A group of researchers recently published a systematic review of studies of hepatitis C virus (HCV) treatment that reached a shocking conclusion: Curing the virus does not prolong life or reduce rates of HCV-related illness. Experts in the field have slammed the paper by the Cochrane Collaboration, a network of experts that reviews medical research, as compromised by critical limitations.

“Our review of this Cochrane publication suggests significant flaws in this analysis, yielding a misleading and a harmful conclusion,” says Anna Lok, MD, president of the American Association for the Study of Liver Diseases (AASLD), and William Powderly, MD, president of the Infectious Diseases Society of America (IDSA). The paper’s findings, they say, ignore “mounting published literature supporting the clear clinical benefit” of curing hepatitis C.

The Cochrane authors analyzed the effect of direct-acting antiviral (DAA) treatment for HCV on rates of death, illness and serious adverse health events. The review was based on 138 randomized clinical trials that included 25,232 participants.

The investigators’ meta-analysis found no difference in the rates of sickness and death based on whether individuals received hepatitis C treatment or a placebo. However, just 11 studies including about 3,000 individuals actually provided data on sickness and death.

The overall review had data on only 16 deaths. The study authors did not provide information on how much follow-up time was associated with those study participants’ deaths.

Another major limitation of the review is that the studies in the analysis provided only minimal follow-up, with an average of just 34 weeks—too short to detect any effect on mortality except in cases involving severe liver disease.

Meanwhile, a recent systematic review of 31 studies of interferon and ribavirin treatment for HCV had a median 5.4 years of follow-up. The authors of that paper found that curing hepatitis C reduced the risk of death by 50 to 80 percent over a five-year period. That review included data on 2,210 deaths, lending much more statistical heft to its findings.