Screening for Hepatitis C Virus Infection in Adults: A Systematic Review for the U.S. Preventive Services Task Force
Download the PDF here
Ann Intern Med. 15 January 2013
Roger Chou, MD; Erika Barth Cottrell, PhD, MPP; Ngoc Wasson, MPH; Basmah Rahman, MPH; and Jeanne-Marie Guise, MD, MPH
"The CDC recently recommended that all persons born between 1945 and 1965 be screened for HCV infection, in addition to persons with risk factors for HCV infection (25). The CDC based its recommendation on the prevalence of patients with HCV infection in this birth cohort (accounting for about three quarters of patients with HCV infection in the United States), the high proportion of patients with undiagnosed HCV infection, projected disease burden after several decades of infection, and estimated benefits from antiviral treatments. Although cost-effectiveness analyses suggest that the birth cohort screening approach is highly cost-effective, no clinical data are yet available (13). The CDC's birth cohort approach was not evaluated in the studies included in our review on the yield of alternative screening strategies. Clinical studies that prospectively evaluate the accuracy, yield, and outcomes of alternative HCV screening strategies, including the birth cohort approach, are needed."
Background: Identification of hepatitis C virus (HCV)-infected persons through screening could lead to interventions that improve clinical outcomes.
Purpose: To review evidence about potential benefits and harms of HCV screening in asymptomatic adults without known liver enzyme abnormalities.
Data Sources: English-language publications identified from MEDLINE (1947 to May 2012), the Cochrane Library Database, clinical trial registries, and reference lists.
Study Selection: Randomized trials and cohort, case-control, and cross-sectional studies that assessed yield or clinical outcomes of screening; studies reporting harms from HCV screening; and large series reporting harms of diagnostic liver biopsies.
Data Extraction: Multiple investigators abstracted and checked study details and quality by using predefined criteria.
Data Synthesis: No study evaluated clinical outcomes associated with screening compared with no screening or of different risk- or prevalence-based strategies. Three cross-sectional studies in higher prevalence populations found that screening strategies that targeted multiple risk factors were associated with sensitivities greater than 90% and numbers needed to screen to identify 1 case of HCV infection of less than 20. Data on direct harms of screening were sparse. A large study of percutaneous liver biopsies (n = 2740) in HCV-infected patients with compensated cirrhosis reported no deaths and a 1.1% rate of serious adverse events (primarily bleeding and severe pain).
Limitations: Modeling studies were not examined. High or unreported proportions of potentially eligible patients in the observational studies were not included in calculations of screening yield because of unknown HCV status.
Conclusion: Although screening tests can accurately identify adults with chronic HCV infection, targeted screening strategies based on the presence of risk factors misses some patients with HCV infection. Well-designed prospective studies are needed to better understand the effects of different HCV screening strategies on diagnostic yield and clinical outcomes.
Primary Funding Source: Agency for Healthcare Research and Quality.
The prevalence of anti-hepatitis C virus (HCV) antibody in the United States is about 1.6% (1). Approximately 78% of affected patients have viremia, indicating chronic infection. About two thirds of patients with HCV infection were born between 1945 and 1964, with the highest prevalence (4.3%) in people 40 to 49 years of age in 1999-2002 (1). There were 16 000 new cases of HCV infection in 2009 (2).
In 2007, HCV infection was associated with an estimated 15 000 deaths in the United States (3). Liver disease related to HCV is the most common indication for liver transplantation among U.S. adults (4 - 5) and is a leading cause of hepatocellular carcinoma (6).
The virus is primarily acquired via percutaneous exposures to infected blood, such as injection drug use (7 - 13). Transfusions before 1992 and high-risk sexual behaviors are also associated with increased risk, although the efficiency of sexual transmission seems to be relatively low (7 - 8,14 - 15).
The natural course of HCV infection varies. Studies of community cohorts estimate cirrhosis in 7% of people after 20 years of infection, with rates about twice as high in clinical and referral cohorts (16 - 17). Studies with longer follow-up suggest that disease progression accelerates after 20 years (18).
Screening for HCV infection could identify persons at earlier stages of disease, before they develop serious or irreversible liver damage, and lead to treatments to improve clinical outcomes or reduce transmission risk. Up to three quarters of HCV-infected persons are unaware of their status (19).
In 2004, the U.S. Preventive Services Task Force (USPSTF) recommended against HCV screening in adults not at increased risk (D recommendation) and found insufficient evidence to recommend for or against screening in high-risk adults (I recommendation) (20). Although the USPSTF found that screening tests are accurate and that antiviral treatments improve viremia (21), the recommendations were based on the lower prevalence of HCV infection in persons without risk factors; the relatively low rate of long-term progression, potentially resulting in overtreatment; and lack of evidence that screening improves important health outcomes or reduces transmission risk. Other groups recommend screening in higher-risk patients (22 - 24). The Centers for Disease Control and Prevention (CDC) also recently recommended screening all persons born between 1945 and 1965 (25).
