Changes in the short-term relationship between air pollution and mortality in New York City, 1990–2019

Few studies have examined how short-term associations between air pollution and mortality have evolved over recent decades when air quality has improved. Public health policy can benefit from timely information. We applied time-series models to estimate the mortality impacts of ambient nitrogen dioxide (NO2), ozone (warm season only), and fine particulate matter (PM2.5) in 5-year moving time windows between 1990 and 2019 (2000–2019 for PM2.5) in New York City (NYC). We modeled full-year, warm (May through September) and cold (October through March) season NO2 and PM2.5, adjusting for temperature, temporal trends, day-of-week, and holidays. We also estimated Total Risk Index (TRI) to characterize changes in the combined risk from exposure to two and three pollutants. All three pollutants showed the strongest association at one lag day. Despite major declines in PM2.5 and NO2 levels over the study period, risk estimates showed no apparent trend, remaining generally positive, but became less precise over time as concentration variability also declined. The estimated overall 1-day lag percent excess risk for PM2.5 was 0.49% (95% confidence interval: 0.12, 0.86) per 10 µg/m3 24-hr average, and for NO2, 0.90% (0.30, 1.50) per 30 ppb daily 1-hr maximum for full year models. Ozone, which slightly increased over the period, had a 1-day lag risk estimate of 1.43% (0.56, 2.30) per 30 ppb daily 8-hr maximum. The TRI followed a similar pattern to individual pollutants’ estimates. With no clear evidence of risk per unit increase changing over time, the reductions in PM2.5 and NO2 concentrations imply declines in excess deaths. Notably, ozone levels and health burden persist. NO2, which was most robustly associated with mortality and represents two major combustion sources—traffic and fossil fuel combustion in buildings—may be the most relevant indicator of energy transition progress in urban areas like NYC in the coming decade.