A time-series dataset of multi-pollutant air pollution, meteorological variables, and cerebrovascular mortality in Thailand from 2014 to 2024.

This dataset integrates multi-pollutant air quality measurements, meteorological variables, and cerebrovascular disease (CeVD) mortality across the Bangkok Metropolitan Region (BMR) from 2014 to 2024. Air pollution data were obtained from 19 urban monitoring stations in Bangkok and 4 suburban stations in surrounding provinces operated by the Thailand Pollution Control Department. Meteorological conditions, including ambient temperature and relative humidity, were sourced from the same network to ensure spatial and temporal consistency in exposure assessment.A structured pre-processing protocol was applied to prepare analysis-ready exposure data. Hourly pollutant records were screened for missing entries, placeholder symbols, and values below detection limits, which were recoded as NA or assigned the minimum detectable level. Missing hourly observations were imputed using a sequential moving-average approach: first with a 168-hour (7-day) window, followed by a 360-hour (15-day) smoothing step. Daily pollutant concentrations were calculated only when at least 75% of hourly values were available for that station-day. For spatial exposure assignment, 5-km buffer zones were defined around urban stations and 20-km zones around suburban stations; in overlapping areas, urban buffers were prioritized to reflect higher population density. Daily exposures were then averaged across all active stations within each buffer, producing representative urban and suburban pollutant concentrations.CeVD mortality data were obtained from the Ministry of Public Health, Thailand, covering January 2014 to December 2024. Deaths were classified using ICD-10 codes I60–I69 and included information on date of death, sex, age, and area of residence. All records were fully anonymized, containing no personal identifiers. Ethical approval was granted by the Graduate Program Committee, Faculty of Science, Chulalongkorn University, and the use of anonymized secondary data was exempt from institutional review board review.The air pollution and mortality datasets were subsequently merged by date and residential area. After merging, the Air Quality Health Index (AQHI) was computed using daily concentrations of PM2.5, NO2, and CO, following an approach adapted for Thailand. The resulting AQHI values were appended to each observation and used for downstream analyses.Each row in the final dataset represents a single day of observation and includes detailed variables describing mortality, exposures, and environmental conditions. Key fields include the calendar date (day, month, and year), ICD-10–coded CeVD mortality counts, demographic characteristics (sex and age group), and regional identifiers. Environmental exposures are represented by daily concentrations of multiple pollutants (PM2.5, NO2, SO2, CO, O3), meteorological variables (temperature and relative humidity), and the calculated AQHI for that day. Together, these variables provide a comprehensive framework for evaluating short-term associations between air quality and cerebrovascular mortality in both urban and suburban settings.The final merged datasets consist of two files: an urban dataset with 23,700 rows and 82 variables, and a suburban dataset with 10,092 rows and 96 variables. These files together form a complete, analysis-ready time-series database suitable for advanced epidemiological modeling, including AQHI evaluation, lag-specific exposure-response estimation, and subgroup analyses.