Background
As the number of older people in western populations rises, stroke prevalence is expected to increase [
1]. In both men and women, stroke rates increase exponentially with age [
2]. Stroke is a main cause of disability and care dependency in adults [
3,
4], and age is also known to play an important role in post-stroke outcomes. Glader et al. analysed data from 19,547 patients included in the Swedish National Quality Register for Stroke Care [
5]. Their results show that, among patients who lived at home before their stroke, age was a strong predictor for living in institutional care 3 months after stroke—with and without control for other variables. In a study of patients recovering from stroke in a long-term rehabilitation hospital, Koyama et al. [
6] found that older age increased the odds of discharge to a nursing home rather than directly to the patient’s home. Their correlation analyses further revealed that older age was associated with female sex, ischaemic stroke, lower scores on the Functional Independence Measure (FIM-m), smaller household size, and a higher number of sons/daughters.
As pointed out by Hankey et al. [
7], although advancing age is known to be strongly associated with increasing levels of disability and an increasing number of comorbidities, few studies have examined the independent effect of age on post-stroke outcomes. Irrespective of stroke, the burden of comorbidity is known to be substantial in older adults. In Germany, it is assumed that one in three adults aged over 70 years has five moderately severe conditions and that almost one in four is in treatment for five conditions concurrently. In the German Ageing Survey 2002, 24 % of respondents aged 70+ years reported that they suffered from five or more conditions, whereas only 7 % did not report any conditions at all [
8]. Stroke survivors are often affected by stroke-related conditions such as aphasia and hemiparesis. Thus, it could be assumed that older stroke patients are at a higher risk of becoming care dependent because they have more comorbidities.
Consequently, in our study we addressed two research questions: (1) How strong is the association between age and care dependency risk 1 year after stroke? (2) Can this association be (partly) explained by burden of disease?
Discussion
This study analysed associations between age and care dependency risk after stroke. Together with findings from other studies, our results indicate increasing care dependency rates with older age [
4,
6,
7,
9]. Relative to the reference group (0–65 years), patients aged 86+ years were 11.30 times more likely (95 % CI: 7.82–16.34) to be care dependent 1 year after stroke, and patients aged 76–85 years were 5.10 times more likely (95 % CI: 3.88–6.71). Even after the stepwise inclusion of gender, stroke subtype, history of stroke in the previous year and comorbidities, these age effects remained largely stable. In other words, the care dependency risk increases with age, irrespective of the presence of all other variables. Only the inclusion of history of stroke in the previous year and gender minimally reduced the effect of age. Age-adjusted gender was not associated with care dependency risk after stroke. Several studies investigating gender-specific differences in stroke care have attributed these differences to female patients’ more advanced age [
4,
13]. Other age-adjusted studies have also found that gender is not significantly associated with further disease progression after stroke [
14,
15].
Furthermore, we examined whether age effects could be explained by burden of disease. It is well known that older people typically have several chronic conditions concurrently [
7,
8] and that the prevalence of multimorbidity increases substantially with age [
16]. Multimorbidity affects more than half the elderly population, with increasing prevalence in very old persons [
12]. At the same time, consistent with our results, studies suggest an association between post-stroke care dependency and burden of disease. A large European study of hospital admissions for acute stroke found that swallowing problems and urinary incontinence were significantly related to disability and handicap 3 months after stroke [
17]. Other studies have found stroke subtype, prior stroke event or pre-stroke disability to have a significant effect on care dependency risk [
4,
6,
7,
18]. Against this background, we hypothesised that the association between age and care dependency risk the first year after stroke would be mediated by burden of disease. However, the results of our stepwise regression analyses revealed that only with the inclusion of prior stroke events the odds ratios of age were reduced minimally. Apart from this result, age associations were not reduced after adjustment for burden of disease. On the contrary, inclusion of both comorbidities and geriatric mulitmorbidity increased the effect of age on care dependency risk (by between 1.1 and 28 %).
