Changes in Functional Outcome Over the First Year After Stroke
An Observational Study From the Swedish Stroke Register
Background and Purpose—Large longitudinal studies on stroke outcome are scarce. The aim of this study was to analyze predictors and changes in functional outcome during the first year poststroke.
Methods—Data on patients who were independent in activities of daily living (ADL) and hospitalized for acute stroke in 2008 to 2010 were obtained from the Swedish Stroke Register. Case fatality was assessed by linkage to the Swedish Population Register. ADL was defined by independence or dependence in dressing, toileting, and indoor mobility and assessed at 3 and 12 months. Predictors of ADL dependency were assessed through multivariate analysis.
Results—In total, 64 746 patients were included. Case fatality at 3 months was 13.1% (men 11.6% versus women 14.8%; P<0.0001) and at 12 months 18.2% (men 16.4% versus women 20.3%; P<0,0001). In the 35 064 followed-up survivors, ADL dependency rates at 3 and 12 months were 16.2% (men 15.9% versus women 19.2%; P<0.0001) and 28.3% (men 22.7% versus women 34.9%, P<0.0001), respectively. Factors predicting deterioration to ADL dependency between 3 and 12 months were female sex (relative risk [RR]=1.56; 95% confidence interval [CI], 1.50–1.70), diabetes mellitus (RR=1.50; 95% CI, 1.05–1.60), comatose at admittance (RR=2.34; 95% CI, 1.79–3.05), previous stroke (RR=1.52; 95% CI, 1.43–1.61), hemorrhagic or unspecified stroke (RR=1.14; 95% CI, 1.05–1.25), and atrial fibrillation (RR= 1.11; 95% CI, 1.04–1.17).
Conclusions—Transition from ADL independence to dependence was observed in a high proportion of patients between 3 and 12 months, challenging the common belief that functioning after stroke is stable beyond 3 months. Deterioration occurred more commonly in women, among whom 1/6 converted to dependency.
Stroke is the second most common cause of death globally and the third most common cause of disability-adjusted life years lost. Over the last 2 decades, the absolute numbers of stroke survivors, disability-adjusted life years lost, and stroke-related deaths have been increasing. This trend is strongest in people aged ≥75. In 2010, the number of stroke survivors was estimated to 33 million, illustrating the effect of stroke on society.1–3
There are many reports on short-term survival and functional outcome after stroke,4–7 and assessment of outcomes at 3 months is common standard in acute stroke trials. Few studies have assessed functional outcome in the longer term, and in particular, longitudinal studies with multiple assessment points are scarce.8–11 Stroke recovery is heterogeneous but usually follows a common pattern, in which the largest regain of function occurs during the first weeks poststroke.12 Although recovery after stroke is usually regarded a phase of improvement until a plateau is reached,13 deterioration in disability level may occur further on. The proportions of patients who do deteriorate, and their characteristics, have not been well delineated in any large recent studies.
The aim of this study was to analyze case fatality and disability levels at 3 and 12 months, as well changes in functional outcome between 3 and 12 months, and predictors of dependency in activities of daily living (ADL), based on data from the Swedish Stroke Register.
Data were obtained from Riksstroke, the Swedish Stroke Register.14 In Sweden, the proportion of stroke patients treated in hospital is estimated at 84%,15 and in 2010, the estimated coverage rate of Riksstroke events was 88%.16
Patients >18 years, hospitalized between January 1, 2008, and December 31, 2010, with any one of the diagnoses (ICD-10) cerebral infarction (I63), intracerebral hemorrhage (I61), or unspecified cerebrovascular event (I64) were included. Only patients who were ADL-independent prestroke were included.
Data on vascular risk factors, functional ability, and living conditions were registered during the hospital stay, as well as secondary prevention, medical care, and planned rehabilitation at discharge.14 Validations of the Riksstroke data have shown over 90% consistency for most of the information from medical records and data entered into the Riksstroke database.17 Consciousness level at admittance was registered using the Reaction Level Scale RLS-85,18 with categories of fully awake, somnolent, and comatose. Data on living conditions and functional ability were collected using a questionnaire at 3 and 12 months (see online-only Data Supplement). At 3 months, some patients were offered a nurse appointment to help them fill out the questionnaire by interview (depending on local resources). The 12-month follow-up consisted of a postal questionnaire.
