Survival, Hazard Function for a New Event, and Healthcare Utilization Among Stroke Patients ≥65 Years
Background and Purpose— The natural history of stroke is still incompletely understood. The aim of this study was to present detailed data on survival, recurrence, and all types of healthcare utilization before and after a stroke event in patients with stroke.
Methods— Three hundred ninety stroke survivors constituted the study population. Information on survival data during 5 years of follow-up, all hospital admissions since 1971, all outpatient and primary care consultations, and all municipal social service support during the year before and after the index stroke admission and patient interviews 1 week after discharge were obtained.
Results— The risk of death or a new stroke was high in the early phase after admission but then decreased rapidly during the next few months. Mortality during the first 5 years was influenced by age and functional ability, whereas the risk of stroke recurrence was influenced by number of previous strokes, hypertension diagnosis, and sex. On a day-by-day basis, 35% were dependent on municipal support before and 65% after the stroke. The corresponding proportions in outpatient care were 6% and 10%, and for hospital inpatient care 1% to 2% and 2% to 3%. Of the health care provided, nursing care dominated.
Conclusions— The risk of dying or having a new stroke event decreased sharply during the early postmorbid phase. Healthcare utilization increased after discharge but was still moderate on a day-by-day basis, except for municipal social service support, which was substantial.
The natural history of stroke is still incompletely known, although there are a number of studies in which risk factors for stroke1,2⇓ and stroke outcome predictors3–12⇓⇓⇓⇓⇓⇓⇓⇓⇓ are analyzed. Most of the high mortality in patients with ischemic stroke has been reported to appear within the first months.12 However, the basis for this statement is unclear, because no hazard function has been published. A hazard function is a measure of the risk for a new event or for death in relation to time from the previous event, providing detailed information on the prognosis.
To study the natural history, or course of events, more knowledge is needed of the premorbid and postmorbid state of the patients, healthcare utilization before and after the stroke event, and more detailed information on the prognosis, especially its timing. A number of studies have reported on specific aspects of healthcare utilization after a stroke event,13,14⇓ whereas studies of the total healthcare utilization panorama before and after the event are needed for a better understanding of the course of events.
The aims of the present study were to compute hazard functions for survival and stroke recurrence in elderly stroke subjects and to analyze the total panorama of healthcare utilization and municipal social service support provided before and after a stroke incident.
The study was performed in the cities of Falun (population 55 000) and Borlänge (population 47 000), in central Sweden, with a similar age and sex distribution as the national Swedish population.15 The 2 cities are served by Falun General Hospital, the only one in the area. Because admission to the hospital in Sweden is free of charge for the patient, virtually all patients with clinical signs and symptoms indicating stroke are admitted to the hospital (except some already institutionalized).16,17⇓ Moreover, patient fees for hospital outpatient clinics, general practices, and municipal support are heavily subsidized by central and local governments, which means that private financial resources are seldom an obstacle to healthcare utilization. Moreover, all Swedish residents have a unique 12-digit personal identification number given at birth or immigration and used in all official documents and registers. The personal identification number is an excellent and highly reliable tool for record linkage.
The study population was previously described in detail.18 Briefly, the inclusion criteria were ≥65 years of age; discharged alive from the Department of Internal Medicine (stroke unit or general ward) during the period September 1, 1999, to May 31, 2001, after an acute stroke (index admission), defined as intracerebral hemorrhage, brain infarction, or stroke of undetermined pathological type (International Classification of Diseases, 10th Revision codes I61, I63, and I64)19; living in regular housing before admission; and having no preadmission dementia diagnosis. Of 432 potential participants, 42 died while in the Department of Internal Medicine. The 390 survivors constitute the study population of this report. Among both men and women, 88% had a cerebral thrombosis, and 97% had a CT scan.
Patients with stroke <65 years of age were not included to simplify data collection (no need of social insurance or work-related data), no institutionalized persons (not always admitted to the hospital), and no demented persons (to avoid interview problems). However, >80% of the total stroke population at the time was included in the study.20 All subjects in the study population gave oral informed consent to participation. The Research Ethics Committee at Uppsala University approved the study.