This result raises a question of which other factors may account for the remaining age effects. It seems probable that social factors such as living in a partnership or alone, social contacts or social support may play a role. As several studies have reported associations between social factors and care dependency risk and between age and social factors, it seems likely that social factors mediate the association between care dependency risk and age. Appelros et al. [
18] examined the association between living setting and need for assistance with activities of daily living before and 1 year after a first-ever stroke. Living alone at baseline increased the odds of living in a service flat or nursing home 1 year after stroke 2.7-fold (95 % CI:1.4–5.1). The authors concluded that this finding to some extent could be attributed to the fact that there are more single women than single men: Spouses take great responsibility for helping their partners after a stroke, and males receive more personal help from their female spouses than vice versa. By the same token, Koyama et al. [
6] found that patients without a spouse at home who lived in smaller households were more likely to be discharged to a nursing home after a stroke. Furthermore, as mentioned above, they found that age was negatively correlated with the number of household members [
6]. To sum up, social conditions such as living together with a spouse are protective factors regarding (post-stroke) care dependency risk. As older age leads to a loss of social contacts and partnerships, it seems likely that the association of age with care dependency after stroke found in our study may be partly explained by such social factors. Unfortunately, we were not able to investigate the effects of social contacts and partnerships on care dependency risk after stroke, as the health insurance data did not include this information for data protection reasons.
Other possible explanations for the strong age effects revealed in our study are neurochemical and physiological changes in older age. Although it is well known that vulnerability for negative outcomes increases with age due to a decreasing reserve of the physiological systems, little is known about the underlying age-dependent mechanisms. Findings from a recent study indicate that hormones modulate the age-dependent differential stroke outcomes [
19]. Further studies are needed to investigate whether such neurochemical effects associated with age also increase the risk of becoming care dependent (after stroke).
Our results also show that geriatric multimorbidity has a significant effect on post-stroke care dependency risk. This finding has clear practical implications, because there is potential to influence the progression and extent of comorbidities. Internationally, treatment in a specialized interdisciplinary stroke unit, followed by inpatient rehabilitation, is the gold standard [
20]. Outpatient services are also essential in safeguarding continuity of care and mitigating the long-term effects of stroke. However, in Europe, including Germany, outpatient rehabilitation is underutilized [
20]. The provision of in- and outpatient rehabilitation services should therefore be increased to reduce stroke-associated comorbidities and thus decrease the post-stroke care dependency risk. Further studies are needed to explore the pattern and characteristics of geriatric morbidities in more detail.
The limitations of our study include the lack of data on social factors in the data set used. Moreover, it was not possible to differentiate between comorbidities that were already present before the stroke and those that were caused by the stroke. Furthermore, it is questionable whether the list of symptom complexes characterizing geriatric multimorbidity covers all aspects of disease burden; we cannot rule out the possibility that the number of diagnoses was underestimated. Divergences from the findings of other studies may also be attributable to differences in methods of data assessment: the diagnoses were ascertained from claims data and not made by specially trained personnel in the context of a primary data assessment. However, the approach to assess geriatric multimorbidity applied by the German Geriatric Association has proved to be generally practicable and successful [
21]. Finally, it remains unclear to what extent the present data can be considered representative of the general German population. The Deutsche BKK is a German company health insurance fund, with around 1.1 million members [
22]. Previous analyses of the Deutsche BKK data have shown nearly the same distribution of the stroke subtypes as reported in other sources (e.g., regional stroke registers and, in particular, national DRG data). Regarding sex distribution, however, women in the age groups older than 60 years have been found to be overrepresented [
23]. The company health insurance funds have been found to provide among the best representations of the German population with respect to socio-demographic indicators such as age distribution orEast/West ratio [
24].
We used the standard definition of care dependency applied in Germany, i.e., the patient was receiving long-term benefits from their care insurer. Although this definition is not in line with major international operational definitions of care dependency, the prevalence identified in our study (17.5 %) is similar to other studies. In international studies, the onset of care dependency is defined in terms of the degree of physical impairment (modified Rankin Scale), reduced independence (Barthel Index) or place of residence (at home, at home with carers’ support, nursing home). The available data are mainly drawn from retrospective surveys. The findings of these studies indicate that 15–19 % of previously independent persons are so impaired by stroke that they are reliant on others’ help for daily living (long-term care) [
18,
13,
25].
The strengths of our study include the fact that claims data are not subject to non-response. This is a major advantage when investigating populations with a high proportion of people with limited communication abilities. Use of claims data also excludes memory errors, which is particularly important in studies with very old respondents. Moreover, the study design excludes the possibility of institution-related selection bias and drop-out at follow-up. As such, the data examined are guaranteed to be complete [
26].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SS, AK, LS participated in the conceptualization and design of study. OK helped to specify the research question and to shape the analyses. MK, DP participated in data acquisition. SS conducted the statistical analysis. SS interpreted the data. SS drafted the manuscript. OK, MK, DP, AK, LS critically revised the manuscript. All authors have read and approved the final manuscript.