Aim, design, and settings of the study
We conducted this case-control study to assess the association between malnutrition and GD, and ascertain the associated risk factors in rural elderly of Bangladesh. The study was conducted from January to October 2019 in three rural villages; Dhankhali, Tiakhali, and Lalua in Kalapara Upazila of Patuakhali district in Bangladesh. The villages are located in remote areas in the coastal belt of the country bedsides the Bay of Bengal [Figure 1].
Participants
The cases were defined as elderly aged 60 years or more having depression, and controls were elderly having no depression. We identified the cases and controls based on the score estimated by the Geriatric Depression Scale-15 (GDS-15). An elderly with a GDS score of <5 was classified as a control while an elderly participant with a score of ≥5 was classified as a case. The study participants were included in the study having criteria such as; (i) a permanent resident of the selected rural community; (ii) aged ≥60 years; (iii) capable of giving informed written (or verbal) consent, and (iv) absence of severe illness.
Sample size and sampling
Without considering the matching and considering 10% non-response rate, and rounding, a sample size of 600 elderly individuals would allow identifying an odds ratio of 2 for exposure of 10.6% among controls (confidence level = 0.95 and power = 0.80). We recruited 300 elderly individuals having depression as cases and 300 individuals without depression as controls who were individually matched for age (±2 years) and neighborhood (residence of cases and controls in the same rural community). We recruited 200 elderly individuals (100 cases and 100 matched controls) from each village and a total of 600 elderly individuals (300 cases and 300 matched controls) from three villages following systematic random sampling. In the beginning, the principal investigator estimated the sampling interval using the total households and sample size in each village. Considering the sampling interval, the data enumerators attended the selected households for finding the elderly individuals living there. Based on the selection criteria, we identified the expected number of cases and matched controls for recruiting in the study.
Data collection
A semi-structured questionnaire was applied for data collection by face-to-face interviews by trained interviewers of both sexes. The questionnaire was pretested among the elderly individuals in a similar rural community of neighbor Upazila (sub-district) and finalized the following necessary corrections and modifications based on the findings. All interviewers were medical graduates and had a Master of Public Health (MPH) degree. Interviews took place at the houses of cases and controls. The questionnaires consisted of a common set of questions, and information was obtained from cases and controls. These included baseline and personal characteristics, and malnutrition-related variables. We took the informed consent from each elderly individuals to interviews at their residences.
Measures
Baseline characteristics:
Information concerning baseline characteristics included age, gender, religion, marital status, monthly income, level of education, employment status, and family type. Although the country has four major religions, only two (i.e., Islam and Hinduism) were found in the present sample. Family income was used as an indicator of social class: less than 10,000 BDT (lower class), 10,000 to 20,000 BDT (middle class), and more than 20,000 BDT (upper class) based on the recommendations of Mamun MA et al., [24]. The marital status of being single included those who were divorced, separated, and widowed elderly individuals.
Personal characteristics:
Information concerning personal characteristics included psychosocial, physical health, lifestyle, and dietary factors. Relevant variables included peer group support history of suffering from chronic illnesses, active in daily life, having a poor diet, and tobacco use. For assessing peer group support, participants were asked if they received any type of support from friends and others. Regarding tobacco use, participants were asked about including smoking and smokeless tobacco use. For active in daily life, participants were asked if they performed daily activities on their own such as shopping, household chores, washing clothes, etc. For having a poor diet, participants were asked if they had consumed the foods regularly, timely, adequately, and nutritious for health.
Geriatric Depression
Depression in the elderly was assessed with GDS-15, the most well-established scale for assessing GD, which has been validated and widely used in both community and clinical settings [25]. The GDS assessed depressive symptoms experienced in the preceding week. The scale comprises 15 questions requiring a binary (‘yes/no’) response. Out of the 15 items, 10 indicated the presence of depression when answered positively, while the other remaining (Items 1, 5, 7, 11, and 13) indicated depression when answered negatively. The scale has a total score of 15 (ranged 0 to 15), where depression level is classified into normal (scores 0–4) [26] and a score of ≥5 indicates probable depression [27]. The internal consistency of the scale in the present study was very good (Cronbach’s alpha = 0.85).
Malnutrition:
The Bangla version of the Mini-Nutritional Assessment-Short Form (MNA-SF) was used for the collection of data on malnutrition. The revised MNA-SF of the MNA was developed and validated in Bangla especially for use among elderly individuals [≥60 years] [28]. It comprised of six questions (relating to appetite, weight loss, mobility, recent illness/stress, dementia/depression, and BMI), and was scored from 0 to 14. A score of 0-7 was used as the cutoff score for malnutrition [29]. The internal consistency of this scale in the present study was very good (Cronbach’s alpha= 0.89).
Statistical analysis
The data were analyzed using SPSS STATISTICS (Version 25.0, IBM Statistical Product and Service Solutions, Armonk, NY, USA) software. Descriptive statistics included frequency, percentage, means, and standard deviation. To assess the association between depression status and variables related to malnutrition, baseline, and personal characteristics of the elderly individuals, inferential statistics (Chi-square test and regression analysis) were carried out. All the variables statistically significant in bivariate analysis were entered into the binary logistic regression models [30]. The results of the logistic regression are reported as adjusted odds ratios with 95% confidence intervals (CI), and a p-value less than 0.05 was accepted as the significant level for this study.