Life satisfaction among elderly patients in Nepal: associations with nutritional and mental well-being

The current study, NepEldQOL II, is the second of a two-part study which aimed to evaluate the overall quality of life of Nepalese elderly. The NepEldQOL I research is fully described elsewhere (Ghimire et al, Health-related quality of life among Nepalese elderly, forthcoming). Briefly, a cross-sectional study was conducted from January–April 2017 in the Out-Patient Department (OPD) of the Nepal Medical College and Teaching Hospital in Kathmandu, Nepal.

The sample size for this cross-sectional study was calculated in StatCalc in Epi Info 7, (sample size = z²pq/d²). Selecting 0.05 as an acceptable error rate (alpha) and 95% confidence intervals, and using the 24% prevalence of malnutrition among Nepalese elderly found in another study [20], we obtained a sample size of 289. Criteria for inclusion in the study included being at least age 60, attending clinic on one of the survey days, and consenting to participate. Those too frail physically or mentally to respond, and/or with hearing or speech impairment were excluded. Participants were selected by systematic random sampling; from the OPD patient list, every third patient attending clinic on one of the survey days was approached for eligibility and consent. If the approached patient was deemed ineligible, then the next eligible respondent from the OPD list was approached. Written and/or verbal consent was obtained from participants, depending upon the literacy status of the subject; there were no refusals. Data were collected until required sample size was reached.

Data were collected during patients’ wait time using individual face-to-face interviews conducted by trained graduate students in medicine and public health.

Participants’ life satisfaction was measured with the widely-used Satisfaction with Life Scale (SWLS) [20]. SWLS has good convergent validity with other types of assessments of subjective well-being [21]. Cronbach’s alpha, a measure of scale reliability, for the SWLS in this study was 0.93. In short, the SWLS is a 5-item scale; response for each item is captured by a 7-point Likert scale of agreement. The cumulative SWLS score ranges from 5 to 35 with higher scores indicating higher levels of life satisfaction. Cumulative SWLS scores are dichotomized into two categories: satisfied (SWLS> 20) and dissatisfied (SWLS≤20) [20].

Nutritional status was assessed using the Mini Nutritional Assessment – Short Form (MNA-SF) [22], a tool previously validated in multiple settings [22], including Nepal [23]. The Cronbach’s alpha for the MNA-SF scale was 0.59. A detailed description of the MNA-SF tool is provided elsewhere [22]. Briefly, the cumulative scores of the six MNA-SF items (ranging from 0 to 14) are classified into three groups: malnourished (MNA-SF score < 8), at risk of malnutrition (MNA-SF score of 8–11), and normal nutritional status (MNA-SF ≥ 12) [22].

Depression was measured using the short form of the Geriatric Depression Scale (GDS) [24], another tool which has been validated in Nepal specifically [25], as well as many different settings [26]. The GDS has been extensively described elsewhere [24]. Briefly, it is a 15-item instrument; each item has a dichotomized response in “Yes/No” format. The cumulative score of the GDS ranges from 0 to 15: 0–4 suggests no depression; 5–9 suggests mild depression; 10–15 suggests severe depression [24, 27]. In this study, the Cronbach’s alpha for the GDS scale was 0.81.

Height and weight were measured using a mechanical stadiometer and a digital weighing scale, respectively. Subsequently, body mass index (BMI) was calculated. Participant’s weight status was characterized as underweight (BMI < 18.5) normal weight (18.5 ≤ BMI < 23), overweight (23 ≤ BMI < 27.5), and obese (BMI ≥27.5) as per the World Health Organization (WHO) recommendation for Asians [28]. Socio-demographic variables assessed by self-report were age (continuous and categorized in 5-year intervals), sex, ethnicity (three groups: Upper Caste, Janjatis, or Dalit/other ethnic minority), marital status (married or separated/widow/single), family’s monthly income (continuous and categorized as < $200, $200- < $300, and ≥ $300), family structure (nuclear, joint or extended), smoking, alcohol use, and education level, which was categorized as illiterate, informal (no formal schooling but some literacy), and formal (any formal education). Also assessed by self-report were participants’ self-perceived health status (better/similar/worse) compared to peers of their age as well as perception of being ignored/hated for being old (yes/no).

Data management and analyses were done in EpiData and IBM SPSS22 (SPSS Inc. Chicago IL, USA), respectively. Continuous variables were described with mean ± standard deviation (SD); categorical variables with frequencies. Differences in mean and frequency distributions between the groups were assessed using independent t-tests and Pearson’s chi-square (χ²) tests, respectively. The correlation of the SWLS score with the MNA-SF and GDS scores was given by Spearman’s correlation coefficient (ϱ). Predictors of life satisfaction were assessed in linear regression models using cumulative SWLS scores as the dependent variable. Models were adjusted for age, sex, marital status, and family type. We used bootstrap models with 5000 replications to calculate stable estimates of correlates (Tables 2, 3 and 4).

