Higher Yield from Large Trials of Universal vs. Risk-Based HCV Screening

Three strategies for achieving one-time screening for hepatitis C virus (HCV) show variable success, but all detected more cases of HCV than current practice. During a plenary session of the Liver Meeting held November 11, 2014 in Boston, MA, CDC Foundation Fellow Anthony Yartel, MPH, reviewed results from three large randomized controlled trials of the effectiveness of hepatitis C virus (HCV) testing for persons born from 1945-1965.

Recent CDC recommendations for one-time universal screening for individuals born in this time period stem from knowledge that the prevalence of HCV is four times higher among this group than in other adults, and 77% of all HCV cases (and 74% of HCV-associated deaths) occur within this age cohort. Further, universal screening is recommended because 46% of infected individuals do not report having any exposure risk, such as risky sexual behavior or intravenous drug use.

The CDC study, conducted from December 2012 — March 2014, consisted of three independent randomized controlled trials at separate large primary care centers. Each trial utilized a different recruitment method, and each had a different structure for the randomized controlled trial. The intervention was a systematic, one-time HCV test, compared with continuation of usual care, usually consisting of risk-based or symptom-based testing.

The first trial initiated the intervention by sending a series of mailings to 3,000 randomly sampled patients, inviting them to go to the nearest affiliated lab for testing. Mailings were sent at weeks 0, 1, 4, 8, and 12, with the window for testing within the intervention closing at 16 weeks. The control group was an additional 6,000 randomly selected patients who received no intervention; their HCV testing status was followed via electronic medical record (EMR) data collection. Patients in this group were well-balanced between arms, as would be expected by this randomization method

The second trial, a cluster randomized trial of a total of 10 clusters, had providers and medical assistants at five of these clusters receive an EMR Best Practice Alert pop-up notification to offer HCV screening. The remaining five control clusters did not have a best practice pop-up activated, so patients continued to receive usual care. Patients were not well-matched between arms in this study, a known shortfall of cluster randomization.

The third trial, a cluster randomized crossover trial, involved four clusters, with the study design having each cluster serve as its own control. For the first period, 2 clusters initiated waiting room solicitation for HCV testing; this was conducted by a trained and experienced recruiter. The remaining two clusters continued usual care without a waiting room recruiter. After the first study period was completed, a one-month washout with no change from usual practice was followed by a second study period in which each cluster switched to the other study arm.

Using an intention-to-treat analysis and adjusting for effects of correlated data, risk ratios of identification of HCV+ patients were calculated for each study, comparing birth cohort testing to each center’s usual care. Additionally, a meta-analysis pooling results from all sites was conducted; however, Yartel explained, data from the third site are being re-examined and so have not yet been included in the meta-analysis.

Trial #1 yielded a 26.9% testing rate for the intervention arm, with a risk ratio of 8.0 vs. usual care for detecting HCV, but with only 2.7 vs. 0.3 HCV+ individual identified per 1,000 eligible patients. The second trial resulted in a 31.0% testing rate and a risk ratio of 3.2 for detecting HCV compared with usual care. A similarly low incidence of overall cases were also seen at the second site (3.0/1,000 intervention and 1.1/1,000 control).

Analysis for the third site is not complete, but this site saw the highest success rate for testing, probably related to the personal interaction with a trained recruiter: 60.2% of the intervention group was tested. The rate of HCV+ individuals was 7.6 per 1,000 eligible patients.

Overall, said Yartel, the birth cohort testing strategy resulted in four times the likelihood of identifying HCV+ patients, according to a meta-analysis using just the first two sites. Vigorous discussion following his presentation, however, also brought out the fact that the overall numbers of HCV+ individuals among the intervention population was less than 1%, significantly less than the predictions from which the universal testing recommendations were drawn. Whether achieving higher screening numbers would significantly increase the number of cases detected is not yet known.