# Using Measles Outbreaks to Identify Under-Resourced Health Systems in Low- and Middle-Income Countries: A Predictive Model

## TL;DR

This study explores whether measles outbreaks can reveal weaknesses in health systems in low- and middle-income countries.

## Contribution

The paper evaluates if measles outbreaks can serve as an indicator of under-resourced health systems using a predictive model.

## Key findings

- No association was found between health system indicators and large measles outbreaks.
- The predictive model had low accuracy with a Brier score of 0.21.
- Missing data did not impact the model's results.

## Abstract

Background/Objectives: Measles is a vaccine-preventable disease with a high level of transmissibility. Outbreaks of measles continue globally, with gaps in healthcare and immunisation resulting in pockets of susceptible individuals. Measles outbreaks have been proposed as a “canary in the coal mine” of under-resourced health systems, uncovering broader system weaknesses. We aim to understand whether under-resourced health systems are associated with increased odds of large measles outbreaks in low- and middle-income countries (LMICs). Methods: We used an ecological study design to identify measles outbreaks that occurred in LMICs between 2010 and 2020. Health systems were represented using a set of health system indicators for the corresponding outbreak country, guided by the World Health Organization’s building blocks of health systems framework. These indicators were: the proportion of births delivered in a health facility, the number of nurses and midwives per 10,000 population, and domestic general government health expenditure per capita in USD. We analysed the associations using a predictive model and assessed the accuracy of this model. Results: The analysis included 78 outbreaks. We found an absence of any association between the included health system indicators and large measles outbreaks. When testing predictive accuracy, the model obtained a Brier score of 0.21, which indicates that the model is not informative in predicting large measles outbreaks. We found that missing data did not affect the results of the model. Conclusions: Large measles outbreaks were not able to be used to identify under-resourced health systems in LMICs. However, further research is required to understand whether this association may exist when taking other factors, including smaller outbreaks, into account.