Of the 390 subjects, 13 died in the Geriatric Department, 6 died at home before the interview, and 2 moved from the area. One week after discharge, a registered nurse interviewed 351 patients, 18 declining interview because of severe illness. A Mini Mental State Examination (MMSE) was undertaken (normal score range, 24 to 3021). Self-rated health was measured with an ordinal scale ranging from 1 (poor) to 5 (excellent). Data on independence regarding transfer, personal hygiene, and dressing immediately before the stroke incident were obtained for 365 subjects from the National Stroke Register22 and the situation after discharge was assessed with Katz Activity of Daily Living index.23 Those with activity of daily living degree A were classified as independent, degrees B to E as partly dependent, and degrees F to G as completely dependent.
Information on day of admission, day of discharge, and all diagnoses for all admissions to the hospital from 1971 onward was obtained from the National Hospital Discharge Register, covering all hospital admissions in Sweden. Information on date of death and underlying cause of death until June 30, 2006, was obtained from the National Cause of Death Register. All diagnoses were coded according to the International Classification of Diseases, Versions 8 to 10.19
Information on date, diagnosis, and clinic for all hospital outpatient appointments from 1 year before to 1 year after the index admission was obtained from the Regional Hospital Outpatient Database. Information on date, diagnosis, and care provider for all primary healthcare center (PHCC) appointments and house calls were obtained by scrutiny of PHCC records in the area. Furthermore, municipal social service records and nurse records were scrutinized regarding type of support given and time period.
Data were analyzed with the SAS software package.24 Survival, healthcare utilization, diagnoses, and discharge destination data were 100% complete, and postmorbid state data, based on interviews, were 95% complete.
The stroke recurrence and death hazard functions were calculated by means of the SAS “lifetest” procedure providing coordinates for the hazard function and estimates of the influence of various factors on the hazard function. The actuarial method was used, and the Wilcoxon and log rank tests were used to test survival function across various groups. The cumulative hazard function was computed with Cox proportional hazards regression censoring for end of follow-up and death from causes other than stroke using age, sex, independence, and comorbidity (hypertension, myocardial infarction, number of previous strokes, other cardiovascular diseases, and diabetes) as exposure variables and providing hazards ratios (HRs) and their 95% CIs. Based on hazard function plots, the relation between the hazard function of the various age, sex, and stroke number groups was found to be approximately proportional.
The proportion of the study population in hospital care any day during the 5 years before the index admission was calculated. Being an inpatient (1) or not (0) was indicated in a set of variables, each variable representing a specific day during the 5-year period (day-by-day matrix). Summing the individual matrices and dividing each day variable by the number of exposed subjects produced the daily proportion of subjects in the hospital. The same procedure was used for the 5 years after the index admission except that exposure to hospital admissions was adjusted for nonsurvival.
The proportion of subjects using hospital outpatient and primary health care and municipal support during the year before and after the index admission was computed accordingly. During the year after index admission, exposure was adjusted for hospital admissions and nonsurvival. All tests on the matrices were done with logistic regression. Only 2-tailed tests were used. Probability values <0.05 were regarded as significant.
Characteristics of the Study Population
At the index admission, mean age was 78 years; half of the women and one third of the men were living alone in their own homes; and 84% to 90% could dress, maintain their personal hygiene, and transfer without assistance (Table 1).
Based on hospital record data from 1971 onward, 174 (78.0%) of the women and 117 (70.1%) of the men were hospitalized for their first-ever stroke. Half of the patients were cared for in the stroke unit, the remaining in a general ward, and more than half were later transferred from the internal medicine department to a geriatric department. The average length of hospital stay was 26 days. At discharge, 73% returned home, whereas the remaining subjects were discharged to various institutions or died while in the geriatric department.
One week after hospital discharge, mean MMSE level was 22.3 (interquartile range, 19 to 29) and well-being score was 3.7 (interquartile range, 3 to 4). The proportion of subjects independent in terms of transfer, personal hygiene, and dressing was substantially lower at discharge than before admission.