The purpose of this report is to review the evidence on HCV screening in asymptomatic adults without known liver enzyme abnormalities (26). This review focuses on research gaps identified in the 2004 USPSTF review (21) and will be used together with a separate review on antiviral treatments (27) by the USPTF to update its HCV screening recommendations.
We developed a review protocol and analytic framework that included the following key questions:
1. Does screening for HCV infection in nonpregnant adults without known abnormal liver enzymes reduce mortality and morbidity due to HCV infection, affect quality of life, or reduce incidence of HCV infection?
2. What is the effectiveness of different risk- or prevalence-based methods for screening for HCV infection on clinical outcomes?
3. What is the sensitivity and number needed to screen to identify 1 case of HCV infection of different risk- or prevalence-based methods for screening for HCV infection?
4. What are the harms associated with screening for HCV infection, including diagnostic liver biopsies?
Detailed methods and data for the review, including search strategies, detailed inclusion criteria, data abstraction tables, and tables with quality ratings of individual studies, are available in the full report, which includes the analytic framework and additional key questions (26). The protocol was developed using a standardized process with input from experts and the public. The analytic framework focuses on direct evidence that HCV screening improves important health outcomes compared with not screening, as well as the chain of indirect evidence (diagnostic accuracy of screening, clinical utility and harms of subsequent testing in HCV-infected persons, and benefits and harms of treatments) linking screening with improved health outcomes. Key questions related to risk modification of mother-to-infant transmission are presented in the full report (26) and in a separate article (28). We did not re-review the diagnostic accuracy of HCV antibody testing, which the 2004 USPSTF review found to be high (21).
Data Sources and Searches
A research librarian searched Ovid MEDLINE (1947 to May 2012), Embase, the Cochrane Library Database, Scopus, and PsycINFO; clinical trial registries (including ClinicalTrials.gov); and grants databases. We supplemented electronic searches by reviewing reference lists of retrieved articles.
At least 2 reviewers independently evaluated each study to determine inclusion eligibility. Papers were selected for full review if they were relevant to a key question and met the predefined inclusion criteria. For screening, we included randomized trials, cohort studies, case-control studies, and cross-sectional studies that compared different screening strategies in asymptomatic adults without known liver enzyme abnormalities and reported clinical outcomes or sufficient information to compute the sensitivity and number needed to screen to identify 1 HCV-infected person. We also included large studies (sample size >1000 participants) reporting harms associated with diagnostic liver biopsy published since 2004 and uncontrolled or controlled studies reporting direct harms associated with screening.
Clinical outcomes were mortality, end-stage liver disease, cirrhosis, hepatocellular carcinoma, need for transplantation, quality of life, HCV transmission, harms associated with screening (such as anxiety, labeling, and effects on quality of life), and harms associated with liver biopsy (including death, bleeding, and severe pain).
We restricted inclusion to English-language articles and excluded studies published only as abstracts. We excluded studies of posttransplant patients, HIV-infected patients, patients undergoing hemodialysis, and persons with occupational exposures, in whom screening and treatment considerations may differ from those in the general population (29 - 33).
Data Abstraction and Quality Rating
One investigator abstracted details about the study design, patient population, setting, interventions, analysis, follow-up, and results. A second investigator reviewed data for accuracy. Two investigators independently applied predefined criteria (34 - 36) to assess the quality of each study as good, fair, or poor. Discrepancies were resolved through a consensus process.
For studies reporting the diagnostic yield of different screening strategies, we computed the number needed to screen to identify 1 case of HCV infection by dividing the number of screening tests performed by the number of HCV cases identified. The proportion screened was the number of patients screened upon application of a particular screening strategy, divided by the total number of patients assessed.
We assessed the overall strength of each body of evidence as "high," "moderate," "low," or "insufficient" in accordance with the AHRQ "Methods Guide for Effectiveness and Comparative Effectiveness Reviews" (37), based on the quality of studies, consistency between studies, precision of estimates, and directness of evidence.
Role of the Funding Source
This research was funded by AHRQ's Effective Health Care Program. Investigators worked with AHRQ staff to develop and refine the scope, analytic framework, and key questions. AHRQ staff had no role in study selection, quality assessment, synthesis, or development of conclusions. AHRQ staff provided project oversight, distributed the draft report for peer review, and reviewed the draft report and manuscript. The investigators are solely responsible for the content of the manuscript and the decision to submit for publication.
