similar services in all parts of the country. In the UK, the needs of the population. After all, this estimate does stroke services are tending to evolve out of existing not indicate the actual resources these people will require, rehabilitation facilities, which were traditionally pro- and of course at some stage somebody has to make a vided by geriatric medicine or general medicine services. political decision about how completely the popula- • The needs of the patient population. Within stroke care, tion’s needs are to be met. Inevitably, where there are as within healthcare in general, there is frequently a limited resources for health services, choices have to be tension between the need to centralize services for made about allocating resources between areas (e.g. should individuals with very specialist needs (for example, one build a new hospital or a new school?). Having been subarachnoid haemorrhage) and the aspiration to given limited resources, however, it is important that provide as much high-quality care as close to the indi- those planning stroke services should use them in as an vidual’s home as possible. efficient a manner as possible and should prioritize (i.e. • Patients’ wishes. Once again there is a tension between ration) to make sure that sufficient funds are available for pressures to centralize and fragment care, for example, whatever are perceived to be the most important aspects between acute and rehabilitation services and to pro- of the service. These decisions are usually taken by polit- vide continuity throughout the patient journey which icians but, because they depend on information about is usually valued by patients and carers. the effectiveness of components of the stroke services, it • Costs and resources. There are increasing pressures to is essential for the politicians to receive sound medical control the costs of healthcare services and this is par- advice. ticularly relevant when considering emergency cover The perceived pressures to develop particular com- by senior medical staff. Once again centralized services ponents of stroke services and the way in which these meet these challenges by having large groups of staff, services develop are likely to reflect a number of factors or at least networks of staff, who are available to be apart from the effectiveness of the service offered: on call. • Local healthcare culture and economy. The types of services provided for stroke patients vary greatly between the 17.11.5 Planning a stroke service developed economies. In the US and Germany services often focus around centralized neurological centres or Having established the aims of the service, the needs of stroke centres where selected patients receive acute the population and the degree to which one expects to care but not ongoing rehabilitation. There is evidence fulfil these, one can then plan the development of ser- that access to these services 127 may be reduced for vices. Usually, one starts with the existing services, even certain patient groups such as the elderly. In contrast if these are inadequate and chaotic. It is then useful to the Scandinavian countries have tended to approach consider the following two questions prior to making stroke services with a more devolved delivery offering any changes: .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 934 934 Chapter 17 The organization of stroke services • What are the current resources committed to the man- The immediate priorities for improving a service will agement of patients with TIA and stroke? also depend on how easily or cheaply particular prob- • What are the major gaps (i.e. unmet needs and failure lems can be solved. For instance, poor standards of to provide cost-effective interventions) in the present medical assessment may be improved by the introduc- provision of services? tion of a protocol and education for junior medical staff One needs to establish the strengths and weaknesses with little implication for resources, while the provision of the current services in order to identify the most of an occupational therapy department where there isn’t important areas for improvement. Priority should be one has major resource implications. Planning a stroke given to providing patients with basic care (e.g. nursing unit is likely to be less challenging in a well-resourced to provide for basic needs) and to delivering those inter- hospital setting where reorganization of staff and work ventions that are of proven effectiveness. Although patterns is needed. However the obstacles are formidable information about current services may already be avail- if the key staff members are not available at all, and even able, it is likely that in addition to routinely collected if they are many staff do not like being reorganized data, it will be useful to carry out a survey to determine: without very good reason, much explanation and proper • How many and what sort of patients are currently support. being managed (i.e. demographic and clinical data)? • Where are they being managed (i.e. in the hospital or community, in accident and emergency departments, neurology, internal medicine or geriatric medicine)? • By whom are they being managed (i.e. general or fam- 17.12 Evaluating and monitoring stroke ily practitioner, neurologist, general physician)? services • How are they being managed (i.e. the process of care)? • What resources are currently being used? A community-based register that identified all patients We believe that the most reliable way of determining the in the population who had a stroke or TIA would be relative effectiveness of interventions is an appropriately an ideal but expensive and impractical