Survival and Risk of New Stroke
During the first year after the index discharge, 71 (18.2%) subjects died, and altogether 224 (57.4%) died during the first 5 years. During the first year, 55 (14.1%) subjects died from cardiovascular disease, in which 29 (7.4%) died from stroke, 18 (4.6%) from myocardial infarction, and 7 (1.8%) died from malignant disease. During the first 5 years, the corresponding numbers were 123 (31.5%), 58 (14.9%), 48 (12.3%), and 25 (6.4%), respectively. All other causes were infrequent.
Mortality during the first 5 years was significantly influenced by age (HR, 1.06; 95% CI, 1.04 to 1.08; P<0.0001) and functional ability (HR, 2.15; 95% CI, 1.81 to 2.54; P<0.0001), but not by sex, stroke number, or any comorbidity. The cumulative mortality rate in groups according to age and functional ability is shown in Figure 1. The cumulative mortality after 5 years ranged from 32.2% for subjects 74 years old at index admission, independent at index discharge, to 94.5% for those 83 years old and completely dependent on help at discharge.
The mortality hazard function is shown in Figure 2. The risk decreased rapidly during the first 90 days from 14.0%, achieved stability on the 4% to 5% level during the next 2½ years, and then increased slowly. The hazard function for stroke recurrence, also shown in Figure 2, decreased rapidly from 14.1% at discharge to a stable level of approximately 2% from 3 years on. The risk of stroke recurrence was mainly affected by the number of previous strokes (HR, 1.39; 95% CI, 1.27 to 1.53; P<0.0001), a hypertension diagnosis (HR, 0.72; 95% CI, 0.55 to 0.94; P=0.02), and sex (male to female HR, 0.79; 95% CI, 0.66 to 0.96; P=0.02), whereas functional ability, age, and other comorbidity variables had no significant effect.
Healthcare Utilization and Municipal Social Service Support
During the 5 years before the index admission, 269 (69.0%) subjects had at least one hospital admission, and 37 (12.1%) subjects had a new hospital admission within 28 days after the index admission, 201 (65.5%) within 365 days, and 291 (74.6%) had a new one within 5 years.
During the 5 years before the index admission, 1% to 2% of the study population were in hospital any given day (Figure 3). At index, the proportion became 100% and then fell back to a level of 2% to 3% 5 months after index. The corresponding proportions of hospital outpatient and PHCC appointments during the year before the index increased from 5% to 6% in the beginning of the period to 8% to 9% toward the end (P for trend <0.0001). After the index, it fell from 14% to 15% to 9% to 10% (P for trend <0.0001). The proportion receiving municipal support increased during the year before the index from 31% to 37% (P for trend <0.0001), fell to 0% during the index admission, and then increased rapidly to 60% 3 months after the index admission and to 65% over the rest of the year (P for trend <0.0001).
Overall, >90% had some contact with PHCC staff before and >99% after the index admission, and 58% and 75%, respectively, attended hospital outpatient clinics (Table 2). The most common contacts were with physicians and nurses or district nurses. The corresponding proportions using municipal support were 39% and 70%. The type of support given was approximately the same before and after index admission, but the levels were considerably higher after the index. The dominant types of support were related to home help, personal safety alarm, and transportation service as well as special housing services after the index. Age, number of previous strokes, and female sex increased the level of municipal support provided.
The mortality risk and the risk of recurrence were highest in the early postmorbid phase and then decreased considerably. Although a substantial proportion of the study subjects were admitted to the hospital in the period before and after the index admission, fairly few were hospitalized on a day-by-day basis. Considerably more subjects used outpatient hospital services, PHCCs, and municipal support, more after than before the index admission. Municipal social service support was by far the largest source of care and support.
All data were based on official registers with complete coverage or based on interviews with validated questionnaires and instruments and with a low attrition rate (5%). We therefore have no reasons to believe that the results were affected by selection or other bias to such an extent that the conclusions were affected.
The strengths of the study include the completeness of the stroke patient segment covered, the reasonably complete coverage of the total stroke population, and the broad and complete coverage of healthcare utilization data. The weaknesses include the fact that the study was performed in a northern European setting with fairly unlimited access to low-cost health care and municipal social service support. The results might have been different in other settings. On the other hand, the situation at hand permits a broad review of accurate needs.
