The influence of pre-existing health conditions on short-term mortality risks of temperature: Evidence from a prospective Chinese elderly cohort in Hong Kong

Both cold and hot temperatures are associated with adverse health outcomes. Less is known about the role of pre-existing medical conditions to confer individual's susceptibility to temperature extremes. We studied 66,820 subjects aged ≥65 who were enrolled and interviewed in all the 18 Elderly Health Centers of Department of Health, Hong Kong from 1998 to 2001, and followed up for 10–13 years. The distributed lag nonlinear model (DLNM) combined with a nested case-control study design was applied to estimate the nonlinear and delayed effects of cold or hot temperature on all natural mortality among subjects with different pre-existing diseases. The relative risk of all natural mortality associated with a decrease of temperature from 25th percentile (19.5°C) to 1st percentile (11.3°C) over 0–21 lag days for participants who reported to have an active disease at the baseline was 2.21 (95% confidence interval (CI): 1.19, 4.10) for diabetes mellitus (DM), 1.59 (1.12, 2.26) for circulatory system diseases (CSD), and 1.23 (0.53, 2.84) for chronic obstructive pulmonary disease (COPD), whereas 1.04 (0.59, 1.85) for non-disease group (NDG). Compared with NDG, elders with COPD had excess risk of mortality associated with thermal stress attributable to hot temperature, while elders with DM and CSD were vulnerable to both hot and cold temperatures. Elders with pre-existing health conditions were more vulnerable to excess mortality risk to hot and/or cold temperature. Preventative measures should target on elders with chronic health problems. •Temperature mortality effects are modified by pre-existing health conditions.•Elders with CSD and DM are vulnerable to both cold and hot temperature.•Elders with COPD are vulnerable to hot temperature.•DLNM could be combined with a nested case-control study design.