Screening Algorithms to Reduce Burden of Pediatric HIV Testing: A Systematic Review and Meta-analysis.

BACKGROUND

The accuracy of symptom screening to identify children eligible for further HIV testing in generalized epidemics has been examined in several studies. We performed a systematic review and meta-analysis of these studies.

METHODS

We screened 5 databases and abstracts from 4 HIV/AIDS conferences. Studies were included if they were performed in clinical settings, included children of 0-15 years old, and used a signs/symptoms screen to determine eligibility for HIV testing. The primary outcomes were sensitivity and specificity of the screening tools. A meta-analysis was performed to evaluate the utility of a screening tool in the outpatient setting.

RESULTS

Our search returned 5529 database results and approximately 6700 conference abstracts, of which 36 articles were reviewed and 7 met criteria for inclusion. All were prospective or cross-sectional studies that developed and/or validated a screening tool to identify children at higher risk for being HIV infected. Sensitivity of the screening tools ranged from 71% to 96%, whereas specificity ranged from 25% to 99%. Meta-analysis of studies evaluating outpatient screening tools revealed a sensitivity of 81.4%, with a specificity of 69.4% for detecting HIV infection.

CONCLUSIONS

Few studies have evaluated the use of screening tools for HIV diagnosis in children. Screening tools that exist showed only moderate sensitivity and specificity and missed a substantial number of HIV-infected children in high-prevalence areas. In outpatient settings, the use of a screening tool may help reduce the number of HIV tests needed to identify an HIV-infected child, but at the cost of missed diagnoses. Further studies are needed to determine whether this represents a resource-saving mechanism.