The postmorbid state was considerably worse than the premorbid, especially regarding independence in transfer, personal hygiene, and dressing, similar to previous findings.25 MMSE levels at discharge were on average below normal. We had no access to MMSE levels at admission, but no subjects had a dementia diagnosis and we are therefore inclined to believe that the low MMSE level may be caused by the stroke event as previously suggested.26
Despite their weakened postmorbid state, three fourths of the subjects could return to their homes after discharge. Men were more likely to be discharged to their homes and less likely to be discharged to chronic care facilities than women, similar to previous findings,27 whereas women were more dependent on others and more often institutionalized.9 One explanation might be that the men were younger and often had someone to care for them at home.
According to the hazard function, the risk of death or recurrence decreased sharply during the first 3 months and then leveled off. Although the hazard function is a well-known (and the best) statistical measure of risk relative to follow-up time, it seems not to have been reported previously in patients with stroke. The finding in this report is therefore a novel one. Similar shapes of the hazard function for new events have been reported for acute myocardial infarction with similar timeframes.28 Others have drawn similar conclusions based on cumulative survival curves.12,29⇓ However, although the survival curve is based on the hazard function, the latter provides a more direct measure of mortality or recurrent event risk over time.
The cumulative mortality function, shown in Figure 1, illustrates the influence of age and independence at discharge on mortality. Similar findings have been reported from other studies.4,30⇓ We found no significant effect of sex or comorbidity such as previous strokes, other cardiovascular disease, or diabetes, although others have.3,6,8,12⇓⇓⇓
Despite the fact that a majority of the patients were admitted to the hospital during the years before and after the index admission, fairly few were hospitalized on any randomly chosen day. The same argument applies to hospital outpatient and primary health care. This is another novel finding of this study and reflects the fact that most of the patients stay in the hospital for fairly short periods, usually only a few days, and outpatient appointments last usually <1 hour. The day-by-day utilization chart gives a more correct view of the “burden” of these patients on the healthcare systems than cumulative proportions of subjects using health care over time.
The implications of this study include, as shown by the hazard functions, that stroke victims are highly vulnerable in the early phase after the incident. The risk then decreases substantially over the next few years rather than remaining at the same level or even increases. This is good news for patients and for doctors counseling their patients. Second, the largest burden of support to the patients with stroke, excluding support from relatives, was borne by the municipal health and social support services. The proportion of the subjects cared for in municipal care was >6 times higher than the proportion cared for by the regular healthcare services on a day-by-day basis. Because the municipal social service support and a considerable part of the health care given by the PHCCs was nursing care, this type of care dominated.
The risk of dying or having a stroke recurrence decreased rapidly during the early postmorbid phase. Healthcare utilization, both in hospitals and primary health care as well as municipal social service support, was considerably higher after the index admission than before. Among the various types of care given, nursing aspects dominated.
Sources of Funding
This study was supported by grants from the Vårdal Foundation, Centre for Clinical Research Dalarna, the Dalarna County Council, and Uppsala University.
- Received May 3, 2009.
- Revision received June 22, 2009.
- Accepted July 20, 2009.
- ↵Gorelick PB, Sacco RL, Smith DB, Alberts M, Mustone-Alexander L, Rader D, Ross JL, Raps E, Ozer MN, Brass LM, Malone ME, Goldberg S, Booss J, Hanley DF, Toole JF, Greengold NL, Rhew DC. Prevention of a first stroke: a review of guidelines and a multidisciplinary consensus statement from the National Stroke Association. JAMA. 1999; 281: 1112–1120.
- ↵Wolf PA, D'Agostino RB, Belanger AJ, Kannel WB. Probability of stroke: a risk profile from the Framingham study. Stroke. 1991; 22: 312–318.
- ↵Appelros P, Nydevik I, Viitanen M. Poor outcome after first-ever stroke: predictors for death, dependency, and recurrent stroke within the first year. Stroke. 2003; 34: 122–126.
- ↵Bae HJ, Yoon DS, Lee J, Kim BK, Koo JS, Kwon O, Park JM. In-hospital medical complications and long-term mortality after ischemic stroke. Stroke. 2005; 36: 2441–2445.
