Citywide re-evaluation of the Bangkok air quality health index: classification performance, nonlinear same-day respiratory associations, and implications for precautionary health communication

Information from the abstract

Air quality health indices (AQHIs) translate multipollutant exposure into graded public-health messages, but their performance as morbidity classifiers or nonlinear lag-informed indicators remains uncertain in tropical megacities. We evaluated the Bangkok AQHI using city-wide outpatient morbidity data from 2021 to 2024. We conducted a retrospective city-wide time series analysis of outpatient department (OPD) data from 23 hospitals across 26 residential postcodes in Bangkok, Thailand, from 2021 to 2024. Fixed AQHI coefficients from the original model were applied without recalibration. The outcomes included broad operational, clinically selected restricted, disease specific, and sensitivity groups. Receiver operating characteristic analyses were used to assess the discrimination of high-morbidity days according to the outcome-specific 90th percentile. Negative binomial models estimated global log-linear associations per interquartile range increase in AQHI. Distributed lag nonlinear models (DLNMs) evaluate nonlinear same-day and cumulative associations over lags 0–7, contrasting AQHI 7 with the citywide median. Sensitivity analyses included post-COVID-19 restriction, PM 2.5 comparator analyses, alternative exposure aggregation, postcode fixed effects, spline sensitivity, stratified ROC analyses, and leave-one-year-out validation. Across the 1,461 study days, ROC analyses revealed limited discrimination for all broad and restricted outcomes. The fixed-direction AUCs ranged from 0.353 to 0.539 for the AQHI and from 0.421 to 0.562 for the PM 2.5 . Negative binomial models revealed no consistent global log-linear associations between the AQHI and morbidity. DLNMs identified small same-day associations at AQHI 7, most coherently for clinically selected respiratory morbidity (RR 1.023, 95% CI 1.004–1.042), which remained significant after family-specific Bonferroni and false discovery rate correction. Nominal same-day signals were observed for asthma, COPD, and ischaemic stroke, with supportive evidence from J44-coded COPD. The cumulative associations over lags 0–7 were close to null. Post-COVID-19 analyses supported the temporal consistency of restricted respiratory, asthma, COPD, and ischaemic-stroke signals. The Bangkok AQHI was best positioned as a graded precautionary communication indicator rather than a universal morbidity classifier or global linear predictor of daily citywide OPD morbidity. The strongest citywide evidence supports targeted respiratory health messaging when the AQHI reaches the high category., whereas cardiovascular-vulnerable groups warrant continued precautionary inclusion pending subgroup-specific evaluation.

Why this record is monitored

This record has an Impact Signal of 73/100 based on recency, source, collaboration, and bibliographic signals. It prioritizes monitoring and is not a judgment of research quality.

Related topics: Air Quality and Health Impacts · Air Quality Monitoring and Forecasting · COVID-19 epidemiological studies