way of answer- designed randomized trial where feasible, or a systematic ing these questions. A hospital-based stroke register is review if more than one trial is available. This of course is a practical alternative that can help to answer most not an option in the evaluation of a local stroke service of the questions, although clearly it cannot provide de- rather than stroke services in general, so we have to rely tailed information about patients who are not referred on less robust methodologies. to hospital. A register is an invaluable tool for mon- Non-randomized comparisons of the process of care, itoring the performance of services as well as planning or of patient and carer outcomes achieved by services, them. are the only practical methods of evaluation. If one is setting up a new service in a hospital, the process, or out- Set up a hospital-based stroke register to get some comes, can be compared with those in a nearby hospital idea of the current state of the local stroke service – without a new service. Alternatively, one could measure however fragmented and chaotic it is. the process and outcomes achieved by the existing service and then measure whether these are improved The simplest, quickest and most practical approach after the new service has been established (i.e. a before- is to carry out a survey of the current services against and-after study). However, such evaluations can be certain standards. Those working within a service are misleading. They may demonstrate improvements (or often fully aware of its deficiencies, and their knowledge worsening), or differences in the process or outcome of should not be ignored (although sadly it often is). This care, but they cannot provide reliable data to indicate will only work where there is a willingness by the import- that any changes are actually due to the new service. ant parties to acknowledge deficiencies and to change One cannot rely on such non-randomized evaluations practice. This approach can identify areas of strength to influence practice elsewhere, but they may fulfil and weakness, and may determine which area to concen- important local functions. One could reasonably argue trate on first (e.g. Sentinel Audit 128 ). that, as long as the new service is not much more expen- sive to run, it does not matter whether any improve- An honest, objective appraisal of services against some ments observed can be attributed directly to that service. agreed standards by those involved in providing them, Obviously, if the new service was very costly, one would perhaps facilitated by an independent observer, may want reassurance that the improvements had not occurred be the best stimulus for service development. spontaneously. But if a non-randomized evaluation .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 935 17.12 Evaluating and monitoring stroke services 935 demonstrates either no improvement or even a worse • prompt access to non-invasive vascular imaging, and outcome, it is difficult to know how to respond. Were the carotid surgery if necessary (section 16.11); changes due to the new service, in which case it ought • continuing care facilities, both community-based and to be modified, or were they due to some unforeseen institutional. confounding factor? However, care has to be taken in defining these stand- There are several other methodological problems, ards. For example, what does ‘prompt access’ really mean, which can affect randomized as well as non-randomized what is a specialist, what does a stroke unit consist of? comparisons, and which need to be considered: One could very quickly fulfil a requirement for a stroke • small numbers of patients; so that any change observed unit by simply relabelling a general medical ward, but may be accounted for by the play of chance or missing of course one is unlikely to accrue the benefits of stroke a real and worthwhile change because too few patients unit care. were studied; • observer bias in assessing the process or outcome; 7.12.2 Process often, the observers have an interest in the result of the evaluation, which may influence their judgements. Some aspects of the process of care are easily monitored, So what aspects of a service can we measure? The e.g. waiting times for appointments. It may be relatively simplest way to monitor the service is to measure the easy to define a standard (for example, patients with amount and nature of work being carried out. Unfort- TIAs and ischaemic strokes should be given aspirin to unately, politicians and those who fund healthcare place reduce the risk of further vascular events, unless there are too much emphasis on the volume rather than quality of contraindications), but for other procedures for which service. A stroke service may include some components there is less scientific justification, it is more difficult. (e.g. processes) for which there is little doubt of their The lack of scientific justification can, and frequently is, effectiveness (e.g. aspirin for cerebral infarct, low-risk overcome by using the combined views of recognized carotid surgery). When assessing performance, it is there- experts, i.e. a consensus. However, what appears to be a fore most important to monitor how well the service fairly straightforward standard cannot be applied sens- delivers those components whose effectiveness is ibly to every patient because – for example – they cannot established. comply with the assessment. The UK National Sentinel In assessing the quality of care (or ‘clinical audit’, Audit of Stroke 71 addressed this problem by having a ‘no as it tends to be called) one should consider three but . . .’ clause attached to each standard (Fig. 17.8). This aspects of the service: 128 the structure, or facilities avail- is essential for the process of care to be compared in dif- able; the process of care; and the outcomes for those ferent groups of patients. treated. Another difficulty is that, by directly observing the care, one is likely to alter its delivery (the so-called Hawthorne effect). Also, such an approach is likely to be Structure very costly if performed on all stroke admissions. The From all that has been said above, it should be fairly alternative is to audit the records of care, but this imme- obvious that a stroke service needs certain essential facil- diately raises the question of the validity of the medical ities to provide all the components necessary to care for record, i.e. whether the records reflect the actual care patients with stroke and TIA (Tables 17.1 and 17.2). This provided. However, most people would agree that good makes the setting of standards and the measurement records probably do reflect good care given, and that this of performance for structure relatively straightforward. is a reasonable method of measuring the process of care. Some basic standards (incorporated into Scottish service The other methodological problems involved in audits standards; http://www.nhshealthquality.org/nhsqis/ of case notes are summarized in Table 17.18. 1288.html) for structure might include: • an identified individual who is responsible for the Monitoring the process of care by case note review organization of stroke services; raises a number of important methodological • early access to a stroke unit; problems that must be addressed if one’s assessment is • multidisciplinary staffing of the stroke unit (section to be useful and valid. 17.6); • early access to outpatient assessment by a specialist Although one may be measuring the performance for patients who do not need admission to hospital against some ‘ideal’ standard, one is likely to want to (i.e. a ‘one-stop’ neurovascular outpatient clinic; compare the performance in the same service over time, • prompt access to CT scanning; or to compare performances between services. To do this .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 936 936 Chapter 17 The organization of stroke services Patient assessment 2.1.0 First 24 hours Neurological assessment in the first 24 hours 2.1.1 (a) Have the following been specifically recorded in the first 24 hours? Yes No No but (i) Conscious level (eg Glasgow Coma Scale, alert / oriented) (ii) Eye movements (eg Cranials 3, 4 and 6 intact, doll's eye response if unconscious) (iii) Limb movements (response to pain if conscious) (b) If patient is noted to be conscious, are the following recorded? Answer No, but if....... impaired level of consciousness is documented Yes No No but (i) Screening for swallowing disorders (not gag reflex) (ii) Communication (iii) Trunk control or gait Fig. 17.8 One of the sections from (c) If patient is alert and able to communicate, is there a formal the National Sentinel Audit of Stroke. assessment of: This one relates to the recording of the Answer No. but if..... impaired level of consciousness/communication neurological examination. The interval is documented. Yes No No but (e.g. 24 h) since admission at which (i) Formal mental test (eg mental test score) these data should be recorded, and the (ii) Visual fields (iii) Visual inattention circumstances in which it is acceptable (iv) Sensory testing for the information not to be recorded, are given in each section. 132 Table 17.18 Important methodological issues in the audit of meaningfully, one has to have a valid and reliable meas- case notes. ure of performance, one needs to audit enough cases to produce statistically robust results, and one must be able Patient selection bias to take into account differences in case mix between ser- The case notes audited should be a representative sample of vices or changes over time. The Stroke Audit Package was all those treated (either a consecutive series or a random originally developed by the Royal College of Physicians sample). Beware the missing notes because so often these to overcome these methodological problems and enable patients are more ‘interesting’ to somebody, perhaps 129 valid comparisons to be made. But because this package because they had a rare form of stroke, or died. only addressed a limited number of the more medical Case note retrieval bias aspects of care, the Royal College of Physicians of London Poor-quality case notes or those of dead patients may be therefore also developed the National Sentinel Audit, more difficult to retrieve, which might bias the audit in a favourable direction. A high proportion retrieved is an which includes other important aspects of care delivered important step in reducing bias. 189 by different members of the multidisciplinary team. 130 Lack of precision A sufficiently large number of case notes should be audited 17.12.3 Outcome to provide a precise estimate of performance and to allow precise comparisons to be made with other centres, or The term ‘outcome’ is used in different ways and so causes with audits performed at different times in the same confusion. Clinicians use the term to refer to the clinical centre. outcome of the patient or carer. Outcomes therefore Observer bias include survival, functional status, complications, or less Auditors may have an interest in the result of the audit, easily defined concepts such as quality of life. Others, which may influence their assessment of performance; particularly those with a management background, use blinded, or at least impartial, observers should be the term ‘outcome’ to refer to any result of an interven- employed