The differential effect of ambient temperature on age-specific and sex-specific mortality in the 300 largest cities of Russia, 2000-19: a first national time-series study

Abstract

Background: Despite a substantial body of evidence on the association between ambient temperature and mortality worldwide, there has not yet been a comprehensive country-wide assessment of the health effects of temperature in Russia. Moreover, there is no consensus on the effect of non-optimal temperatures on age-specific and sex-specific mortality. Our study aimed to provide the first analysis of temperature-related mortality in a large assembly of cities located in different geographical and socioeconomic zones of Russia.

Methods: We analysed 19 044 538 non-accidental deaths in the 300 most populated cities in Russia between 2000 and 2019. A two-stage analysis strategy was used. First, a quasi-Poisson time-series model with distributed lag non-linear model was fitted to estimate city-specific associations. Second, these associations were pooled with multivariate multilevel meta-regression, from which we also calculated temperature-attributable mortality.

Findings: Relative risks were generally higher for cold than for heat, except for cities in southern European Russia. Cold had a similar effect in both sexes, with a varying age gradient across cities. Although the effect of heat was generally stronger in women than in men, with the relative risk increasing steadily with age in both sexes, men younger than 60 years had a significantly higher risk of dying from heat than women of the same age. With a total of 106 007 (95% empirical CI [eCI]: 88 942-121 318) temperature-attributable deaths, there was a higher mortality attributable fraction for cold (10·74%, 95% eCI 8·80-11·99) than for heat (0·67%, 0·42-0·88).

Interpretation: Russia has a high temperature-related mortality burden, with large differences in risk between cities and subpopulations. This information should be taken into account when planning public health interventions.

Funding: European Research Council, National Health and Medical Research Council, and Australian Research Council.

Figure 1

Pooled cumulative exposure–response associations between daily temperature and non-accidental mortality for the 300 largest Russian cities in 2000–19, stratified by sex (A) and age group (B) for the total population, by age group for females (C) and males (D), and by regions for the total population (E) Dashed vertical lines denote the 1st and 99th percentiles of the temperature distributions. Solid vertical lines denote minimum mortality temperature for the subgroups. Shaded areas denote 95% CIs of relative risks.

Figure 2

Geographical variation in minimum mortality temperature (A) and relative risks of death associated with cold (at the 1st percentile; (B) and heat (at the 99th percentile; (C) in 2000–19 Based on best linear unbiased predictions of the cumulative exposure–response association for the 300 largest Russian cities (non-accidental deaths, all ages, both sexes). Here and below, the borders of Russia are shown according to the 2019 delineations as defined by the Russian Government.

Figure 3

Association between mean temperature and cold-related relative risk (A) and heat-related relative risk (B) in 2000–19, across cities, both sexes, and all ages