Background
The proportion of the world’s population aged 60 and older is increasing rapidly. Within 35 years, it will have spiraled from 12% (in 2015) to 22% (in 2050). Virtually all the regions in the world are witnessing this population aging process [
1]. From a public health perspective, older adults (OA) account for 23% of the global burden of disease, particularly as regards chronic diseases (CDs) such as cardiovascular, pulmonary, musculoskeletal, neurological, mental, and neoplastic conditions [
2]. They have also been found to be at higher risk for multimorbidity and geriatric syndromes such as frailty [
3].
Regarding multimorbidity, associations with greater use of health-care services and adverse health events such as polypharmacy, increased health expenditures, disability, a low quality of life, and even mortality have been reported [
4‐
6]. However, multimorbidity research has traditionally focused on counting diseases, thus hindering the detection of co-occurrences. Notwithstanding, recent literature has highlighted the importance of taking into account the CDs combinations in OA health studies [
7].
Frailty, on the other hand, defined as a biological syndrome resulting from cumulative declines across multiple physiologic systems, with impaired homeostatic reserve and a reduced capacity of the organism to withstand stress [
8], has been identified as an independent predictor for adverse health outcomes including falls, a diminished quality of life, disability and death [
9‐
14].
Although OA can suffer from multimorbidity and frailty simultaneously, a causal link has been suggested between them [
15], empirical evidence has focused on the independent effects that they have on adverse health outcomes. Among the scarce evidence that has explored the joint effects of multimorbidity and frailty, one study with OA residents of Hong Kong found that combination of frailty and multimorbidity increased the risk of disability and death [
16]. However, it is still pending to analyze this interaction effect on other outcomes such as health-related quality of life. Therefore, our main aim in this study was to estimate the independent associations of multimorbidity and frailty with three different outcomes: disability, quality of life and all-cause mortality. A secondary aim was to determine whether exist a significant interaction effect of the multimorbidity and frailty on those same outcomes.
Statistical analysis
We used bivariate analyses to examine the relationships between the independent variables (frailty and multimorbidity patterns) and the dependent variables (disability, quality of life and mortality). We employed the following tests: Chi-square for categorical variables and ANOVA or Kruskall-Wallis for continuous variables.
Multimorbidity patterns
As we were interested in the CDs combinations more than a simple count of them, we performed a principal-components analysis of the nine CDs mentioned above to identify possible multimorbidity patterns. Given that the CDs were coded dichotomously, we built a polychoric correlation matrix, with the optimal number of patterns (components) determined according to the rule of eigenvalues greater than one. Then each CD was assigned to the component where its coefficient yielded the highest factor loading (score).
Disability and quality of life
To assess the relationships of the multimorbidity patterns and frailty with (1) disability and (2) quality of life, we applied linear regression models.
Mortality
To analyze mortality, we explored participant survival time based on the Cox proportional hazards model. We performed exploratory analysis by estimating Kaplan-Meier curves for all categorical predictors, and log-rank test for equality of strata to assess the predictors in the final model. We selected model predictors according to the exploratory analysis and prior evidence on OA survival.
Finally, we assessed the interaction effects between frailty and each multimorbidity patterns. We tested the statistical significance of this potential interaction effect by including a multiplicative term between frailty and multimorbidity patterns into the three regression models.
Data were weighted using post-stratified individual probability weights based on the selection probability at each stage of selection. Differences were considered statistically significant if p < 0.05. All statistical analyses were performed using STATA version 15.1 software (StataCorp. 2017. College Station, TX.).
Discussion
The results of this study show that frailty and multimorbidity were independently associated with the disability, quality of life and mortality after 5-years follow-up among a national sample of older Mexican adults. Frailty and multimorbidity rose the mean scores of the WHODAS and WHOQOL and also increased the risk of death. However, no significant interactions between frailty and multimorbidity were found.
Our study identified three multimorbidity patterns in Mexican OA, which were: cardiopulmonary, vascular-metabolic and mental-musculoskeletal. The evidence related to multimorbidity patterns has led us to conclude that these patterns are composed of diseases sharing a number of similarities from a clinical perspective.
Regarding the first pattern, cardiopulmonary condition, comprised of asthma, chronic obstructive pulmonary disease (COPD), and angina, a clinical and molecular link had been previously established [
21,
22]. For the second pattern identified, vascular-metabolic conditions, evidence support the relationship between metabolic syndrome (chronic inflammation, adiposity, etc.) and aging. This pattern includes CDs such as diabetes, arterial hypertension, stroke and cataracts [
23,
24]. Finally, the third pattern, comprising arthritis and depression (mental-musculoskeletal), can be explained by the link reported several decades ago between pain and mental illness, which suggests that they share a number of biological pathways [
25‐
27].
