Ethics
The Institutional Review Committee of Kathmandu University School of Medical Sciences (IRC-KUSMS), Dhulikhel Hospital, Kavre, Nepal approved the study protocol.
All invited participants were informed about the nature, purpose and procedures of study. They were also informed that they were free to withdraw their consent and discontinue the research interviews at any time. In addition, they had the opportunity to ask any questions they would like regarding the study. Before the interviews began, written consent was given by those participants who could read and write, while fingerprints were collected from those who were unable to do that.
Study design and sampling
Data collection was made in January and February 2019. The design was a cross-sectional, population-based, door-to-door survey based on field interviews conducted in the Kavre district, which falls under Province no 3, Nepal. It comprises seven rural and six urban municipalities. According to the Central Bureau of Statistics (CBS) 2011[15], the total population was 381,937 (males 182,936 [47.9%] and females 199,001 [52.10%]. In this population, 36,912 were elderly aged 60 years or above (9.6% [males 9.7% and females 9.6%]).
Inclusion and exclusion criteria
The inclusion criteria for age was 60 years and above. The participants should live in the Kavre district and be able to speak Nepali. Moreover, elderly persons who had lost the ability to speak or had major hearing loss, who were disoriented were excluded, i.e., they were unable to give correct answers to at least two of the three following questions (i) time of day [morning, day and evening], (ii) location [outside or inside the house], and (iii) weekday of the data collection [Sunday to Saturday], and those who were residing in old-age homes.
Sample size and sampling procedures
We estimated a sample size of 460 persons by using the formula Z2pq/d2 assuming a prevalence of depression (p) 60.0%, a precision (d) 5.0% of p with 95% confidence interval (CI), and adding a 20% non-response rate.
We used a multistage systematic random sampling procedure to recruit eligible participants. In the first stage, four municipalities, two out of six urban and two out of seven rural municipalities of the Kavre district were selected by using a lottery method. We allocated equal proportions of participants (115 elderly) for each municipality to ensure a balanced representation from rural and urban areas. In the second stage, we selected at least 50% of the wards (administrative units) from each selected municipality also by using the lottery method.
The field interviewers sketched the ward map. They listed and numbered households with elderly inhabitants. This was achieved by help from the Female Community Health Volunteers (FCHVs); they are local women working voluntarily to enrich the outreach health care services at the community level. From the list, we selected households using a systematic random sampling technique. In the third stage, we selected one elderly person from each household. If there were two or more elderly in the household, one of them was selected randomly using the lottery method.
Participants
Of the potential 460 selected elderly individuals, 13 declined to participate, six had impaired hearing, and two were unavailable. These 21 elderly (10 male, 11 female; age 60–85 years [13 urban and 8 rural]) were regarded as non-participants. Finally, 439 persons participated in the study. Thus, the proportion of participation was 95.4%.
A comparison of representativity has been made in Table 1 between the age and gender distribution of our selected samples of the elderly and the figures from the CBS [15]. The selected sample proved to be quite representative of the elderly population of the Kavre district, despite a rather modest overrepresentation of females and persons of age 70 years and above.
Study instruments
Face-to-face interviews were done by trained nurses (field interviewers) using a structured questionnaire. The questionnaire was developed in English and translated into Nepali and tested in the field before initiating the data collection (Additional file 1). The final structured interview consisted of six parts (i) personal and demographic characteristics (age, gender, habitation, educational and marital status); (ii) life style related questions (alcohol consumption); (iii) family support (time provided by family, financial support, perceived respect from family, and verbal and/or physical abuse by family); (iv) questions that map the elderly person’s physical condition with regard to chronic physical health problems; the subjective report of any of the following diseases: diabetes, chronic obstructive pulmonary diseases [COPD], heart disease, cancer, paralyses and/or mobility problems; (v) questions from the Geriatric Depression Scale short form (GDS-15) [24]. A detailed description of the scale is given below. The last part (vi) consisted of biometric measurements such as height, weight, waist circumference and blood pressure (BP).
Geriatric Depression Scale short form (GDS-15) - Nepali version
We used the Geriatric Depression Scale short form (GDS-15) to scan the potential presence of depressive symptoms [24]. The instrument had already been used in Nepal [22, 23] and in other Asian countries [25-27] including India [28] and Pakistan [29].
The original English version GDS-15 [24] was translated into Nepali. For the validation, the Nepali version was administered to the total of 106 participants from the Kavre district (mean age 68.1) by trained nurses. The participants were later blindly interviewed by the local consultant psychiatrists for possible geriatric depression according to the criteria of the International Classification of Diseases-10 (ICD-10).
In the validation study, 5/6 was found to be the optimal cut-off point also in Nepal. The internal consistency was found to be 0.79, and the sensitivity (Se) was 86.3% and the specificity (Sp) was 74.5%. Further details of the validation study will be published elsewhere (Risal et al.).
The GDS-15 consists of 15 items; they are focusing on the psychological symptoms that the person felt during the past week. Each item is rated in a yes/no format. Among them, 10 items (2, 3,4,6,8.9,10,12,14 & 15) indicate the presence of depression when answered “yes” (positive), while the remaining 5 items (1, 5, 7, 11 & 13) indicated depression when answered “no” (negative). The potential total score ranged from 0 to 15 [24].
Assessment of geriatric depression
A score of 5 or less was considered to be within the normal range and classified as a “no case” of geriatric depression, while 6 or more endorsements were considered to indicate a caseness of geriatric depression according to the validation study.
Statistical analysis
We estimated the crude prevalence of geriatric depression with a 95% confidence interval (CI), and we adjusted the prevalence for gender and age according to the population distribution of the elderly aged ≥ 60 years of the Kavre district, as found in the data from the CBS [15].
The depression status of the GDS-15 was used as the dependent variable and the responses were dichotomized into yes (case of geriatric depression) or no (no case of geriatric depression).
Moreover, we dichotomized the socio-demographic variables like age (<75 or ≥75 years), gender (male or female), habitation (urban or rural), educational status (literate or illiterate), and marital status (married or unmarried/widowhood/separate); life style factors including alcohol consumption (yes or no); family support in terms of the time provided by the family (yes or no), source of financial support (her/himself or family), perceived respect from family (yes or no), and verbal and/or physical abuse (no or yes); and any physical condition that included limited mobility (‘able to go out of home’ or ‘not able to go out of home’); and finally, any chronic physical health problem (yes or no).
Bivariate and multivariate logistic regression analyses with odds ratios [ORs] and adjusted ORs [AORs] respectively were used, each with 95% CIs to investigate the associations of geriatric depression with the above mentioned variables.
The p-values <0.05 were considered statistically significant.
The statistical analyses were carried out using the Statistical Package for Social Science software (IBM SPSS Statistics 21, Chicago, USA).