if possible. tion, e.g. reduced waiting times or readmission rates. Poor inter-observer reliability of measure If the measure of performance is not reliable, then it will be In this section, we use ‘outcome’ to refer to the clinical more difficult to demonstrate real differences between outcome of patients (i.e. physical, functional, cognitive, centres. Also, if there is a consistent difference in the way emotional, etc.). an audit measure is applied by different auditors, this may Since the main aim of stroke services is to optimize the produce invalid comparisons. outcome for patients and carers, the measurement of out- Differences in case mix come is obviously the most relevant criterion by which A standard that is applicable to one patient may not apply to judge the performance of a service. This recognition to another. It is important that standards should be has prompted the development of outcome analyses adjusted for differences in case mix. such as Dr Foster’s notes (drfoster.co.uk). Unfortunately, .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 937 17.12 Evaluating and monitoring stroke services 937 the use of outcomes to reflect the quality of care is the most before and after its introduction to eliminate the effects challenging area of stroke audit, and there are many of chance, and in itself contribute nothing to the elim- well-known difficulties to be overcome. 131–136 Until they ination of biases of various sorts. The impact of stroke are, those involved in providing and monitoring health unit care is far less than this (Table 17.6). services must be extremely careful not to misinterpret There are several examples of studies of stroke unit outcome data. services 71,91 where treatment in a hospital with a stroke The observed outcome in a group of patients treated unit was associated with improved outcomes even after by a particular service will be determined by four factors: adjusting for case mix. However, they were based on • the quality and effectiveness of the care provided; large patient samples of several thousands, and smaller • the method of measurement of outcome (e.g. who is studies have often failed to show a clear link between measuring it, and how?); process and outcome. 137 • chance (or random error); • case mix (or mix of prognostic factors at baseline). 17.12.5 The method of measurement of outcome Many attempts to monitor the quality of service by 17.12.4 The quality and effectiveness of care measuring patient outcomes have relied on mortality This is the aspect we hope outcomes will reflect. How- data, presumably because they are often routinely avail- ever, it is important to remember that most interven- able, reasonably objective and may indicate where there tions have only small or moderate-sized effects, which are major problems. Unfortunately, mortality is unlikely may be difficult to detect even in large randomized trials. to be influenced by many components of care (discharge For example, a wildly implausible 50% relative reduction planning being one obvious example). Some outcome in the death rate from 30% to 15% after the opening of a measures that may better reflect the quality of care are stroke unit would require a sample of 200 patients both shown in Table 17.19. They measure outcome at different Table 17.19 Aspects of outcome that may Outcome Promising measurement tools be relevant in assessing stroke services, and some tools for measuring these Survival Case fatality during a defined time period, e.g. at 30 days or outcomes. 6 months after stroke onset Complications Proportion of patients developing pressure ulcers or fractures; there are difficulties in defining these and reliably recording them; paradoxically, better services may identify more and record them more often Residual Probably not very useful, and not easily collected after hospital impairments discharge Mobility 10-m walking speed Arm function Nine-hole peg test Psychological Hospital anxiety and depression scale outcome General health questionnaire Many of the most disabled patients will not be capable of responding to measures of psychological outcome Disability Barthel Index Functional Independence Measure (FIM) Three simple questions (Fig. 17.9) Oxford Handicap Scale (also known as the Modified Rankin Scale) – should be measured at a defined point after the stroke, e.g. 6 months; easily collected after hospital discharge Handicap London Handicap Scale 190 – a difficult area with no well-tested measures; the Oxford Handicap Scale does not really address handicap in isolation Patient or carer Hospsat and Homesat 99 – is this an outcome satisfaction or process measure? General health Nottingham Health Profile or health-related Short Form 36 quality of life Euroqol – potentially interesting because it could allow comparisons with other disease states; however, many stroke patients cannot complete the questionnaires because of cognitive problems .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 938 938 Chapter 17 The organization of stroke services Table 17.20 Important features of scales Validity, different types include: for the measurement of outcome after • Criterion validity, when the measure is related to an accepted ‘gold standard’. stroke. • Construct validity, where the measure is related to existing measures of similar aspects of outcome. • Content (or face) validity, which relies on expert agreement that the measure is a reasonable reflection of what it is supposed to be measuring. There can be considerable difficulties in demonstrating the validity of a particular measure. 