Commentary on “Symptom-Based Screening is not the Solution to Improve Pediatric HIV Testing”

In the current issue of the Pediatric Infectious Disease Journal, Sara Clemens et al report the results of a systematic review and metaanalysis of studies examining the use of symptom screening to reduce the number of children needing to receive an HIV test in clinical settings. This study is timely as HIV testing programs in high-burden countries are being pushed to both dramatically reduce the volume while simultaneously improving the yield of testing. To accomplish these goals, several countries have looked to adopt some form of symptom screening, in which children are only administered an HIV test if they screen positive for one or more clinical symptoms. Unfortunately, the results reported in this exhaustive systematic review and metaanalysis suggest that symptom screening is not the solution that policymakers and national testing programs are seeking.

The first and most dramatic result from this study is how limited the evidence is supporting the utility of symptom-based screening. After a search that included over 12,000 articles, only 7 met the criteria for inclusion in this review. Five of the studies were from outpatient, while 2 were from inpatient settings. Of these 7 studies, only 2 report findings from the post-B+ era, in which a dramatic decrease in the number of new pediatric HIV infections has been observed.¹ Even if the results reported in these studies had been robust, policymakers would need to take pause before endorsing an approach with such a slim body of evidence. Even more problematically, though, the results reported by the studies included in this metaanalysis were wholly inadequate to support endorsement of symptom-based screening as national policy. The 2 inpatient studies demonstrated such low specificity (25% and 40%) that the majority of patients screened would need a confirmatory HIV test, meaning that the screen was not an effective tool for reducing the numbers of HIV tests administered, one of the main goals for having a screen in the first place. Metaanalysis of 4 outpatient studies that used a similar screening tool showed a sensitivity of 81.4% and a specificity of 69.4%. As Clemens et al detail in their metaanalysis, adoption of such tools in high-burden countries as national policy would result in an unacceptable number of missed diagnoses.

What are policymakers to do then? Is there any alternative approach to symptom-based screening that might reduce the number of HIV tests done on children, while still maintaining adequate sensitivity and specificity? Work currently being done in Malawi suggests that there may be. But before discussing this approach, it should first be recognized that in some clinical settings and pediatric populations, universal testing is still required. All patients being evaluated in inpatient and specialized clinical services such as malnutrition and TB in high-prevalence settings need an HIV test, to both identify new cases but also, most importantly, to provide timely diagnosis to guide clinical management.² Adolescents are another population where universal and repeat testing would be the favored approach. Adolescents and young people, particularly girls and young women, represent some of the most at risk populations for incident HIV infection. In 2018 alone, an estimated 190,000 adolescents between the ages of 10 and 19 newly acquired HIV infection, representing 11% of all new HIV diagnoses.¹ Given that only 19% of adolescent girls and 14% of adolescent boys ages 15–19 years in Eastern and Southern Africa have been tested for HIV within the last 12 months,¹ all opportunities to test adolescents accessing health services should be capitalized on.

Outside of populations requiring universal HIV testing, what can be done to reduce the amount of low-yield testing taking place in pediatric outpatient and well-child clinical settings? An increasingly large body of evidence suggests that screening for maternal HIV status, as opposed to clinical symptoms, may be a more effective approach for both reducing the volume while simultaneously increasing the yield of pediatric HIV testing.³ Data from the Sustainable East Africa Research in Community Health trial in Kenya and Uganda, for example, revealed that of the over 75,000 children tested in the study, HIV testing yield was highest in those children whose mothers tested HIV positive (6.0%–7.1%) or were of unknown status (0.4%–2.0%), varying between region, whereas children with mothers who tested HIV negative had dramatically lower yields of 0.04%–0.22%.⁴ Screening utilizing HIV exposure as opposed to clinical risk factors has the added advantage of identifying asymptomatic patients as well as mothers who acquired HIV late in pregnancy or during breast-feeding.

Building on this evidence, a pilot project spearheaded by the Ministry of Health in Malawi Provider Initiated Testing and Counseling Subgroup with technical assistance from the Clinton Health Access Initiative endeavored to develop a screening approach based on HIV exposure status of the child. This work is based on the simple premise that healthy children of mothers who test negative for HIV do not also need to be tested for HIV. In this pilot, the mothers of all children 12 years old and under presenting to outpatient settings were screened for their HIV status. Over 95% of the close to 20,000 children screened in this pilot were accompanied by their mothers. The roughly 35% of mothers, who did not have a current and documented HIV negative status, were retested with over 80% accepting same-day testing. Only children of mothers who were new positive diagnoses, known positives but had for any reason missed infant/child testing, or with deceased/unavailable mothers were tested. Using this approach, only 4% of infants and children screened in outpatient and well-child clinic required an HIV test, with a 4.4% yield in those who were tested.^5,6 While these results are still preliminary and require validation and peer review, logic dictates that this approach may have superior sensitivity and specificity to symptom-based screening.

Although the pilot being conducted in Malawi is the first attempt we are aware of to explicitly utilize HIV exposure status to determine eligibility for HIV testing in outpatient and well-child clinics, confirming maternal status before pediatric testing in outpatient and community settings has already been captured as policy in national testing guidelines in most high-burden countries for over a decade.³ Clinton Health Access Initiative is conducting further qualitative work in Malawi with focus group sessions with patients and providers that aims to identify challenges and areas for improvement, but the experience so far suggests that most of these barriers to implementation can be surmounted and that now is the time for screening of mothers to move out of the guidelines into actual implementation in the field. Ironically, it may turn out that the approach to both reduce the amount and improve the yield of pediatric HIV testing is the one we have had all along.