The main outcome was dependency in ADL at 3 and 12 months, respectively. Questions 3 (How is your mobility now? Independent/independent indoors/need help), 4 (Do you need help from someone to visit the toilet? Yes/no), and 5 (Do you need help getting dressed and undressed? Yes/no) were used in both the 3- and 12-month follow-up questionnaires (online-only Data Supplement).14,19 ADL-independent was defined as independent in dressing, toileting, and indoor mobility (being able to get around independently indoors). Dependent was defined as needing help with dressing, toileting or indoor mobility. Mortality status at 3 and 12 months was assessed by data linkage to The Swedish Population Registry. Analyses on case fatality were performed on the whole cohort.
The local ethics approval committee approved the project in 2012 (2012/453).
Patients Lost to Follow-Up
Patients lost to follow-up consisted of those who did not return the questionnaire or did not have valid social security numbers, valid addresses or those under protected identity. The 12-month follow-up was only sent once, meaning that if a patient had a recurring stroke the same year, only one 12-month questionnaire was sent.
Statistical analyses were performed in SPSS 21.0. Baseline data were analyzed using student’s t test for parametric normally distributed variables and χ2 test for categorical variables. We dichotomized the variable functional outcome into ADL-dependent and ADL-independent. Poisson regression was used for multivariate analyses of relative risks (RRs) of ADL dependency.20 Huber/White/Sandwich estimation was used to estimate covariance in the model.
The number of stroke events recorded from 2008 to 2010 was 75 048. A total of 8703 (11.6%) were ADL-dependent at baseline and excluded from the study. An inclusion/exclusion flowchart is presented in Figure 1. Within the first 90 days, 8483 died and an additional 3296 died between 3 and 12 months. At 3 months, 56 263 survivors were found in The Swedish Population Register and were contacted for follow-up, and 49 684 subjects completed it. The number of 12-month survivors who received the 12-month survey was 46 299. A total of 35 064 were followed up at both 3 and 12 months. The response rate at 3 months was 85.9% and at 12 months was 80.7%.
Baseline data including stroke characteristics and vascular risk factors in men and women are presented in Table 1. Among the 35 064 patients, there were 18 943 (54%) men and 16 121 (46%) women. Missing data for each variable was under 1%, except for smoking (6.6%).
In the total study population including 64 746 subjects fulfilling the basic inclusion criteria, case fatality was 8483 (13.1%) at 3 months (11.6% in men versus 14.8% in women; P<0.0001) and 11 779 (18.2%) at 12 months (16.4% in men versus 20.2% in women; P<0.0001).
Functional Outcome at 3 and 12 Months and Modes of Responding to Survey
At 3 months, 2767 (14.6%) men and 2896 (18%) women (P<0.0001; in total 16.2%) were ADL-dependent, whereas at 12 months, 4290 (22.6%) men and 5620 (34.9%) women (P<0.0001; in total 28.3%) stated dependency. Data on ADL status was missing in <2% of patients.
At 3 months, 17 356 (49.5%) completed the follow-up independently and 8721 (24.9%) completed it with help. A total of 5831 (16.7%) completed the form by nurse interview, either over telephone or in the clinical setting. In the remaining cases (8.1%), an assisting person (care giver, next of kin, other) completed follow-up without participation from the patient.
At 12 months, no nurse follow-up was offered. Patients completed follow-up independently in 21 847 (n=64.1%) cases, and in 8784 (25.1%) cases, patients completed the form with help. An assisting person completed the remaining 10.1%.
Data on who completed the questionnaire was missing in <3% of cases.
Predictors of ADL Dependency at 12 Months
Data from a multivariate analysis are presented in Table 2. The RR of ADL dependency at 12 months was higher in women compared with men (RR=1.31; 95% confidence interval [CI], 1.26–1.36). Other factors predictive of an unfavorable outcome were current smoking habit, atrial fibrillation, diabetes mellitus, decreased consciousness level at admittance, previous stroke, and stroke other than ischemic.
Changes in Functional Outcome Between 3 and 12 Months
Of the 28 683 patients who were ADL-independent at 3 months, 4544 (16.3%) deteriorated to ADL dependency at 12 months. ADL status in men and women, under and over 75 years, is presented in Figure 2. Deterioration occurred in all groups, but was most pronounced in women >75 years.
A comparison of those who deteriorated between 3 and 12 months to those who were stably independent between 3 and 12 months is shown in Table 3. Patients who were ADL-dependent at 3 months were not included in the comparison. Mean age was higher in those who deteriorated (79.01 compared with 71.13), the majority were women (61.4% compared with 41.8%), and a larger proportion was living in a 1-person household (54.8% versus 38.2%). The deteriorated group showed a higher prevalence of vascular risk factors (except for smoking), and especially atrial fibrillation (28.6% compared with 19.6% in the stable group).
