Prediction of subnational-level vaccination coverage estimates using routine surveillance data and survey data

Area of Science:

• Epidemiology
• Public Health
• Vaccinology

Background:

• Measles vaccination significantly reduced disease burden, yet coverage disparities persist.
• Accurate subnational vaccination coverage data is vital for targeted interventions.
• Existing methods for estimating coverage have limitations in accuracy, timeliness, and spatial resolution.

Purpose of the Study:

• To explore the use of routinely collected case-based surveillance data for predicting subnational measles vaccination coverage.
• To develop and validate a model using surveillance data to estimate vaccination coverage.

Main Methods:

• Utilized aggregated measles case data from 18 WHO African region countries.
• Derived three surveillance indicators: mean age of suspected cases, proportion vaccinated, and proportion IgM-negative.
• Developed a beta regression model using Demographic and Health Surveys (DHS) as the gold standard, comparing predictions with withheld DHS estimates.

Main Results:

• Surveillance indicators showed stronger correlation with DHS coverage than administrative estimates.
• Out-of-sample predictions achieved a high correlation (rho = 0.74) with DHS-based coverage.
• The model demonstrated the effectiveness of surveillance data in predicting vaccination coverage.

Conclusions:

• Routinely collected measles surveillance data can effectively predict subnational vaccination coverage.
• This approach provides yearly estimates superior to administrative data and more accessible than surveys.
• Enables timely identification of low-coverage areas to facilitate targeted public health interventions.