BACKGROUND: Approximately 2.4% of adults in New York City (NYC) are infected with chronic hepatitis C virus (HCV). The NYC Department of Health and Mental Hygiene (DOHMH) developed algorithms that use the pattern of reported HCV RNA test results to estimate HCV treatment and cure rates among NYC residents; both positive and negative RNA results are reportable. Previous algorithm validation used data from DOHMH direct-service programs, which involve intensive care coordination services and frequent patient follow-up. To validate the algorithms in the general population, where the frequency of provider visits and RNA testing may be less ideal, treatment and cure status was determined through chart review of HCV-infected patients in care across NYC.
METHODS: We randomly selected 250 HCV-infected patients from 15 healthcare facilities with high HCV treatment rates. Facilities were grouped as large hospitals (≥500 beds), small hospitals (<500 beds), and private practice/clinics; five facilities of each type were selected. Individuals were considered patients of a given facility if they had an HCV test at that facility during January–August 2015 and their most recent RNA test as of February 29, 2016 came from the same facility. Twenty-five patients were selected per large hospital, 15 per small hospital, and 10 per private practice. We compared treatment and cure status from chart review to algorithm results, and used testing patterns and outcomes observed in the chart, such as accounting for previous treatment failures, to inform further algorithm refinement.
RESULTS: Chart reviews were completed for 235/250 (94.0%) selected participants. Evidence of treatment in 2014 or later was found for 160 patients (68.1%), with 126 (78.8%) of those completing treatment. Ninety-six patients had evidence of cure, as documented in the chart (25.0%) and/or from a negative RNA test >12 weeks after treatment completion. Using the newly revised definitions, the algorithm identified treatment status with a sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of 93.1%, 82.7%, 92.0% and 84.9%, respectively. The revised algorithm identified cure with a sensitivity, specificity, PPV, and NPV of 93.8%, 88.6%, 88.2%, and 93.7%, respectively.
CONCLUSIONS: Chart review indicated that the algorithms accurately estimate treatment and cure using surveillance data and that accuracy can be improved using documented treatment outcomes. Practical applications of the algorithms include developing HCV care cascades and creating clinic/provider dashboards. Estimating treatment and cure will allow for enhanced monitoring, resource allocation, and program planning by DOHMH to address the ongoing HCV epidemic.