Data Source
Data from the National Family Health Survey [NFHS-5] was retrieved, which is a house assessment that gathers demographic and health surveys [DHS] for all 29 states and 7 union territories of India. It was conducted by the Ministry of Health and Family Welfare [MoHFW], managed by the International Institute for Population Sciences [IIPS], Mumbai, and implemented by a group of survey organizations and Population Research Centres. The sample design for NFHS-5 was uniform design implemented by using a two-stage cluster sampling approach, where primary sampling units [villages in rural areas and census enumeration blocks in urban areas] were chosen with a probability proportionate to the size of the population, and following a mapping and household listing, households were chosen by simple random sampling
Study Variables
· Dependant variables:
The primary outcome of interest was the waist-to-hip ratio. For the first time, NFHS-5 included waist and hip circumference measurements which were recorded by using Gulick tapes for eligible women and men for measurements of abdominal obesity. WHR is categorized as high risk in women at≥0.85 cm and men at ≥0.90 cm.
· Independent variables:
We considered independent or predictor variables as per the following broad domains.
a) Socio-demographic variables:
Caste, age, sex, and marital status, region[urban/rural], religion, type of family, wealth index, education and place of residence, region of country.
b) Anthropometric indicators:
High waist circumference [cut-off values 90 cm for men and 80 cm for women], Hip circumference, high waist-to-hip ratio [cut-off value 0.90 cm for men and 0.85 cm for women], levels of BMI [underweight, normal, overweight calculated by using formula Weight in kg/Height in m2]
c) Co-morbidity and behavioural habits:
We have included smoking/tobacco consumption, and alcohol consumption to understand the behavioural habits among the study population. Additionally, morbidity patterns such as hypertension were also considered. In NFHS-5, three blood pressure [systolic and diastolic pressure] Subjects' observations were collected at least 5 min between each BP measurement with a standardized OMRON BP monitor. We have considered the average of the three measurements of blood pressure to decide whether a participant was hypertensive. If a single measurement was absent for an individual in the dataset, then the average of the remaining two measurements was used. We used the remaining measurements if two measurements were missing. The person was considered hypertensive if the average of all three systolic pressure was between 140-200 mmHg or diastolic pressure was between 90-140 mmHg.
Sample Selection:
Data from 2,022,037 individuals [household member recode file] was available from the DHS datasets. The information regarding particular variables was downloaded and then checked to identify and address errors or inconsistencies, such as missing values.
The missing information was managed by two approaches: If data of a particular subject was absent for multiple variables, it was deleted and not considered in the analysis and If data was lacking for one or two variables, it was imputed using appropriate statistical methods [by substituting the mean value]. The identification of outliers was achieved through visual inspection and was corrected by computing the mean and removing extreme values.
Some variables were transformed to ensure compatibility with statistical analysis. This included the categorization of the age variable into 3 sub-groups, and creating composite indices such as hypertension status by using three measurement values of BP, BMI using weight and height variables and WHR using waist and hip circumference.
A final inspection of the cleaned dataset was conducted to validate and ensure that each problem discovered during the cleaning process was resolved. The cleaned data collection was ready for statistical analysis with a comprehensive data dictionary documenting all variables, coding schemes, and transformations applied.
Data Analysis
Initially, for the visualization of national and state-level data on waist-hip ratio, the maps were drawn to study the patterns of WHR across India's states and union territories using the software Tableau and the prevalence of waist-hip circumference studied trends. The states representing darker shades in the map indicate a higher prevalence of WHR in the region while the lighter shades indicate low prevalence. In addition, the prevalence of WHR was further connected with the predicting factors using individual-level data. All the variables were coded and categorized using the raw data in Excel and the data sheet was exported to statistical software SPSS version 25.0 to run the statistical analysis. The descriptive statistics were used to examine the socio-demographic features of the research's subject using frequency and percentage. Pearson’s chi-square test of significance is used to test the possible relationship between men's and women's independent and dependent variables separately. Logistic regression was utilized to determine the factors substantially connected to higher abdominal obesity. The binary logistic regression test was performed by the ‘enter’ method, considering waist-hip ratio as the dependent variable and entering the following variables as the independent variables in Step 1: age in years, gender, type of residence, wealth index, Household Structure, smoking, alcohol, BMI Status, and BP. Bivariate analysis and logistic regression were reported as a chi-square test and adjusted odds ratio [OR], the statistical analysis was undertaken at a 95% confidence level with significance at a p-value of <0.05.
Ethical Consideration:
Since there was no human engagement with the study and it made use of secondary information from a national survey, an ethical review was not considered essential.