J Clin Hypertens (Greenwich). 2026 Sep;28(9):e70358. doi: 10.1111/jch.70358.
ABSTRACT
This study was designed to identify the changing patterns of the potential covariates that influence hypertension stages (SHT) over time. The study used two cross-sectional datasets (BDHS 2017/18 and 2022), which had 12,189 and 13,843 respondents aged ≥18 years. The χ2analysis examined the association between the selected covariates and SHT as well as their influence on SHT using multinomial logistic regression (MLR) models (p < 0.05). Moreover, a SHAP-based approach was used to quantify the contribution of each covariate to the predictive performance for SHT classification. By adjusting for age and gender, the MLR-based model showed that seven significant covariates were found for SHT in both BDHS 2017/18 and 2022. Both MLR and SHAP-based approaches identified education, BMI, wealth index, diabetes status, and division as the principal determinants of SHT for BDHS 2022 dataset. Education and BMI were a significant and consistent predictors of SHT in both survey years. Respondents with secondary education and obese had 1.4 times (AOR: 1.4, 95% CI: 1.1-1.8) and 5.3 times (AOR: 5.3, 95% CI: 3.2-8.7) higher odds of having controlled hypertension (CHT) and severe hypertension in 2022. The respondents from middle and rich households also had higher odds of mild hypertension in 2017/18, whereas only rich respondents were found to have CHT and mild hypertension in 2022. Marital status showed a changing association with SHT over time, with divorced respondents showing a significantly higher risk of developing severe hypertension in 2022. These findings highlight the need for geographically targeted and socioeconomically tailored public health interventions in Bangladesh.
PMID:42703909 | DOI:10.1111/jch.70358