By Anne Friedland, MD, Infectious Disease Program Fellow and Susanna Naggie, MD, Associate Professor of Medicine and Principal Investigator for the HERO-HCQ trial. (Both with the Duke School of Medicine.)
Our historical understanding of medicine and treatment is backed by carefully conducted research to prove what does – and doesn’t – work. Over the years, the scientific community has aligned on core responsibilities for those conducting research to ensure we consistently adhere to the highest standards in the design, conduct, and reporting of research. A key feature of research accountability is to ensure the results of research studies are accurately shared.
One way we ensure accountability is through a peer-review process, meaning any time a researcher has results of a study, those results are reviewed by other, objective researchers to be sure all agree the study was properly conducted and that the results being reported are supported by the study data. In the era of COVID 19, this process of sharing and validating research findings has been quickened – and in some cases short-changed.
A recent pre-print article (meaning it has not been through this peer-review process) reporting results from a study of veterans with COVID-19 hospitalized at VA medical centers across the country was released to the public. This article is worth highlighting because it is the first reporting on the early experience of COVID-19 in the VA population. It also highlights a broader issue. Early results may have much uncertainty or significant issues which later, larger studies will not. In general, the earliest studies out are those that are retrospective, observational, or small studies. These carry significant limitations and have to be viewed cautiously while the large well-designed, randomized clinical trials are underway. All agree that it is critical to understand whether hydroxychloroquine is effective in treating or preventing COVID-19. What seems to be murky is how to interpret the mix of data either showing potential benefit or potential risk of hydroxychloroquine on clinical disease.
In this VA study, researchers looked back at 368 veterans who had been hospitalized with COVID-19. The veterans were grouped according to whether they had been treated with:
- hydroxychloroquine alone
- hydroxychloroquine combined with azithromycin
- no hydroxychloroquine.
They looked at the rates of death and whether patients needed mechanical ventilation. Rates of death in the hydroxychloroquine, hydroxychloroquine + azithromycin and no hydroxychloroquine groups were 27.8%, 22.1% and 11.4%, respectively. Rates of ventilation in the hydroxychloroquine, hydroxychloroquine + azithromycin and no hydroxychloroquine groups were 13.3%, 6.9% and 14.1%, respectively. Use of hydroxychloroquine alone was associated with increased risk of mortality (but not hydroxychloroquine with azithromycin.) Hydroxychloroquine was not associated with increased risk of needing mechanical ventilation.
These results could be interpreted in one of two ways at least. One way is that the patients do worse with hydroxychloroquine. An alternative and likely way is that clinicians selected the sickest patients for treatment and so not surprisingly they fared worse. How to know which answer is right can only be addressed through prospective, well-controlled studies. Specifically, the veterans were observed retrospectively (after the fact) and thus were not randomly assigned to groups to receive treatment. Therefore, there are numerous potential factors that could have confounded (skewed) the results. For example, the veterans who received hydroxychloroquine (both arms) had more severe disease than the veterans who were not prescribed hydroxychloroquine including lower oxygen saturations, higher liver enzymes, higher creatinine and lower albumin. In addition, veterans who received hydroxychloroquine were more likely to have cardiovascular disease, which has been associated with COVID severity and outcome.
Therefore, it is not necessarily surprising that the patients treated with hydroxychloroquine had increased mortality. While the study attempted to control for these differences using statistical methods, the differences in these study populations highlight the difficulty of drawing conclusions from uncontrolled studies.
Our conclusion from this study is aligned with that summarized by the FDA communication: We need randomized clinical trials. Hydroxychloroquine should not be used for the treatment of COVID disease outside of clinical trials. But without clinical trials we will not know whether this drug can be used as part of our toolkit to treat patients with COVID. Furthermore, we need studies for inpatients, outpatients, and for prophylaxis, where we have no current public data but a clear need.