Background:Diagnosis for HIV in infants is hard to determine, particularly in limited- resource areas. A delay in the diagnosis of HIV-infected infants will lead to high morbidity and mortality. The purpose of this project is to construct a model of an HIV-positive infant and develop a useful and practical scoring system to estimate the likelihood of mother-to-child transmission that can be applied in the field.

Methods:A cross-sectional study on 100 subjects through medical records of infants born to HIV-infected mothers was conducted at four hospitals and one community health center. Several models of risk prediction scores of HIV-infected infants were then made. Furthermore, the performed validation was performed on 20 subjects of infants born to mothers with HIV in three hospitals by comparing the scoring system and the result of the PCR RNA examination performed at the age of 6 weeks old.

Results:The risk of HIV-infected infants was higher in mothers who did not receive ARV through PMTCT programs (OR 33.6; 95% CI 4.0 to 282.2), pulmonary TB infection (OR 5.1; IK95% 1.6 to 16.0) and vaginal delivery (OR 9.2; IK95 2.2 to 38.0%). Two models can predict the occurrence of infected HIV infants effectively. Model 1 consists of maternal age, maternal ARVs, lung TB infection, gestational age, mode of delivery, and sex of the infants with sensitivity and specificity of 78.9% and 70.8% (AUC=0.817 [95% CI 0.709 to 0.926]) and likelihood ratio score of 4. Model 2 consists of ARVs to the mother, pulmonary TB infection, and mode of delivery with sensitivity and specificity of 73.7% and 86.1%; AUC value of 0.812 (95% CI 0.687 to 0.938) and likelihood ratio of 5. External Validation gave similar results to the Model 2 scoring system with PCR RNA.

Conclusion:The prediction score of HIV-infected infants in Model 2 can be used in newborns of HIV-positive mothers as an effective and practical risk screening tool for HIV-infected infants before the gold standard examination by PCR.