Introduction
Environmental gerontology proposes that people’s home environment (HE) influences their quality of life at older ages, since older people spend most of their time at home [
28]. If an individual’s loss of competence can no longer be compensated for by the existing HE conditions, or by the person’s subjective HE experience because of limited coping options, a so-called person-environment misfit occurs [
28]. This imbalance can lead to increased stress, restricted autonomy and health decline [
28]. As frailty is characterized by an increased vulnerability to external stressors [
13] and is associated with adverse health outcomes, including mortality [
2,
11], hospitalization [
16,
19] and nursing home admission [
6,
16], we assume that unfavorable objective HE conditions and negative HE experiences are associated with frailty status in very old adults (VOA). In line with the “Challenges and Potentials Model of Quality of Life in Very Old Age” [
27], this study focused on the link between environmental conditions and individual resources.
Two systematic reviews [
9,
12] have confirmed that living in socioeconomically deprived communities is associated with frailty; most of the studies used Fried’s definition of frailty [
11], known as the frailty phenotype (FP). Caldwell et al. [
5] reported that the physical deterioration of buildings in close proximity to peoples’ places of residence can have a negative impact on FP, whereas feeling secure and connected to one’s neighborhood is associated with lower FP levels [
5,
12]. A longitudinal study [
30] found that having a higher proportion of green spaces in the HE can be linked directly with improvements in FP, as well as indirectly through increased physical activity. In addition, higher levels of subjective esthetic quality and walkability of the HE was reported to have a protective effect against frailty [
29]. A few studies found that neglected interior living space can have a negative impact on an individual’s health status, physical and cognitive functioning, suggesting a poorly maintained HE as an indicator of a person’s frailty status [
8,
10].
There is broad evidence that frailty increases with age [
7,
21]. In Germany, a representative study of community-dwelling older adults reported a FP prevalence of 2.6% in the population aged 65–79 years [
4], while another study found a FP prevalence of 12.1% in those over age 65 years [
24]. The differences in these findings are attributable in part to their methodologies [
4]. As people who are aged 80 years or older or are institutionalized tend to be underrepresented in or even excluded from frailty research [
7], the first aim of this study was to analyze FP prevalence in the very old population of North Rhine-Westphalia (NRW). Based on the given research evidence, the second aim was to examine the association between subjective and objective aspects of the HE and the FP levels in VOA.
Results
Of the VOA in NRW, 24.3% (95% confidence interval, CI, 22.3–26.5%) were classified as robust, 57.0% (95% CI 54.6–59.4%) were found to be prefrail, and 18.7% (95% CI 16.9–20.7%) were found to be frail. Sample descriptions according to the four frailty criteria are provided in Supplement 2. As shown in Table
1, the frailty groups differed significantly in most characteristics. Frail persons were predominantly older, women, had lower socioeconomic status, no partner, poorer health status, more chronic diseases, needed more help with instrumental activities of daily living, rated walkability of external living environment less frequently as rather high/high, lived more frequently in poor residential areas, in neglected interior living space, were not very closely/not closely at all attached to their outdoor place, lived in nursing homes, and in communities with 500,000 or more inhabitants compared to robust and prefrail very old adults.
In the fully adjusted model (Table
2), holding all other variables constant, an improvement in the residential area (e.g., from rather poor/very poor to moderate) was statistically significant associated with a 41.6% decrease in odds of being frail. Being not very closely/not closely at all (compared to rather close/very close) attached to the outdoor place increased the odds of being prefrail by 62.7% and being frail by 131.7%. Living in communities with less than 50,000 (compared to 500,000 or more) inhabitants decreased the odds of being frail by 71.2%. Similarly, actual residence in communities with 100,000–499,999 inhabitants was linked to a 49.7% decrease in odds of being frail. The model estimates are presented in Table
2. A sensitivity analysis performed with community-dwelling individuals only revealed comparable results. For more details see Supplement 2.
