Additive interaction of family medical history of diabetes with hypertension on the diagnosis of diabetes among older adults in India: longitudinal ageing study in India

The data from the Longitudinal Ageing Study in India (LASI) Wave 1 (2017–2018) were used in this study. The survey collected data on the health, economic, and social factors, and consequences of India’s population ageing. The LASI is a full-scale, nationally representative survey that included 72,250 individuals aged 45 years and older and their spouses (irrespective of age) across all states and union territories (UTs) of India except Sikkim. The LASI uses a multistage stratified area probability cluster sampling to select the eventual units of observation. This study presents scientific evidence on chronic health conditions, biomarkers, symptom-based health conditions, and functional and mental health. The LASI survey was conducted with a three-stage sampling design in rural areas and a four-stage sampling design in urban areas. In each state/UT, in the first stage, Primary Sampling Units (PSUs) were selected, and in the second stage, villages in rural areas and wards in urban areas were selected in the selected PSUs. In the third stage, households were selected from each selected village; however, sampling in urban areas involved an additional stage, i.e., the random selection of one Census Enumeration Block (CEB) in each urban area. In the fourth stage, households were selected from each CEB. The main goal was to select a representative sample at each stage of sample selection. The detailed methodology and extensive information on the survey’s design and data collection are available in the report [29]. The present study is based on 65,562 respondents aged 45 years and above excluding those less than 45 years (n = 6,688). Additionally, after removing respondents with missing information on self-reported diabetes (n = 181) and those with incomplete information in any of the selected variables and biometric measurements (n = 6,769) (including family history: 394; hypertension: 2; physical inactivity: 53; ADL: 121; IADL: 180; biometric measurements of body mass index: 6,328 and any selected variables), the total sample size for the analysis was 58,612 respondents. (Fig. 1 presents the inclusion and exclusion criteria for the study sample.

The main outcome variable was self-reported diabetes. In the study, respondents were asked, “Has any health professional ever diagnosed you with diabetes?”. The responses were coded as no and yes. In LASI, respondents were asked additional questions to those who reported being diagnosed with a disease by a medical professional, including the diagnosing physician, the date of diagnosis, and whether they are currently receiving treatment. Participants were asked, “In order to treat or control your diabetes or high blood sugar, are you currently taking medications?”. Interaction effects were estimated in two different study sample, (1) self-reported diabetes and (2) Individuals with diabetes, who were taking treatment.

The main explanatory variable was self-reported hypertension, and family medical history of diabetes. In the study, respondents were asked, “Has any health professional ever diagnosed you with high blood pressure or hypertension?”. The responses were coded as no and yes. In the LASI, to understand the genetic risk factors for diabetes, information was collected about the respondent’s family medical history; the family medical history of the father, mother, brother, and sister were selected for the analysis.

In LASI, respondents were asked additional questions concerning medication to those who reported being diagnosed with a disease by a medical professional including whether they are currently receiving treatment. Participants were asked, “In order to control your blood pressure or hypertension, are you currently taking any medication?”.

Age was coded as 45–54 years, 55–64 years, 65–74 years, and 75 + years. Gender was coded as male and female. Education was recoded as no education, primary, secondary, and higher. Marital status was coded as currently married, widowed, and others, not in a union. Working status was coded as never worked, currently working, and currently not working. Alcohol use was coded as ‘no’ and ‘yes’; smoking and chewing tobacco were coded as ‘never’, ‘former, and ‘current’ The BMI was computed by dividing the weight (in kilograms) by the square of the height (in meters). BMI was coded according to the criteria of the World Health Organisation’s classification; as underweight (< 18.5 kg/m2), normal weight (18.5–24.9 kg/m2), overweight (25.0–29.9kg/m2), and obesity (≥ 30.0 kg/m2), for the analysis, overweight and obesity were combined [30]. The monthly per capita expenditure quintile (MPCE) or consumption quintile was categorized into five quintiles, poorest, poor, middle, rich, and richest. Religion was categorized as Hindu, Muslim, Christian, and Others. The social group (caste) was categorized as Scheduled Castes (SC), Scheduled Tribes (ST), Other Backward Classes (OBC), and others. The ‘other’ category in caste is identified as non SC/ST and OBC. The place of residence was coded as urban and rural. The regions were categorized as North, Central, East, Northeast, West, and South. In the study, respondents were asked, “Has any health professional ever diagnosed you with high cholesterol”. The responses were coded as no and yes.