In patients who were functionally stable between 3 and 12 months, 18 893 (80.2%) were living in their own home without assistance at 3 months and 20 525 (85%) at 12 months. In the deteriorated group, 2196 (48.3%) were living in their own home without assistance at 3 months and 1802 (41.9%) at 12 months. The proportion living in a nursery home was 312 (1.3%) and 404 (1.7%) at 3 and 12 months, respectively, in the stable group, although rising from 373 (8.2%) to 726 (16.9%) in the deteriorated group.
In the stable group, 18 893 (80.2%) completed the form independently at 12 months, whereas in the deteriorated group, only 1713 (39.1%) could fill out the form themselves.
Factors Predicting Deterioration Between 3 and 12 Months
Table 4 shows factors predictive of deterioration from ADL independency to dependency between 3 and 12 months in patients who were ADL-independent at 3 months. Female sex predicted deterioration with a RR of 1.60 (95% confidence interval, 1.50–1.70). Other factors predicting deterioration were current smoking habit, diabetes mellitus, decreased consciousness level, previous stroke, and hemorrhagic stroke.
Lost to Follow-Up
Reasons for lack of follow-up are described under Methods. Baseline characteristics were compared for patients lost to follow-up versus patients who completed follow-up at 3 and 12 months, respectively (Tables I and II in the online-only Data Supplement).
At 3 months, 6579 (11.8%) of survivors did not return the 3-month follow-up questionnaire for reasons not further specified. In unadjusted data, there were higher proportions of patients with previous stroke (30% versus 22.1%) and patients living alone (51.8% versus 45.3%) than in the group not followed up. The proportion of patients with decreased consciousness levels at admittance was higher in the group not followed up (17.3% versus 8.5%).
Correspondingly, of 12-month survivors, 15 574 patients were lost to follow-up, of whom 6668 patients never received the questionnaire because of missing addresses, invalid social security numbers, protected identity, or recurrent stroke. Compared with the group that was followed up, the proportion of previous stroke was higher (34.6% versus 18.3%) in the group not followed up. The proportion of patients living in a 1-person household was higher in the group not followed up (52% versus 42.4%).
Our results show that from 3 months onwards, functional outcome is not stable over time. In this large longitudinal study of 12-month survivors, 1 in 6 patients deteriorated in functional outcome between 3 and 12 months poststroke. Women >75 years were most susceptible to deterioration, with dependency proportion levels rising from 23.2% to 45.5%. The RR of ADL-dependency at 12 months was 31% higher in women, and women were also more susceptible to deterioration. Other independent predictors of unfavorable outcome were diabetes mellitus, atrial fibrillation, smoking, decreased consciousness level at admittance, and hemorrhagic or unspecified stroke.
Previous longitudinal studies have shown that neurological recovery in terms of body functions is largely complete within 3 months poststroke.12,21 However, there are only a few studies on stroke cohorts with repeated measurements of functional long-term outcome, and all were confined to patients admitted to rehabilitation units rather than unselected cohorts.12,22–24 Deterioration in mobility was seen in 43% between discharge and 1 year follow-up.22 The smaller proportion (16%) that deteriorated in the present study might reflect differences in study population, that is, the majority in an unselected stroke cohort consists of minor strokes, hence a large proportion of ADL independent. Another problem is the different outcome measures used in the studies, making direct comparisons difficult. Motor recovery in terms of functioning (ADL) was not stable beyond 3 months in this cohort, whereas motor function in terms of indoor mobility showed little change between 3 and 12 months (data not shown). Dressing and toileting, perhaps requiring a larger complexity of motor activity, accounted for all deterioration from ADL independency to dependency.
In the present study, the group that deteriorated between 3 and 12 months showed a higher proportion of females and individuals with higher age, higher proportions of atrial fibrillation, diabetes mellitus, previous stroke, and subjects living alone. Probably because of high age, those who deteriorated were discharged with a lower proportion of statins in unadjusted data. In patients with atrial fibrillation, the proportion treated with warfarin was only 35% in those who deteriorated compared with 58% in the stable group (Ullberg T, personal communication).
