Estimates of case-fatality ratios of measles in low-income and middle-income countries: a systematic review and modelling analysis

Abstract

Background: In the 21st century, increases in immunisation coverage and decreases in under-5 mortality have substantially reduced the global burden of measles mortality. However, the assessment of measles mortality burden is highly dependent on estimates of case-fatality ratios for measles, which can vary according to geography, health systems infrastructure, prevalence of underlying risk factors, and measles endemicity. With imprecise case-fatality ratios, there is continued uncertainty about the burden of measles mortality and the effect of measles vaccination. In this study, we aimed to update the estimations of case-fatality ratios for measles, to develop a prediction model to estimate case-fatality ratios across heterogeneous groupings, and to project future case-fatality ratios for measles up to 2030.

Methods: We did a review of the literature to identify studies examining measles cases and deaths in low-income and middle-income countries in all age groups from 1980 to 2016. We extracted data on case-fatality ratios for measles overall and by age, where possible. We developed and examined several types of generalised linear models and determined the best-fit model according to the Akaike information criterion. We then selected a best-fit model to estimate measles case-fatality ratios from 1990 to 2015 and projected future case-fatality ratios for measles up to 2030.

Findings: We selected 124 peer-reviewed journal articles published between Jan 1, 1980, and Dec 31, 2016, for inclusion in the final review-85 community-based studies and 39 hospital-based studies. We selected a log-linear prediction model, resulting in a mean case-fatality ratio of 2·2% (95% CI 0·7-4·5) in 1990-2015. In community-based settings, the mean case-fatality ratio was 1·5% (0·5-3·1) compared with 2·9% (0·9-6·0) in hospital-based settings. The mean projected case-fatality ratio in 2016-2030 was 1·3% (0·4-3·7).

Interpretation: Case-fatality ratios for measles have seen substantial declines since the 1990s. Our study provides an updated estimation of case-fatality ratios that could help to refine assessment of the effect on mortality of measles control and elimination programmes.

Funding: Bill & Melinda Gates Foundation.

Figure 1

Review of the literature published in 1980–2016 of case-fatality ratios for measles in low-income and middle-income countries

Figure 2

Case-fatality ratio for measles, by year of study, extracted from studies published in 1980–2016 Line indicates the trend line of the case-fatality ratio for measles by year, according to a simple linear regression of case-fatality ratio with time (year). 158 observations from community-based studies; 68 observations from hospital-based studies.

Figure 3

Mean predicted case-fatality ratios for measles from community-based and hospital-based studies, 1990–2015 Predicted case-fatality ratios (dots) are averaged arithmetically across all 136 low-income and middle-income countries for all ages in each year to provide a mean predicted case-fatality ratio in that year, with bounds according to the 95% CI. Community-based and hospital-based studies were weighted by the number of measles cases from each included study.

Figure 4

Boxplots of predicted case-fatality ratios for measles from 1990–2015 by World Bank income level (A) and under-5 mortality (B) for 136 low-income and middle-income countries and all ages Under-5 mortality defined as the estimated number of deaths per 1000 livebirths. LIC=low-income country. LMIC=lower-middle-income country. UMIC=upper-middle-income country.

Figure 5

Mean projected case-fatality ratios for measles from community-based and hospital-based studies, 2016–2030 Projected case-fatality ratios (dots) are averaged arithmetically across all 136 low-income and middle-income countries for all ages in each year to provide a mean projected case-fatality ratio in that year, with bounds according to the 95% CI. Community-based and hospital-based studies were weighted by the number of measles cases from each included study.