The evidence reviewed in this report is summarized in Table 3. As in the 2004 USPSTF review (21), we found no direct evidence on effects of HCV screening versus no screening on clinical outcomes, or on the comparison of clinical effects of alternative screening strategies. Retrospective studies found that screening strategies targeting multiple risk factors were associated with sensitivities exceeding 90% and numbers needed to screen to identify 1 case of HCV infection of less than 20 (38 - 41). More narrowly targeted alternative screening strategies (such as screening only persons with a history of injection drug use) were associated with numbers needed to screen of less than 2, but they missed up to two thirds of infected patients.
Although direct harms of screening seem minimal, such harms as labeling, anxiety, and stigmatization remain poorly studied and difficult to quantify (63 - 65). Harms of biopsy include a risk for death of less than 0.2% and serious complications (primarily bleeding and severe pain) in about 1%(48,51 - 55). As detailed in our full report, noninvasive tests have fair to good accuracy for diagnosing fibrosis and good to excellent accuracy for diagnosing cirrhosis compared with liver biopsy (26). Although clinical practice has evolved toward less routine use of liver biopsy before antiviral therapy and the proportion of HCV-infected patients undergoing liver biopsy has decreased overall, no study reported the proportion of screen-detected patients who undergo biopsy. Thus, it is difficult to determine the magnitude of harms associated with liver biopsy subsequent to screening.
In the absence of direct evidence on clinical outcomes associated with screening, an indirect chain of evidence showing the availability of accurate diagnostic tests and effective treatments could link screening with improvements in clinical outcomes. The 2004 USPSTF review found HCV antibody testing to be highly accurate (21). Much of the benefits from screening are likely to be based on the effectiveness of antiviral treatments, including newly approved direct-acting antiviral agents, which are addressed in a separate review (27). Therefore, screening recommendations should be based on the evidence for screening and treatment in totality (27). Studies showing that screening or subsequent interventions are associated with decreased transmission risk could also significantly affect estimates of potential benefits, but these are not yet available (26).
Our study has limitations. We excluded non-English-language articles, which could result in language bias, although we identified no non-English-language studies that would have met inclusion criteria. We could not formally assess for publication bias because of small numbers of studies. We also excluded modeling studies, which might be informative for understanding benefits and harms of screening, given the challenges in conducting the large, long-term studies needed to assess clinical outcomes associated with screening. Available evidence regarding screening yield is derived from a few retrospective studies. High or unreported proportions of potentially eligible patients in these observational studies were not included in calculations of screening yield because of unknown HCV status.
The CDC recently recommended that all persons born between 1945 and 1965 be screened for HCV infection, in addition to persons with risk factors for HCV infection (25). The CDC based its recommendation on the prevalence of patients with HCV infection in this birth cohort (accounting for about three quarters of patients with HCV infection in the United States), the high proportion of patients with undiagnosed HCV infection, projected disease burden after several decades of infection, and estimated benefits from antiviral treatments. Although cost-effectiveness analyses suggest that the birth cohort screening approach is highly cost-effective, no clinical data are yet available (13). The CDC's birth cohort approach was not evaluated in the studies included in our review on the yield of alternative screening strategies. Clinical studies that prospectively evaluate the accuracy, yield, and outcomes of alternative HCV screening strategies, including the birth cohort approach, are needed.
The Appendix Figure shows the results of the search and study selection process. No study compared clinical outcomes between individuals screened and not screened for HCV infection or between individuals screened by using different risk- or prevalence-based strategies.
Summary of evidence search and selection.
The flow diagram summarizes the search and selection of articles addressing the following key questions: 1. Does screening for hepatitis C virus (HCV) infection in nonpregnant adults without known abnormal liver enzymes reduce mortality and morbidity due to HCV, affect quality of life, or reduce transmission of HCV? 2. What is the effectiveness of different risk- or prevalence-based methods for screening for HCV infection on clinical outcomes? 3. What is the sensitivity and number needed to screen to identify 1 case of HCV infection of different risk- or prevalence-based methods for screening for HCV infection? 4. What are the harms associated with screening for HCV infection, including diagnostic liver biopsies? Reproduced from reference (26).
Yield of Risk-Based Screening Methods
Four cross-sectional studies (samples sizes ranging from 985 to 3367) provided data to calculate the diagnostic accuracy and yield of alternative HCV screening criteria (Table 1) (38 - 41). Two studies evaluated patients attending sexually transmitted disease clinics (38,41) and 2 evaluated patients attending urban primary care clinics (39 - 40). Three studies evaluated higher-prevalence populations (HCV prevalence, 4.6% to 8.3%) (38 - 40) and 1 a lower-prevalence population (HCV prevalence, 1.0%) (41). One study of patients in primary care and gastroenterology clinics (n = 429) also evaluated alternative screening criteria but used a case-control design (42). All of the studies applied and evaluated alternative screening criteria retrospectively. Other limitations of the studies were that high proportions of potentially eligible patients were not included in analyses because of unknown HCV status or that the study did not report the proportion with unknown HCV status. Although the studies used different criteria for targeted screening, several factors (a personal history of injection drug use, sexual intercourse with an injection drug user, and pre-1992 blood transfusion) were consistently used across studies to identify higher-risk individuals.