192 Reliability This is the reproducibility of a measurement, most commonly between observers (inter-observer reliability) and over time (intra-observer or test-retest reliability). Relevance The scale should measure some aspect of outcome that is relevant to the patient or carer as well as to the doctor. Thus, the size of a cerebral infarct on a CT brain scan is of little relevance, while patients’ ability to look after themselves is very important to the patient and carer. Practicality Scales vary in their complexity and the time taken to complete an assessment. Studies of long-term outcome involving hundreds of patients need very simple measures which can be completed by postal or telephone questionnaire, while smaller studies in hospital can afford to use more complex measures. Sensitivity A scale should distinguish between patients who have different outcomes or detect important changes in a particular patient. Usually, more sensitive scales are more complex and unfortunately less reliable. Communicability Ideally, the measure will mean something to other health professionals or even patients. It is more useful to know that a patient feels ‘fine’ than to be told that their score on a particular stroke scale was 23 out of 100, for example. levels of disease – pathology, impairment, activity, par- although reduced scales can be completed from case ticipation and quality of life. It is important that they notes. 141 We do not think they are particularly useful should have acceptable validity (i.e. they measure what in evaluating stroke services. They might be used as a they are intended to measure) and reliability (i.e. they measure of case mix. are reproducible in different settings and when used by different people). ‘Far better an approximate answer to the right question, A number of different types of scale have been developed which is often vague, than an exact answer to the and used to measure outcome after stroke and there are a wrong question, which can always be made precise’. 142 number of features that one should look for in choosing an outcome measure (Table 17.20). 138 One can loosely Functional scales (activity) categorize the measures of outcome after stroke under the following headings. These include measures of disability or dependence in activities of daily living (ADL), such as the Barthel Index, the Nottingham ADL Scale and the Functional Stroke scales 143,144 Independence Measure (FIM). Under this heading So-called ‘stroke scales’ (e.g. the Scandinavian Stroke one could also include the so-called extended ADL Scale, Canadian Stroke Scale, National Institute of Health (EADL) scales such as the Frenchay Activities Index and Stroke Scale) were largely developed to describe the the Nottingham Extended ADL Scale 145–147 which iden- severity of acute stroke and to monitor changes in the tify whether patients are participating in more complex patient’s condition. 139,140 Most concentrate on the type activities such as shopping, leisure or work. These scales and severity of the neurological impairments. They have appear to measure relevant aspects of outcome, although been criticized for lacking relevance for patients, being some demonstrate ceiling effects (e.g. the Barthel Index) complex and therefore impractical, and for summing and may not pick up problems in particular areas, ‘apples and pears’. 138 They rely on a clinical examination, e.g. communication. Indeed, one can score maximum .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 939 17.12 Evaluating and monitoring stroke services 939 points on the Barthel and yet be blind, deaf and mute. The scales are in general ‘ordinal’, so that care must be Is the patient alive? taken in choosing the appropriate statistical method to describe or compare groups of patients. Some of these No scales are simple enough to incorporate into a postal or Yes telephone questionnaire, and may therefore be used in Dead large studies of long-term outcome. 148 ‘Do you require help from anybody Handicap (participation) for everyday activities?’ Handicap, or its more acceptable converse, participation, Yes is difficult to define and therefore difficult to measure, No Poor but is undoubtedly of relevance to stroke patients and their carers. The Oxford Handicap Scale (Table 17.19), which is a modification of the Rankin Scale, sounds from ‘Has the stroke left you with its name as if it measures handicap, but it really measures any problems?’ a combination of symptoms, dependency and change in lifestyle. 149,150 However, it has been widely used, is Yes relevant and simple enough to be used reliably over the telephone, and is therefore useful in large studies. 151 Indifferent Precision of assessment can be improved by using a No structured interview. 152 Good Quality of life Like handicap, quality of life is difficult to define and thus measure. A large number of generic measures (otherwise known as multidimensional measures) have been de- Fig. 17.9 ‘Three simple questions’ that can be used to place veloped that attempt to measure outcomes in relation stroke patients into four different outcome categories. 160 to various aspects, including physical function, psycho- logical function, pain and social function. They include the Short Form 36, Nottingham Health Profile, EuroQol, Three simple questions Sickness Impact Profile, Stroke Impact Scale and the Assessment of Quality of Life (AQoL) instrument. 153,154 We have used ‘three simple questions’ to categorize Most provide a profile of outcome rather than an overall patients into those with poor, fair and good outcomes measure, and group comparisons are therefore complex. after stroke (Fig. 17.9). 