To assess difficulty in activities of daily living (ADL), respondents were asked questions on the following six activities: difficulty in dressing, walking across a room, bathing, eating, getting in or out of bed, and using the toilet. Individuals who reported difficulty with any activity for more than three months were coded as “yes” and otherwise “no”.

To assess difficulty in instrumental activities of daily living (IADL), respondents were asked if they had difficulty in performing in any of the following seven activities: preparing hot meals, shopping for groceries, making telephone calls, taking medication, doing household work, managing finances, and getting around or finding an address in an unfamiliar location (Alpha value: 0.88). Those who reported trouble with any of these activities for more than three months were labelled “having difficulty.” Otherwise, they were categorized as having “no difficulty.” ADL and IADL are considered as measures of functional health and a prolonged difficulty in any of the items refers to individuals’ dependence on others and/or instrumental devices.

To assess the level of physical activities, participants were asked about the type and amount of physical activity integrated into daily life. For vigorous activities, participants were asked “How do you often take part in sports or vigorous activities such as running or jogging, swimming, going to the health centre/gym, cycling, digging with a spade or shovel, heavy lifting, chopping, farm work, fast bicycling, and cycling with loads?”. For moderate activities, participants were asked, “How do you often take part in sports or activities that are moderately energetic such as cleaning house, washing clothes by hand, fetching water, or wood, drawing water from a well, gardening, bicycling at a regular pace, walking at a moderate pace, dancing, floor or stretching exercises?”. The available responses for evaluating moderate and vigorous activities were as follows: every day, more than once a week, once a week, one to three times per month, and hardly ever or never. For both moderate and vigorous activities, participants were also asked “On the days you did the activity, how much time did you usually spend doing any activity?”.

Weekly durations of both moderate and vigorous physical activities were computed: Moderate physical activity was defined as those who engaged in a minimum of 150 min of moderate-intensity physical activity in a week, while vigorous physical activity encompassed those who engaged in a minimum of 75 min of vigorous-intensity physical activities in a week. Respondents were subsequently categorized into two groups based on their engagement in moderate and vigorous activities: “Physically active,” denoting those who engaged more than once a week, and “Physically inactive,” characterizing individuals who engaged once a week or less often. Subsequently, a binary variable of physical activity variable was created as “Physically active,” comprising those engaged in either moderate or vigorous physical activities, and otherwise “Physically inactive,” [31, 32].

Multivariable logistic regression was used to calculate the unadjusted and adjusted estimates, aiming to evaluate the joint effect of family medical history of diabetes with hypertension on diabetes. Further, an additive model was employed to assess the additive interaction effect of family medical history of diabetes with hypertension on diabetes by estimating three distinct measures of additive interaction: the relative excess risk due to interaction (RERI), attribution proportion due to interaction (AP), and synergy index (S). Interaction on an additive scale means that the combined effect of two exposures is not simply the sum of their individual effects. Instead, the two exposures interact with each other to produce a combined effect that is greater (or smaller) than the sum of their individual effects [33, 34].

The interaction measures on the additive scale are defined as RERI (excess risk of the outcome due to the interaction between two factors) = OR11 - OR10 - OR01 + 1; AP (proportion of the combined effect that is attributable to the interaction between two factors) = RERI / OR11; S (ratio between combined effect and individual effects) = (OR11–1) / (OR10–1) + (OR01–1). RERI = 0 indicates no interaction, RERI > 0 suggests positive interaction, RERI < 0 suggests negative interaction, AP = 0 suggests no interaction, AP > 0 suggests positive interaction, AP < 0 suggests negative interaction, S = 1 suggests no interaction, S > 1 suggests positive interaction, S < 1 indicates negative interaction [33‐35].

Bivariate analysis was performed to investigate the prevalence of diabetes and proportion of individuals who were using medication in relation to selected variables. A chi-square test and bivariate analysis were also employed to investigate the prevalence of diabetes concerning the combined effect of a family medical history of diabetes with hypertension. Moreover, multivariable binary logistic regression analysis [36] was employed to establish the association between diabetes, and main explanatory variables including hypertension and family history of diabetes (comprising overall family history, parental, sibling, and specific histories related to the father, mother brother and sister).

In the current study, the multivariable logistic regression and additive interaction models were adjusted for potential confounding factors, including age, sex, education, working status, marital status, residence, MPCE, religion, caste, region, physical inactivity, smoking, chewing tobacco, alcohol consumption, body mass index (BMI), ADL, IADL, and high cholesterol. The survey weights were applied during the analysis to account for sample clustering and present population estimates. All the analyses were conducted using Stata version 14.1 [37].