Findings from perinatal HIV prevention study in Uganda are valid

A Ugandan drug trial's findings that the AIDS medication nevirapine is effective and safe in preventing HIV transmission from mother to unborn child during birth were well-supported, according to a new, independent analysis by the Institute of Medicine of the National Academies.

The IOM's analysis of the design and methodology of the 1997 drug study in Uganda, called HIVNET 012, determined that policy-makers and other scientists can rely on the resulting data and conclusions, despite some flaws in record keeping and procedural issues.

"The data from the HIVNET 012 study, which showed that nevirapine effectively prevents many infants from contracting HIV from their infected mothers, are sound and reliable," said James Ware, chair of the committee that wrote the report, and professor of biostatistics, Harvard School of Public Health, Boston. "None of the shortcomings that we discovered upon reviewing the data and conducting our own original analysis of source documents indicates a need to retract or discount the study's findings. Our confidence in the trial's data and findings is based on several factors, including evidence that the study's design was both scientifically sound and ethically implemented, that participants adhered well to the treatment regimens, and that a high percentage of participants remained in the study so that the effectiveness and safety of the drug could be thoroughly assessed."

Previous evaluations of HIVNET 012 left lingering uncertainties about the trial's results, suggesting the need for a definitive, objective review. The IOM focused on the scientific validity of the study's conclusions based on a close examination of how researchers from Johns Hopkins University and Uganda's Makerere University conducted the trial. This independent review was requested and funded by the National Institutes of Health, which also funded the original trial in Uganda.

The committee did not evaluate the trial's oversight by the National Institutes of Health. It also did not examine the impact of other recent studies of potential toxicity or resistance buildup associated with the use of nevirapine either in short- or long-term treatment of HIV-infected individuals.

Because of inconsistencies in and challenges to previous audits, the committee undertook its own assessment of the accuracy and completeness of the trial's reporting through a review of medical records and other primary source documents for a subset of 49 infants involved in the trial. In addition, the committee reviewed information provided by NIH, the original investigators, previous audits of the trial, and other information brought to its attention. The findings of the HIVNET 012 study previously underwent audits by Westat Corp. and by NIH's Division of AIDS (DAIDS).

The Hopkins and Makerere researchers' conclusion that nevirapine is effective is supported by data on rates of survival and HIV infection among newborns in the study, the committee determined, noting that the trial researchers accurately recorded that information in the database created for the study. No evidence was found that the trial researchers either failed to report or mistakenly reported the deaths of any of the infants.

Regarding the trial researchers' findings on nevirapine's safety, the committee's review of source documents for the subset of 49 infants found that deaths, hospitalizations, and serious adverse events observed during clinic visits also were recorded accurately in the trial database. In some instances, however, not all serious adverse events that occurred simultaneously were reported, and some less-serious adverse events were underreported.

However, there was no evidence of a difference in the level of underreporting of adverse events among patients receiving nevirapine versus those receiving zidovudine, a second AIDS drug that also was studied in HIVNET 012. The trial investigators' comparative findings on safety are valid, the committee said.

Questions in previous audits about whether any adverse events had been missed stemmed from the trial investigators' use of a narrow but acceptable interpretation of what counted as "serious." The Hopkins and Makerere researchers used hospitalization as the principal -- but not sole -- determinant to classify clinical events as "serious" to take into account the high prevalence of malaria, tuberculosis, and other concurrent health problems in Uganda. The IOM report finds that the investigators' use of a narrow interpretation was reasonable, but it means that other researchers may not be able to generalize the study's total rate of adverse events to all settings. Other settings -- such as countries with lower rates of endemic diseases -- may have different thresholds for hospitalization and interpretations of what counts as "serious."

Another concern about HIVNET 012 focused on whether cases of jaundice -- or hyperbilirubinemia -- among infants in the study were underreported. While the study investigators reported only one infant with abnormal levels of bilirubin, a subsequent safety report issued by DAIDS initially stated that there were 63 cases of elevated bilirubin. DAIDS later retracted the safety report as incorrect. The IOM committee, based on its own analysis, determined that the DAIDS safety report initially used an incorrect upper limit of the normal range for bilirubin levels in newborns. When the correct upper limit is applied, the trial data confirm the original HIVNET 012 investigators' findings and the subsequent DAIDS retraction.

Overall, the Hopkins and Makerere researchers conducted the trial ethically and in accordance with U.S. and international standards for research and management of patient care, the IOM report says. Although there were some problems with full documentation of compliance with all the requirements for the trial, the committee determined that the HIVNET 012 study was designed and implemented with approval from the boards that oversaw the trial's design and protocols; that the researchers enrolled women only after they gave free and informed consent; and that fathers were involved in the consent process when they were reasonably available.

Blood tests that detected the presence of nevirapine in mothers and infants and other data showed that trial participants received the right drug and there was a high level of adherence to the treatment regimens, the committee found. It also noted that trial investigators achieved high rates of retention and follow-up among participants.

The committee found no reason that medical journals should revise or retract articles that reported on the efficacy and safety of nevirapine for reducing mother-to-child transmission of HIV based on the HIVNET 012 trial.

The Institute of Medicine is a private, nonprofit institution that provides health policy advice under a congressional charter granted to the National Academy of Sciences. A committee roster follows.