Our findings indicate that the three multimorbidity patterns analyzed were independent predictors for increased disability among OA. These results are consistent with the studies that have analyzed this link taking multimorbidity patterns into account. Jackson et al. (2015) studied a cohort of older Australian women and found that patterns similar to mental-neurological and cardiopulmonary conditions were associated with higher levels of disability (measured by the basic and instrumental activities of daily life - ADL and IADL, respectively) [
28]. Furthermore, Arokiasamy et al. reported an association between diabetes and hypertension combination and the presence of disability (+ 1 ADL) in adults 18 and older [
5]. Otiniano et al. (2003) found that the combination of diabetes and stroke increased the risk of disability (1 + ADL and 1+ IADL) in Mexican-American OA aged 65 and older, at five-year follow-up [
29]. Quinones et al. (2016) reported that the combination of arthritis, depression and hypertension in American OA was associated with higher levels of disability (combined ADL and IADL index) at two-year follow-up [
30].
Frailty was also an independent predictor for disability. This finding is also consistent with those reported in the literature [
9,
31], even allowing for different follow-up periods [
8,
10,
32,
33]. In the Mexican context, at 11-year follow-up, the cohort of OA aged 60 years and older from the Mexican Health and Aging Study (MHAS) indicated that frailty was a predictor for disability as regards ADL, but not IADL, while pre-frailty was a predictor only for restricted mobility, but not for ADL or IADL [
9]. Another study of urban OA aged 70 and older in the Mexican Coyoacan Cohort Study found that frailty increased ADL and IADL disability [
31].
In our study disability was measured using the WHODAS 2.0, which traditionally has been done using ADL and IADL criteria in older adult population. The use of WHODAS allowed that functionality spectrum of OA was enhanced [
34]. This means that vulnerability resulting from frailty in OA could affect other spheres beyond the physical dimension, inter alia, their interaction with others and their participation in society.
We also found that frailty and multimorbidity diminished the quality of life as measured in the follow-up study. These results are consistent with those of Arokiasamy et al., who reported that the combination of asthma and hypertension correlated with a diminished quality of life in adults aged 18 and older [
5]. Moreover there is evidence that depression among OA is independently associated with a lower quality of life [
35], and that adults with arthritis have a particularly high probability of suffering from a deteriorated quality of life [
36].
Our results indicate that baseline frailty status were an independent predictor for a deteriorated quality of life in the follow-up. There is scarce longitudinal evidence of an association between frailty and quality of life [
13,
37]. Even so, it has been hypothesized that this association could be bidirectional, baseline frailty could turns out to be predictive of a deteriorated quality of life, just as a low quality of life at baseline could be a predictor of frailty at follow-up [
37]. Our results appears to provide support for the first scenario, which means that a decreased functionality among frail OA affects their satisfaction in various areas (physical and social) that are measured with instruments such as the WHOQOL [
12,
13,
37]. However, this association must be deeply explored in future longitudinal studies with older adults.
Related to frailty and mortality, our results were consistent with previous studies that have identified frailty is an independent predictor for death. A systematic review and meta-analysis of longitudinal studies with OA using the frailty phenotype, found that frail subjects had a risk of dying two times higher than non-frail subjects. Pre-frailty also increased the risk of death, although the association was weaker [
38]. For older Mexican adults, this association was also reported by various studies. Specifically, Mexican Health and Aging Study (
MHAS) and 10/66 Dementia Study found that frailty was a predictor for death [
9,
39].
Observed association between the metabolic-vascular pattern and mortality, supports the reported evidence on the relationship between metabolic syndrome and the likelihood of die [
40], although recently it has been suggested that this association can be mediated by factors such as frailty [
41,
42] and sleep disorders [
43]. Nevertheless, even controlling for frailty, disability and other variables, it has been found that the metabolic-vascular combination has an independent effect on survival rate among OA.
A secondary objective of this study was to evaluate the potential interaction effects of multimorbidity and frailty. Although we did not find a significant interaction between these conditions, evidence suggests that a causal connection could exist, given they share common physiopathological mechanisms [
15]. Even so, few studies have explored the possibility of a combined effect. Among them, Woo et al. (2014) found that combination of frailty and multimorbidity increased the risk of disability and death in OA [
16]. Future research with OA that deeply explore on different combinations of chronic conditions will help to understand the potential interaction of frailty and multimorbidity and their effects on diverse health outcomes.
The results of our study should be interpreted taking into account the following limitations. First, the analysis of multimorbidity was confined to the nine high burden chronic conditions utilized in SAGE; it did not include diseases such as chronic kidney failure, cancer, cardiac conditions and dyslipidemia, all prevalent in older Mexican adults. Nevertheless, various studies using larger, equal or lower number of CDs, have found similar multimorbidity patterns. Second, potential selection bias may have resulted from differences between the analytical sample and excluded OA. Respondents in the study proved somewhat more affluent than respondents excluded (see Additional file
2: Table S3), it is not clear how this could affect our results, although it is possible that people with a higher economic level have greater knowledge about their health status, and then our prevalence of multimorbidity patterns could be underestimated. If the above were true, then our associations could be somewhat biased.