Heat effects of ambient apparent temperature on all-cause mortality in Cape Town, Durban and Johannesburg, South Africa: 2006-2010.

Due to climate change, an increase of 3-4°C in ambient temperature is projected along the South African coast and 6-7°C inland during the next 80years. The objective of this study was to investigate the association between daily ambient apparent temperature (Tapp) and daily all-cause non-accidental mortality (hereafter mortality) in Cape Town, Durban and Johannesburg during a 5-year study period (2006-2010). Susceptibility by sex and age groups (<15years, 15-64years and ≥65years) was also investigated. The associations were investigated with the time-stratified case-crossover epidemiological design. Models were controlled for PM, public holidays and influenza epidemics. City-specific Tapp thresholds were determined using quasi-Poisson generalised additive models. The pooled estimates by sex and age groups were determined in meta-analyses. The city-specific Tapp thresholds were 18.6°C, 24.8°C and 18.7°C, respectively for Cape Town, Durban and Johannesburg. A 3.3%, 2.6% and 2.8% increase in mortality per IQR increase in Tapp (lag0-1) was observed in Cape Town, Durban and Johannesburg, respectively above the city-specific thresholds. The elderly were more at risk in Cape Town and Johannesburg. No difference in risk was observed for males and females in the three cities. In the meta-analysis an overall significant increase of 0.9% in mortality per 1°C increase in Tapp (lag0-1) was observed for all age groups combined in the three cities. For the ≥65year group a significant increase of 2.1% in mortality was observed. In conclusion, the risks for all age groups combined and the elderly are similar to those reported in studies from developed and developing countries. The results can be used in present-day early warning systems and in risk assessments to estimate the impact of increased Tapp in the country due to climate change. Future research should investigate the association between Tapp and cause-specific mortality and also morbidity.