A mediational analysis [29] was conducted to determine the direct and mediating associations of depression and nutritional status on life satisfaction. Two different mediating models were developed to examine whether, and to what extent, the mediator, (nutrition in Model 1 and depression in Model 2), explained the relationship between the exposure (depression in Model 1 and nutrition in Model 2) and SWLS as the outcome. For moderation analysis, an interaction between the GDS and MNA-SF score was added to the regression model with the GDS and MNA-SF scores as predictors and SWLS as the outcome. For testing our mediation-moderation models, we used the non-parametric bootstrap approach which allows decomposition of the total effect of an exposure into natural direct and indirect effects [29]. To be consistent with the terminology used in mediation analysis [29], we use the term “effect” rather than “association” for the findings pertaining to mediational analyses. However, based on the cross-sectional nature of our study, no conclusions should be made regarding the direction of causality. The PROCESS macro for SPSS, which tests mediation models by comparing the observed indirect associations against 5000 bootstrapped resamples, was used for the mediation-moderation analyses [29]. The coefficients of associations and their bias-corrected and accelerated (BCa) 95% confidence intervals (CI) were obtained from 5000 random bootstrap samples. BCa confidence intervals that do not include zero indicate statistically significant estimates [29]. The mediation analyses were first run without any covariates (Model 1), then adjusted for age and sex (Model 2), and further adjusted for age, sex, ethnicity, marital status, educational status, smoking, alcohol use, perception of being ignored/hated, and perceived health status compared to peers (Model 3). For all statistical tests in Table 1, two-tailed p-values< 0.05 were considered statistically significant.

In adjusted analyses (Table 2), married participants were significantly more satisfied with their life than those who were separated/widow/single (β = 2.19; BCa 95% CI = 0.28, 4.18). Compared to participants living in a joint family, those living in a nuclear family were significantly less satisfied (β = − 2.37; BCa 95% CI = − 4.79, 0.01). Family monthly income was positively associated with life satisfaction (β = 0.01; BCa 95% CI = 0.01, 0.02). Likewise, working participants were more satisfied with life than those retired or unemployed (β = 2.24; BCa 95% CI = 0.27, 4.17). The perception of worse health than peers (β = − 3.53; BCa 95% CI = − 4.94, − 2.08) and the perception of being ignored/hated due to old age (β = − 3.95; BCa 95% CI = − 6.07, − 1.88) were associated with a lower life satisfaction score. Life satisfaction was positively associated with the MNA-SF nutritional score (β = 0.90; BCa 95% CI = 0.65, 1.17) and inversely associated with the GDS depression score (β = − 1.00; BCa 95% CI = − 1.14, − 0.85) (Table 2).

In moderation analysis, there was no statistically significant interaction between nutrition and depression (β = 0.03, BCa 95% CI: -0.02, 0.08).

Mediation analysis with depression, as captured by the cumulative score on the GDS assessment, as the mediator, revealed a partial mediation in the relationship between nutrition and life satisfaction, through depression (Table 3). In the final model, adjusted for age, sex, ethnicity, marital status, smoking, alcohol use, educational status, the perception of being ignored/hated and perceived health status compared to peers, a significant indirect association was found in the mediation model (β = 0.30, BCa 95% CI: 0.13, 0.49). Depression mediated 38% of the total associations of nutrition on life satisfaction (Table 3).

Mediation analysis (Table 4), with the MNA-SF nutritional score as the mediator, revealed a partial mediation in the relationship between depression and life satisfaction, through the MNA-SF score. In the final model, adjusted for age, sex, ethnicity, marital status, smoking, alcohol use, educational status, the perception of being ignored/hated, and perceived health status compared to peers, a small but significant indirect association was found in the mediation model (β = − 0.07, BCa 95% CI: -0.14, − 0.03). The nutritional score mediated 7% of the total association of depression on life satisfaction (Table 4).

This is the first study not only to assess the life satisfaction of Nepalese elderly but also to explore the relationship of three important aspects of biological aging- depression, nutritional status, with life satisfaction - in the Nepalese context. Further, to our knowledge, this is the first study to explore the mediating role of depression and elderly malnutrition on life satisfaction among the elderly in any context. The study findings suggest that life satisfaction is truly multi-faceted [40], and depression plays a crucial role. Further prospective studies are needed to identify the direction of this relationship.

Naturally, our study is subject to some limitations. Due to the cross-sectional nature of our data, no inferences should be made with regard to the direction of causality; the possibility of reverse causation cannot be ruled out. We acknowledge that mediation models are best suited for longitudinal data. However, through this analysis, we provide preliminary support to the hypothesized mediating role of depression on life satisfaction. Future research should evaluate this hypothesis with prospective data. There may be other mediators in the pathway between nutrition and/or depression and life satisfaction which we were not able to consider in this study; future studies should explore other potentially important mediators. Our study participants were patients attending outpatient clinics in an urban setting; however, the nutritional status, depression, and life satisfaction of the general population and those in a rural setting may differ significantly from them, which would impact the generalizability of our results. Lastly, the internal consistency of the MNA-SF scale was relatively low in our study (Cronbach’s α = 0.59); however, given that the MNA-SF has already been validated among the Nepalese elderly population [23], we accept it as a valid nutritional assessment tool for this context.