Infants born to HIV-infected women should receive HIV testing

Infants born to HIV-infected women should receive HIV testing to allow early diagnosis and treatment. Recommendations for resource-limited settings stress laboratory-based virologic assays. While effective, these tests are logistically complex and expensive. This study explored the cost-effectiveness of incorporating initial screening with rapid HIV tests into the conventional testing algorithm to screen-out HIV-uninfected infants, thereby reducing the need for costly virologic testing. Data on HIV prevalence, rapid HIV tests sensitivity and specificity, and costs were collected from 820 HIV-exposed children (1.5-18 months) attending 2 postnatal screening programs in Uganda during July 2005 to December 2006. Cost-effectiveness models compared the conventional testing algorithm DNA polymerase chain reaction (DNA-PCR with Roche Amplicor v1.5) with a modified algorithm (initial rapid HIV tests to screen-out HIV-uninfected infants before DNA-PCR). The model estimated that the conventional algorithm would identify 94.3% (91.8%-94.7%) of HIV-infected infants, compared with 87.8% (79.4%-90.5%) for a modified algorithm using rapid HIV tests (HIV 1/2 Determine) and excluding the need for DNA-PCR for HIV antibody-negative infants. Costs per infant were $23.47 ($23.32-$23.76) for the conventional algorithm and between $22.75 ($21.89-$23.31) and $7.58 ($6.41-$10.75) for the modified algorithm, depending on infant age and symptoms. Compared with the conventional algorithm, costs per HIV-infected infant identified using the modified algorithm were higher in 1.5-to 3-month-old infants, but significantly lower in 3-month-old and older infants. Models replicating the whole infant testing program showed the modified algorithm would have marginally lower sensitivity, but would reduce total program costs by 27% to 40%, producing an incremental cost-effectiveness ratio of $1489 ($686-$6781) for the conventional versus modified algorithms. Screening infants with rapid HIV tests before DNA-PCR is cost-effective in infants 3 months old or older. Incorporating rapid HIV tests into early infant testing programs could improve cost-effectiveness and reduce program costs.

For abstract access click here: 1

Editors’ note: In the absence of antiretroviral prophylaxis, between 70 and 85% of infants born to mothers with HIV infection do not acquire HIV during pregnancy, labour, and delivery. However, all infants born to such mothers will have a positive HIV antibody test because their mother’s HIV antibodies cross the placenta before birth and persist in the baby’s blood up to 18 months of age. Because parents want to know whether their baby is infected and because antiretroviral treatment should be started in infants as soon as HIV infection is detected, it is not acceptable to wait until 18 months of age for a negative HIV antibody test to confirm that the baby has no antibodies of its own. Current policy in Uganda recommends testing of all HIV-exposed infants at 10 weeks of age, at their second immunization visit, by DNA-PCR which detects the virus itself, not antibodies. Although cost-effectiveness studies such as this one are context-specific with results varying by HIV prevalence, costs of supplies and personnel, equipment, etc., they are nonetheless of interest for programme planners in similar settings. The idea of using a rapid test to screen out infants that are likely not infected and then use the DNA-PCR test only on those that are rapid test-positive is partly motivated by the cost and complexity of DNA-PCR testing and partly by an appreciation of the visit burden imposed on parents by repeated testing visits should the first DNA-PCR be negative and the baby continues to breastfeed. Given that the algorithm studied here was cost-effective only in infants 3 months or older, there is not much to suggest that Uganda should change its current 10-week DNA-PCR testing policy. Using a rapid test to screen above that age in breastfeeding infants in order to decide whether a DNA-PCR should be done could be considered.