103,156 This approach appears to be However, the EuroQol and AQoL provide a single meas- reasonably valid and reliable, and is certainly practical ure of ‘utility’, and researchers have made some head- when the outcome of very large numbers of patients needs way in deriving summary scores for the other measures. to be measured. Further work is required to establish the Because they are generic (i.e. can be used across many optimal wording of the simple questions and to test different health states) they offer health economists and them in different languages and settings. 157 others the opportunity to compare the utility of different health outcomes in different diseases. Some are long and Patient satisfaction complex (e.g. the Sickness Impact Profile) and are not suited to large-scale studies in which face-to-face admin- Many healthcare systems are being influenced by market istration is not practical. Also, they all rely on patients’ forces and the idea that patients are consumers. This has views of their health status, which limits their use in placed increasing importance on the satisfaction of our patients with severe communication and cognitive diffi- ‘clients’ (i.e. patients) with their healthcare. Many health culties. It is unclear how valid the carers’ responses on service managers regard patient satisfaction as being an behalf of the patient are to these questionnaires. 155 important outcome, although some consider satisfac- tion to be a measure of process; measures of patient and Like motherhood, quality of life is much admired, carer satisfaction with hospital and home care have been difficult to define and even more difficult to measure. 97 developed. .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 940 940 Chapter 17 The organization of stroke services Patient and carer satisfaction appears to reflect the pro- made. Thus, it is important to measure outcomes in a cess of care and the patients’ outcomes. Improvements large, representative sample of patients or carers. This in satisfaction have been observed in trials of occupa- has implications for the type of measure of outcome tional therapy 158 and early supported discharge services. 70 used, since it must be simple and practical to administer As one would expect, those with poorer physical out- to large numbers, exactly as in large randomized trials. comes and depression are likely to report less satisfaction Therefore, to obtain precise estimates of outcome may with care. 159 However, patients – and particularly women take several years for a hospital to accumulate enough and the elderly – appear to have low expectations and are data to provide precise estimates – for example, of its case often satisfied with what professionals would regard as fatality. Thus, there is likely to be a considerable delay poor treatment. 72,159 between a change in the quality of care and any statist- ically significant change in measured outcomes. Figure 17.10 shows the case fatality in five Scottish hospitals When should we measure outcome? with and without adjustment for important case mix Outcomes some months after the stroke are probably variables. This illustrates that even with several hundred most relevant to patients, but are more difficult and patients, the 95% confidence interval around the esti- expensive to measure than at an earlier stage. Many ser- mates of case fatality are surprisingly wide. We suggest vices monitor the patient’s functional status at the time that power calculations should be performed (as for of hospital discharge, and this information is easily and randomized trials) before instigating any audit to demon- cheaply collected. However, because patients usually im- strate changes in the outcomes following modification prove for several months after a stroke, the longer they of a service. stay in hospital the better their outcome at discharge. Furthermore, patients tend to be discharged at much the 1. Unadjusted same level of disability. Thus, such measures are easily A manipulated and difficult to interpret. It is far more B C relevant to measure the outcome at a fixed interval after D the stroke, but after discharge this will inevitably be E more time-consuming and expensive. However, some of the simpler measures can be completed by telephone or postal questionnaire. 148,151,160 Some measures (e.g. 2. Adjusted for age, EuroQol), which seem ideally suited for use as postal A sex and questionnaires, include visual analogue scales, but these deprivation B appear to be particularly unreliable in stroke patients. 161 C D E How to score dead patients? Because many patients die after stroke, outcome meas- 3. Adjusted urements can only be applied to the survivors. If these for study A measures are averaged, and groups of patients compared, variables B then there may be a serious problem of interpretation if C there are more survivors in one group than in the other. D Some attempt to get round this by giving the worst score E to the dead patients and then including them in the –20 –15 –10 –5 0 5 10 15 20 analysis but, depending on the scale used, this is not necessarily valid. One solution is to measure the propor- Fig. 17.10 Number of deaths above or below that predicted per tion of patients who are ‘dead or disabled/handicapped’, 100 stroke patients admitted (w score) to each of five Scottish but this may sacrifice sensitivity. However, in studies hospitals (A through E) at 6 months. Plots are shown for (1) unadjusted case fatality; (2) after adjustment for age, sex and including large numbers, such dichotomized outcomes deprivation; and (3) after adjustment for the study prognostic may be adequate. 162 variables (age; whether the patient lived alone before the stroke and was independent in simple activities of daily living; and, 17.12.6 Chance (or random error) on admission, whether the patient could speak and was orientated in time and place, could lift both arms against With small numbers, the imprecision of the estimate gravity, and could walk without the help of another person). of performance may prevent useful comparisons being Used with permission from Weir et al., 2001. 136 .. ..

