For the present study primary data was collected in Varanasi district of Uttar Pradesh state, India, from December 2017 to May 2018. The data used in the study is cross-sectional, and a multistage simple random sampling procedure was adopted. In the first stage, three blocks were selected out of eight blocks. Three primary sampling units were randomly selected at the second stage from each block, keeping urban and rural composition in view. Further, the third stage adopted a proportional allocation procedure to obtain the desired number of households from every primary sampling unit (PSU). We prepared a list of households with at least one respondent age 50 and above through house listing. Using a systematic simple random sampling method, we selected the targeted number of households, and from every selected household, one older adult age 50 and above was selected for the survey. Respondents with severe cognitive impairment or physical impairment were excluded from the study.
For the current study, the Varanasi district of Uttar Pradesh state was selected, with demographic and socioeconomic indicators similar to the state. Varanasi is located in the south-eastern part of Uttar Pradesh, with a population of 3.6 million. The older adult population aged 50 years and above constituted 14.6% which is relatively higher than the state (13.9%) in the 2011 census . Varanasi district has 1.84% of the total population of Uttar Pradesh, with a considerable proportion of the urban population (43%), which is again higher than the state average of 22%.
Sample size determination
In the absence of morbidity prevalence data for the study area, we took the estimated morbidity prevalence rate of Uttar Pradesh from the WHO-SAGE survey as a proxy for the Varanasi district. The state-level prevalence of multimorbidity was 16.8 % for the older population age 50 and above. Allowing 20% non-response rate, 95 % confidence level, 5 % margin of error with and 1.75% design effect, the estimated sample size was 451. The final sample size of the study was 500.
Quality of Life
In the present study, the QoL (overall wellbeing) is the outcome variable. The WHOQOL-BREF questionnaire has shown a good, reliable and valid cross-cultural measure for assessing QoL . The WHO-BREF consists of four significant domains, namely- physical, psychological, social relation and environment. The four domain consists of 24 questions. Two additional questions were asked related to overall health and QoL.
For the present analysis, we used two sets of QoL measures. The first set, including domain-specific QoL, is based on 24 questions and a second index based on 26 items. The WHOQOL-BREF is a combination of both positive and negative items. The positive items include happiness, content, and energy, whereas the negative items include sadness, sexual difficulty, and pain . The scoring method of domain-specific QoL is presented in the supplementary table at the end of the paper. However, the detailed methodology of WHOQOL-BREF is given elsewhere . The second set outcome variable, which was the overall quality of life, was created from all 26-items. Using the additive method, a raw score was generated, and this raw QoL score rescales on 0-100, where higher values present a higher QoL score.
Three sets of independent variables were used to assess their effects on QoL, namely demographi, health, socioeconomic factors and lifestyle factors.
- Demographic and health factors included are: a) Number of disease (no disease, one disease & two or more diseases), b) Self-rated health (good SRH & poor SRH), c) Age (50-59, 60-69, & 70+), d) Sex (male & female), and e) marital status (currently married & widowed).
- The socioeconomic factors included are: a) years of schooling (illiterate, 1-5 years, 6-9 years and 10 and above), b) religion (Hindu & Muslim), c) caste (Scheduled Caste (SC)/Scheduled Tribes (ST), Other Backward Class (OBC), & others), d) residence (rural & urban), e) wealth index (poor, middle, & rich), and f) economic independence (totally dependent, partially dependent & independent).
- The lifestyle and health factors included are:
a) physical activity based on the WHO recommendation of physical activity which included either 75 minutes of vigorous activity or 150 minutes of moderate physical activity in a week as sufficient physical activity and those failed kept as doing insufficient physical activity 
b) sedentary behavior marked those having more than 4 hours of a day on sitting in leisure time activity (yes & no).
The study was approved by the Student Research Ethics Committee (SREC) of International Institute for Population Sciences (IIPS), Mumbai (Sr.No. 15/1843, Date:13/10/2017). All respondents provided informed consent and were informed that they could withdraw from the study at any stage.
Descriptive statistics, mean proportions with standard deviations and Cronbach alpha were used to present the sample characteristics of the study population. Inferential statistics like t-test and variance analysis (ANOVA) were used to test differences in the mean QoL score of multimorbidity and sociodemographic measures. Multivariate linear regression analysis was performed to study the association of chronic multimorbidity and sociodemographic measures with QoL among older adults. All the results were presented with the beta coefficient. A p-value <0.05 was considered as the level of significance. All analysis was performed with STATA 14.2 software .