Table 2
Multinomial regression model of the associations between home environment and frailty level (N = 1575a)
Walkability (Ref: rather high/high) | 1.010 | 0.754 | 1.352 | 0.790 | 0.524 | 1.191 |
Residential area | 0.807 | 0.607 | 1.074 | 0.584** | 0.396 | 0.860 |
Condition of interior living space | 0.884 | 0.664 | 1.178 | 0.802 | 0.542 | 1.187 |
Attachment to outdoor place (Ref: rather close/very close) | 1.627** | 1.144 | 2.316 | 2.317*** | 1.452 | 3.696 |
Place of residence (Ref: private household) | 0.623 | 0.340 | 1.143 | 0.488 | 0.231 | 1.032 |
Community type (Ref: ≥ 500,000) |
5000–49,999 | 0.944 | 0.613 | 1.454 | 0.288*** | 0.149 | 0.555 |
50,000–99,999b | 1.471 | 0.926 | 2.337 | 0.966 | 0.517 | 1.803 |
100,000–499,999 | 1.077 | 0.789 | 1.470 | 0.503** | 0.315 | 0.802 |
Pseudo R2 | 0.420*** |
Discussion
Our analysis showed that at 18.7%, the FP prevalence of the very old population of NRW was higher than that of younger older adults in Germany [
4,
24]. On the one hand, this confirmed the widespread empirical evidence that the risk of prefrailty and frailty increases with age [
7,
21]. On the other hand, the inclusion of nursing home residents could contribute to the higher FP prevalence (cf. [
6]). In other studies, nursing home residents were excluded [
4,
6,
16]. Furthermore, our findings support the assumptions of environmental gerontology [
28]: after adjusting for relevant individual health and socioeconomic characteristics, the quality of the respondents’ residential location (as assessed by interviewers), as well as their subjective attachment to their HE, remained significant predictors of their FP levels. Consistent with previous studies [
5,
12], not feeling connected to one’s own HE was associated with a higher likelihood of prefrailty and frailty. Wahl and Oswald [
28] argued that particularly at older ages, people’s subjective perception of their HE tends to reflect their personal identities. The failure to establish an attachment to the HE over a longer period of time may have a negative impact on an individual’s health and psychological well-being [
28], which may, in turn, lead to the development of frailty.
In line with previous research [
9,
12], our results showed that VOA who lived in poorer residential areas were more likely to be classified as frail. Thus, following Wahl and Oswald [
28], frailty might be seen as an outcome of person-environment misfit. For example, VOA who are functionally impaired may be unable to overcome an unfavorable HE (such as deteriorated footpaths), which can lead to reduced mobility outside the home or facility, and, in turn, to an increased risk of frailty [
5,
30]. Although there is an empirical evidence that an individual’s neighorhood affects frailty risk, conceptual frameworks explicitly explaining this relationship are still missing [
9,
12]. Unlike in previous studies [
20], our study did not confirm the association between individual socioeconomic status and frailty, supporting the age as leveller theory suggesting that socioeconomic disparities in health diminish in very old age [
17].
Inconsistent with previous studies reporting a higher likelihood of being frail among older population in rural areas [
18,
22], we found that VOA who were frail more likely lived in metropolitan communities. NRW is characterized by high urbanization levels [
3], therefore, rural communities were not included in our study sample. Since there are no existing studies of regional differences in frailty levels in Germany, this association should be further investigated using national samples.
In this study, a few aspects of the HE (walkability, place of residence, condition of the interior living space) were not found to be associated with FP. From a person-environment-fit perspective [
28], older individuals who are able to cope with HE challenges can protect themselves from the negative effects of these shortcomings on their health by adapting their behavior and routines. Hladek et al. [
15] found that a high level of coping self-efficacy can protect older people against prefrailty, independent of the individual health deficits. Thus, our results might indicate that VOA in NRW had sufficient coping capacities to enable them to adjust to these HE barriers. In further research, the moderation effects of coping efficacy on the link between HE and FP should be explored.
The present study has provided initial representative results on the FP prevalence and its relationship to the HE in VOA in Germany. Subjective as well as objective evaluations were used to precisely operationalize the HE. The definition of frailty we used was based on Fried et al. [
11] to ensure that our findings were comparable to those of other studies; however, minor modifications were necessary due to data availability. First, because the walking speed was not assessed in the NRW80+, four FP criteria were considered. Previous research has shown that four symptoms can predict the adverse health outcomes associated with frailty [
2,
19], and the exclusion of the criterion of walking speed has the smallest impact on predicting mortality [
26]. Second, we used frequency measures to operationalize physical activity, which differed from the original questionnaire based on expended kilocalories per week [
11]. The original questionnaire was recognized as being less suitable for vulnerable old individuals, as it was developed primarily for the younger population, and has been validated only for men and healthy older adults [
26]. Furthermore, interviews with proxies had to be excluded from the analyses because it was not possible to measure the handgrip strength of the target person. While the FP prevalence might be underestimated in this study because the proxy interviews were conducted in cases in which the target persons were suffering from health impairments, it was still found to be higher than in other German studies [
4,
24]. To minimize the possible bias, multiple imputation was applied to replace the missing values.