Background/aim Considering the difficulty to separate the effects from individual pollutants present in mixtures, the cumulative risk index (CRI) estimates the combined effect from several pollutants together. We evaluated the association between air pollution exposure and daily elderly mortality using CRI from multipollutant models.

Methods This study was a daily time series of non-accidental and cause-specific mortality among the elderly living in São Paulo, Brazil, between 2000 and 2011. Effects of NO2, particulate matter smaller than 10 µm (PM10), carbon monoxide (CO) and ozone (O3) were estimated in Poisson generalised additive models. The single lag effect (lags 0 and 1) and the cumulative effect (lag 0 to 10) were evaluated in one, two-, three- and four-pollutant models and the CRI was estimated for each model. Air pollution effect estimates are presented as percentage increase or decrease in the number of deaths, and their 95% confidence interval (CI), for the interquartile range of air pollutants.

Results An association between NO2, PM10, CO and O3 exposures and deaths was found in one- and multipollutant models. For circulatory deaths, the CRI of NO2, for lag 1 (1.13%; CI: 0.69 to 1.57) and the cumulative lag 0–10, was close to the CRI of the four-pollutant model (1.49% for lag 1 (CI: 0.91 to 2.06)). For respiratory deaths, the CRI from the two-pollutant model with CO and O3 (12.34% for lag 0–10 (CI: 7.12 to 17.81)) represents the largest fraction of the CRI from the four-pollutant model (12.23% for lag 0–10 (CI: −2.65 to 29.38)). For non-accidental deaths, the pattern differs per lag. For lag 1 the CRI of all two-, three- and four-pollutant (1.49%; CI: 0.91 to 2.06) models was similar.

Conclusion The results suggest that air pollution mixtures have an effect on elderly mortality. The CRI documented that single pollutants did not fully capture the risk of the mixture.