One cross-sectional study of a lower-prevalence population in a Dutch sexually transmitted disease clinic (n = 985; HCV seroprevalence, 1%) found that screening based on presence of 1 or more positive items on a 20-item questionnaire was associated with a sensitivity of 90% for identifying persons with HCV infection and a number needed to screen to identify 1 case of HCV infection of 2.4 (Table 2) (41).
Three cross-sectional studies in higher-prevalence populations found that screening strategies targeting multiple risk factors were associated with sensitivities of more than 90% and numbers needed to screen of 9.3 to 18 (Table 2) (38 - 40). One cross-sectional study in a sexually transmitted disease clinic (n = 3367; HCV seroprevalence, 4.9%) found that screening patients with 1 of 5 risk factors (injection drug user, sex partners of injection drug user, received a pre-1992 blood transfusion, bacterial sexually transmitted disease in last 5 years, or age ³30 years) would have resulted in testing 63% of clinic attendees, with a sensitivity of 97% for identifying HCV infection and a number needed to screen of 13 (38). One study of patients in an inner-city primary care clinic (n = 1000; HCV seroprevalence, 8.3%) found that screening patients with positive findings in at least 1 of 3 domains (medical history, exposure history, or social history) would have resulted in screening 71% of the population, with a sensitivity of 92% and a number needed to screen of 9.3 (39). A study of U.S. veterans (n = 2263; HCV seroprevalence, 4.6%) found that screening patients according to presence of 1 or more of 5 risk factors (Vietnam-era veteran, tattoo/body piercing, blood transfusion before 1992, abnormal liver enzyme levels, past or present injection drug use) would have resulted in screening of 78% of the population compared with screening based on the presence of these or 6 additional risk factors (multiple sexual contacts, intemperate alcohol use, intranasal cocaine use, blood exposure [mucous membranes], unexplained liver disease, hemodialysis), with a sensitivity of 97% and number needed to screen of 18 (40).
More narrowly targeted screening strategies evaluated in these studies were associated with specificities of more than 95% and numbers needed to screen of less than 2, but missed up to two thirds of infected patients (38 - 40). Two studies found screening only injection drug users would have resulted in testing of 3.0% or 5.8% of the population, with sensitivities of 41% and 60%, and numbers needed to screen of 1.6 and 1.9, respectively (38,40). One study found screening patients with positive findings in 3 domains (medical, exposure, or social history) would have resulted in testing of 5.6% of the population, with a sensitivity of 34% and number needed to screen of 2.0 (39).
A case-control study (222 cases) found screening based on presence of 4 or more of 7 risk factors (self-reported history of sex with a prostitute, history of exposure to potentially infected blood transfusion, rejections as a blood donor, refused life insurance, witnessed use of injecting drugs, sexual intercourse with an injection drug user, or self-reported hepatitis B virus infection) would have identified 24% of HCV-infected persons, with a specificity of nearly 100% (203 of 204) (42). Screening patients with 1 or more risk factors would have identified 94% of infected persons, with a specificity of 35%.
The 2004 USPSTF review (21) included a post hoc analysis of National Hepatitis Screening Survey data that found that screening using 1 of 3 risk factor models would have identified 53% to 69% of HCV-infected persons (43).
Potential Harms Associated With Screening
Three studies (n = 15 to 161) found diagnosis of HCV infection associated with some negative effects on psychological status, strain on spousal relationships, or binge drinking, but these studies had important shortcomings, including no control group of HCV-infected persons unaware of their status, reliance on retrospective recall, and poorly defined outcomes (44 - 46). A small, fair-quality cross-sectional study (n = 34) included in the 2004 USPSTF review found that HCV-infected intravenous drug users aware of their status reported worse quality of life than those who were unaware (47) of their status.
One study of percutaneous liver biopsies (n = 2740) in HCV-infected patients with compensated cirrhosis and at least moderate fibrosis reported a 1.1% rate of serious adverse events, most commonly bleeding or severe pain, with no deaths (48). Two other small studies (n = 126 and n = 166) included in the 2004 USPSTF review reported no episodes of bleeding, perforation, or death after percutaneous liver biopsy in HCV-infected persons (49 - 50).
In patients undergoing liver biopsy for various indications, large series (n = 1398 to 61 184) published since 2004 reported periprocedural mortality rates of 0% to 0.2% and major complications (primarily bleeding) in 0.3% to 1.0% (51 - 55), consistent with studies included in the 2004 USPSTF review (56 - 62).