Population-level risk factors related to measles case fatality: a conceptual framework based on expert consultation and literature review

Abstract

Introduction: A better understanding of population-level factors related to measles case fatality is needed to estimate measles mortality burden and impact of interventions such as vaccination. This study aimed to develop a conceptual framework of mechanisms associated with measles case fatality ratios (CFRs) and assess the scope of evidence available for related indicators in the published literature. Methods: Following expert consultation, we constructed a conceptual framework of mechanisms associated with measles CFR and listed population-level indicators potentially associated with each mechanism. We also conducted a systematic literature review to determine and classify the scope of evidence for the listed indicators. Results: Five mechanisms were included in the conceptual framework: health system access and care-seeking behaviors, health system quality, measles control and epidemiology, nutritional status, and risk of secondary infection. Thirty-seven measurable population-level indicators were chosen as proxies for the mechanisms. We found studies showing a significant association with CFR for 15 indicators and non-significant associations only for 5 indicators. Inadequate data were available to evaluate the remaining 17 indicators. A randomized controlled trial provided evidence for a relationship between vitamin A treatment and measles CFR. Average household size, educational attainment, first-dose coverage of measles-containing vaccine, human immunodeficiency virus prevalence, second-dose coverage of measles-containing vaccine, stunting prevalence, surrounding conflict, travel time to major city or settlement, travel time to nearest health care facility, under-five mortality rate, underweight prevalence, vitamin A deficiency prevalence, and wasting prevalence had quantitative observational level evidence. Level of care available had qualitative evidence of a relationship with measles CFR. Conclusion: To reduce uncertainty in measles CFR estimation, population-level factors representative of underlying mechanisms associated with CFR could be used. Our study provides additional insights and a summary of the available evidence of these underlying mechanisms and indicators that could inform future estimations of measles CFR.

Estimating national-level measles case fatality ratios: an updated systematic review and modelling study

“…Previous work identified measles incidence and age to be critical covariates when assessing measles CFR. 3,5,7 We additionally selected covariates based on a systematic review and expert consultation 7 that identified five possible underlying mechanisms contributing to systematic increases or decreases in measles CFR (health system access and care seeking behaviours, health system quality, nutritional status, measles control and epidemiology, and risk of secondary infection) and related population-level indicators with evidence of an association with measles CFR (average household size, educational attainment, first-dose coverage of measles-containing vaccine (MCV1), human immunodeficiency virus (HIV) prevalence, level of health care availability, second-dose coverage of measles-containing vaccine (MCV2), stunting prevalence, surrounding conflict, travel time to nearest health care facility, under-five mortality rate, underweight prevalence, vitamin A deficiency prevalence, vitamin A treatment prevalence, and wasting prevalence).…”

“…Then, in a second stage model, we used the same general model formula previously described, maintaining the spline knot locations identified in the first stage model, and fit our outcome of logit CFR to all data following age-splitting (Supplementary Information Section 3). To ensure the correct direction of association between each covariate and CFR, defined as the direction described previously 7 , we placed priors on each regression coefficient. We generated 1000 samples of the regression coefficients from their fitted joint posterior distribution and predicted country- and age-specific measles CFR in LMICs from 1990 to 2019.…”

“…Covariates were grouped into five mechanisms as previously described. 7 To examine the correlation of covariates, we then calculated the pairwise correlation of each of the covariates in each mechanistic group. If there was a correlation > 0.8 for any pairwise comparison, covariates were removed sequentially based on the average highest collinearity between all other covariates in the mechanistic group.…”

“…6 While a major advancement, these previously published estimates do not include CFR data from the most recent years, nor an underlying conceptual model between CFR and associated covariates. 7…”

“…6 While a major advancement, these previously published estimates do not include CFR data from the most recent years, nor an underlying conceptual model between CFR and associated covariates. 7 Additionally, CFR estimates stratified by broad age categories may obscure important variation within these age groups, particularly for younger children. Both previous systematic review studies have shown that measles CFR is higher in persons younger than five years of age compared with children five years of age or older.…”