9781405127660_4_017.qxd 10/13/07 2:15 PM Page 941 17.12 Evaluating and monitoring stroke services 941 prognostic factors that were routinely collected, and When planning an audit, estimate the likely number could therefore be adjusted for, were age, sex and social of cases that will need to be included in order to deprivation. 136 identify a difference reliably (i.e. do a power If more powerful predictive factors such as those calculation). identified in the Oxfordshire Community Stroke Project (Fig. 17.10) are taken into account, most of the variation Although the evidence that institutions with greater between hospitals with respect to case fatality disappears throughput have better outcomes is conflicting (e.g. for and can be accounted for by chance alone. 136,163 Unfort- carotid endarterectomy), one argument for stipulating unately, these variables are not routinely available to a minimum patient volume per year is to ensure that allow such adjustment. If variations in outcome remain, measures of performance can be reasonably precise. If it is impossible to know whether they are due to failure one’s local surgeon performed 50 operations in the pre- to completely adjust for case mix, or some aspect of the vious year, with only two deaths or perioperative strokes, care given. 84 In accounting for random variation, one this is a very acceptable 4% complication rate. However, must also recognize the imprecision of any statistical the 95% confidence interval extends up to a very model used, which will depend on the size of the cohort unacceptable 14%. It is therefore very difficult to know from which the model was derived. with any certainty whether one’s local surgeon has The problems of adjusting for case mix are even greater results that make carotid endarterectomy worthwhile. if one considers other relevant outcomes such as quality of life, where we know almost nothing about the factors If outcomes are measured in a relatively small number that predict this. Before we use outcomes to reflect the of patients or carers, bad outcomes may reflect bad effectiveness and quality of care, and to alter services as a luck rather than bad care, while conversely, good consequence, we have to develop reliable methods of outcomes may reflect good luck rather than good care. interpreting them. It will be interesting to see whether this will ever be possible. 17.12.7 Case mix Crude measures of patient outcome do not necessarily The most important determinant of outcome is probably reflect the quality or effectiveness of the care provided. not the quality or even the effectiveness of care, but Even adjusting for case mix may not solve this the type of patient treated. The patient’s age, pre-stroke problem. status, comorbidities and the severity of the stroke are bound to have an overwhelming effect on the outcome, Rather than attempting to interpret measures of out- and may well obscure any real effect that our treat- come at a particular interval after the stroke, the change ments may have. That is why large randomized trials are in the patient’s condition can be used as an ‘outcome’. required to demonstrate modest treatment effects. For example, some research groups have used the func- Case mix can vary considerably in different services tional independence measure (FIM) to assess patients and in the same service over time, which means that raw on admission to and discharge from a treatment pro- outcome data simply cannot be used to reflect the qual- gramme. Any change in the FIM might be considered, at ity or effectiveness of care; they have to be adjusted for least in part, to be a measure of the effectiveness of the differences in case mix. Unfortunately, this assumes that treatment programme, although most of the improve- we know how to adjust for case mix in stroke, which is not ment may actually be spontaneous. The change in FIM the case. Good case mix descriptors include those factors can be divided by some measure of the amount of treat- that are highly predictive of outcome. But as we have ment provided, e.g. length of stay, to give an idea of already seen, our ability to predict outcome after stroke, ‘efficiency’. Unfortunately, differences in case mix such even in terms of survival, is relatively poor (section as age, severity and location of the brain lesion and other 10.2.7). Also, we can only correct for those prognostic medical problems, as well as the interval since the stroke, factors that we can identify and measure reliably. After are likely to influence the rate of change in the FIM. To all, we rely on randomized controlled trials to provide interpret the change in the FIM as a reflection of effect- evidence of the effectiveness of interventions simply iveness or efficiency would therefore still require a because randomization ensures that the different treat- measure of case mix. 164 Another problem is that measures ment groups are balanced for recognized, unrecognized such as the FIM are not ‘interval’ scales (Table 17.19). A and unmeasurable prognostic factors and that the treat- change of 10 points at one end of the scale is therefore ment allocation is not biased. In the report comparing not equivalent to a 10-point change at the other. This outcomes after stroke in Scottish hospitals, the only makes changes in score difficult to interpret. .. ..