Previous studies on sex differences in stroke outcome have shown conflicting results that may partly be explained by geographic differences in incidence, risk factor profiles and stroke genetics, and hospital and rehabilitation treatment.11,25–27 The results of our study concur with studies that found female sex to be predictive of ADL dependency after stroke.27–33 In a systematic review by Gall et al, it was hypothesized that women may be more vulnerable than men to worse outcomes because of differences in demographic, social, and medical histories.34 In the present study, women were ≈4 years older than men and a slightly higher proportion had atrial fibrillation. Moreover, in those who deteriorated after 3 months, the majority was female and almost a third had atrial fibrillation. Previous studies have shown more severe strokes in women35 and a higher proportion of cardioembolic strokes,31 and additionally, female sex has been found to be an independent risk factor in cardioembolic stroke.36 That may in part contribute to the higher RR of disability at 12 months in women. Despite the fact that this study excluded patients with prestroke disability and adjusted for possible confounders, there was still a small but significant effect of female sex on risk of long-term disability.
Strengths of the study include the large sample size and the high coverage rate of Riksstroke (all hospitals and >80% of acute stroke patients in Sweden). Therefore, selection bias is expected to be low, and the sample should be representative of the Swedish stroke population. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statements.37 Dependency in activities of daily living is a robust outcome measure, in this case using the core variables of functioning (dressing, toileting, and mobility).
Although we used the core variables of the modified Rankin Scale and the Barthel Index, results might differ slightly from the full scales, and this can be considered a limitation of this study. Another limitation was the difference in follow-up at 3 and 12 months. For practical reasons, the follow-up procedures in Riksstroke are based on questionnaires. Local resources permit nurse follow-up in a minority of participating hospitals. At 3 months, some patients were offered a nurse interview (completed in 17%). Our clinical experience is that, as compared with the answers given during nurse interview, patients tend to underestimate their disability when self-reporting. In the most disabled, a caregiver or next of kin filled out the questionnaires, and it cannot be ruled out that this might have influenced the responses, but it is unlikely that dependency was overrated. Furthermore, we lack information on the deteriorated group, including data on recurrent strokes or hospital admissions for other reasons. There were a fairly large number of patients lost to follow-up both at 3 and 12 months (attrition bias), representing the 1-year survivors with the worse outcome. Therefore, the result of the study is most likely an underestimation of ADL dependency at both 3 and 12 months.
Transition from ADL independence to dependence was observed in a high proportion of patients between 3 and 12 months. This finding challenges current beliefs that a 3-month follow-up can be used as the sole standard time point for evaluation of stroke outcome. Deterioration was common in the elderly and in women, in particular. With a growing proportion of people age ≥75, this becomes an important matter. Probable explanations for deterioration are high rates of comorbidities in the elderly and social isolation.
We thank all participants and their caregivers. We also thank Maria Berglund at Riksstroke.
Sources of Funding
The Swedish Stroke Association and the Region Skåne founded this study.
Guest Editor for this article was Bruce Ovbiagele, MD, MSc, MAS.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.006538/-/DC1.
- Received June 22, 2014.
- Revision received November 6, 2014.
- Accepted November 11, 2014.
- © 2014 American Heart Association, Inc.
- Krishnamurthi RV,
- Feigin VL,
- Forouzanfar MH,
- Mensah GA,
- Connor M,
- Bennett DA,
- et al
- Vaartjes I,
- O’Flaherty M,
- Capewell S,
- Kappelle J,
- Bots M
- Feigin VL,
- Forouzanfar MH,
- Krishnamurthi R,
- Mensah GA,
- Connor M,
- Bennett DA,
- et al
- Andersen KK,
- Andersen ZJ,
- Olsen TS
- Kelly PJ,
- Crispino G,
- Sheehan O,
- Kelly L,
- Marnane M,
- Merwick A,
- et al
- Luengo-Fernandez R,
- Paul NL,
- Gray AM,
- Pendlebury ST,
- Bull LM,
- Welch SJ,
- et al
- Hardie K,
- Hankey GJ,
- Jamrozik K,
- Broadhurst RJ,
- Anderson C
- Kwakkel G,
- Kollen B,
- Twisk J
- Glader EL
- Eriksson M,
- Appelros P,
- Norrving B,
- Terént A,
- Stegmayr B
- Zou G
- van de Port IG,
- Kwakkel G,
- van Wijk I,
- Lindeman E
- Glader EL,
- Stegmayr B,
- Norrving B,
- Terént A,
- Hulter-Asberg K,
- Wester PO,
- et al
- Di Carlo A,
- Lamassa M,
- Baldereschi M,
- Pracucci G,
- Basile AM,
- Wolfe CD,
- et al
- Gall SL,
- Tran PL,
- Martin K,
- Blizzard L,
- Srikanth V
- Friberg L,
- Benson L,
- Rosenqvist M,
- Lip GY
- von Elm E,
- Altman DG,
- Egger M,
- Pocock SJ,
- Gøtzsche PC,
- Vandenbroucke JP