- ↵Campbell SE, Seymour DG, Primrose WR. A systematic literature review of factors affecting outcome in older medical patients admitted to hospital. Age Ageing. 2004; 33: 110–115.
- ↵Di Carlo A, Lamassa M, Pracucci G, Basile AM, Trefoloni G, Vanni P, Wolfe CD, Tilling K, Ebrahim S, Inzitari D. Stroke in the very old: clinical presentation and determinants of 3-month functional outcome: a European perspective. European Biomed Study of Stroke Care Group. Stroke. 1999; 30: 2313–2319.
- ↵Glader EL, Stegmayr B, Norrving B, Terent A, Hulter-Asberg K, Wester PO, Asplund K. Sex differences in management and outcome after stroke: a Swedish national perspective. Stroke. 2003; 34: 1970–1975.
- ↵Hankey GJ, Jamrozik K, Broadhurst RJ, Forbes S, Anderson CS. Long-term disability after first-ever stroke and related prognostic factors in the Perth community stroke study, 1989–1990. Stroke. 2002; 33: 1034–1040.
- ↵Hankey GJ, Jamrozik K, Broadhurst RJ, Forbes S, Burvill PW, Anderson CS, Stewart-Wynne EG. Five-year survival after first-ever stroke and related prognostic factors in the Perth community stroke study. Stroke. 2000; 31: 2080–2086.
- ↵Vernino S, Brown RD Jr, Sejvar JJ, Sicks JD, Petty GW, O'Fallon WM. Cause-specific mortality after first cerebral infarction: a population-based study. Stroke. 2003; 34: 1828–1832.
- ↵Samsa GP, Bian J, Lipscomb J, Matchar DB. Epidemiology of recurrent cerebral infarction: a Medicare claims-based comparison of first and recurrent strokes on 2-year survival and cost. Stroke. 1999; 30: 338–349.
- ↵Statistics Sweden. Population Statistics. Stockholm: Statistiska centralbyrån. Available at: www.Scb.Se/. Accessed October 1, 2002.
- ↵The Board of Directors of the National Stroke Register. Analysed report of the national stroke register for the years 1999 and 2000. Available at: www.Riks-stroke.Org/content/analyser/. Accessed 2002.
- ↵Olai L, Omne-Ponten M, Borgquist L, Svardsudd K. Prognosis assessment in stroke patients at discharge from hospital. Age Ageing. 2007; 36: 184–189.
- ↵World Health Organization. International Statistical Classification of Diseases and Related Health Problems. Tenth Revision: Geneva: World Health Organization; 1992.
- ↵The National Board of Health and Welfare. Statistical databases. Available at: www.socialstyrelsen.se/en/Statistics/Statistical_databases.html. Accessed April 22, 2009.
- ↵Statistical Analysis System. Cary, NC: SAS Institute Inc; 1995.
- ↵Pohjasvaara T, Erkinjuntti T, Vataja R, Kaste M. Comparison of stroke features and disability in daily life in patients with ischemic stroke aged 55 to 70 and 71 to 85 years. Stroke. 1997; 28: 729–735.
- ↵Kase CS, Wolf PA, Kelly-Hayes M, Kannel WB, Beiser A, D'Agostino RB. Intellectual decline after stroke: the Framingham study. Stroke. 1998; 29: 805–812.
- ↵Holroyd-Leduc JM, Kapral MK, Austin PC, Tu JV. Sex differences and similarities in the management and outcome of stroke patients. Stroke. 2000; 31: 1833–1837.
- ↵Gulliksson M, Wedel H, Köster M, Svärdsudd K. Hazard function and secular trends in the risk of recurrent acute myocardial infarction 30 years of follow up of more than 775 000 incidents. Circ Cardiovasc Qual Outcomes. 2009; 2: 178–185.
- ↵Dennis MS, Burn JP, Sandercock PA, Bamford JM, Wade DT, Warlow CP. Long-term survival after first-ever stroke: the Oxfordshire Community Stroke Project. Stroke. 1993; 24: 796–800.