# Development of a prediction rule for incomplete vaccination among children in Indonesia

## TL;DR

This study creates a prediction tool to identify Indonesian children under 2 years old at high risk of incomplete vaccination, aiming to improve vaccination coverage.

## Contribution

The study introduces a novel prediction rule based on multiple factors to identify children at high risk of incomplete vaccination in Indonesia.

## Key findings

- Factors like young maternal age, lack of mobile phones, and low socio-economic status are linked to incomplete vaccination.
- The model achieved moderate discrimination with an AUC of 0.67, showing potential for targeted interventions.
- Using a cut-off score of >20 points, the model identifies 60% of high-risk parents with 64% specificity.

## Abstract

Childhood vaccination is a fundamental public health intervention, playing an essential role in improving health outcomes and preventing serious infections. Despite proven benefits of vaccination programs, its coverage in Indonesia remains inadequate over the years. Therefore, this study aims to develop a prediction rule using intrapersonal, interpersonal, organizational, community, and policy-related factors to distinguish between Indonesian children < 2 years at high and low risk of incomplete vaccination.

The prediction rule was developed using cross-sectional data from the 2017 Indonesia Demographic Health Survey. Data on vaccination status was obtained from a vaccination card, which was filled out by health care providers during vaccination. Multivariable logistic regression was applied to develop a prognostic score based on the regression coefficients of associated parental intrapersonal, interpersonal, organizational, community, and policy-related factors. Discrimination of the model was assessed with Receiver Operating Characteristic (ROC) curve.

The sample population in this study comprised 3,790 respondents, and 2,414 (63·7%) were incompletely vaccinated. Several factors such as a mother at young age, absence of a mobile telephone, limited antenatal care attendance, absence of postnatal checks within two months after birth, had not received tetanus vaccination during pregnancy, and low socio-economic status were independently associated with incomplete vaccination. The area under curve (AUC) of the model was 0·67, which showed moderate discrimination, but was acceptable. Using a cut-off score of > 20 points, only half of the parents with a high probability of incompletely vaccinated children are selected with a sensitivity of 60% and specificity of 64%, and only 41% of parents with incompletely vaccinated children are missed.

This novel, easy-to-use prediction rule could be a useful tool to complement current strategies and further encourage tailored vaccine uptake interventions, particularly to parents with a high chance of incompletely vaccinated children in Indonesia.

The online version contains supplementary material available at 10.1186/s12889-025-23109-0.