Performance of Risk-Based and Birth-Cohort Strategies for Identifying Hepatitis C Virus Infection Among People Entering Prison, Wisconsin, 2014

Abstract

Objectives

The prevalence of hepatitis C virus (HCV) infection among young adults is rising in Wisconsin. We examined the prevalence of HCV infection among male and female inmates entering two Wisconsin prisons and evaluated existing and alternate risk-based strategies for identifying HCV infection at intake.

Methods

We added HCV testing to the intake procedures for all 1,239 adults prison entrants at the Wisconsin Department of Corrections (WDOC) from November 3, 2014, to January 31, 2015. We identified risk factors associated with HCV infection during the routine intake examination and calculated the sensitivity and specificity of risk-based testing strategies for identifying HCV infection.

Results

The prevalence of HCV antibody among prison entrants was 12.5% (95% confidence interval [CI] 10.7, 14.4) overall and was almost two times higher at the women's facility (21.3%, 95% CI 15.4, 27.2) than at the men's facility (11.0%, 95% CI 0.0, 12.9) (p<0.001). The sensitivity and specificity of the WDOC risk-based criteria were 88% (95% CI 83, 93) and 80% (95% CI 78, 83), respectively. Adding a new criterion, the 1945–1965 birth cohort, to the risk-based criteria improved the sensitivity to 92% (95% CI 88, 96) and lowered the specificity to 71% (95% CI 68, 74). Compared with entrants without these risk factors, HCV antibody prevalence was significantly higher among prison entrants who had the following risk factors: injection drug use (prevalence ratio [PR] = 9.9, 95% CI 7.4, 13.2), liver disease (PR=9.7, 95% CI 7.8, 12.0), and elevated levels of alanine transaminase (PR=3.6, 95% CI 2.7, 4.9).

Conclusion

The WDOC risk criteria for HCV testing identified 88% of HCV infections among prison entrants. Including the 1945–1965 birth cohort as a criterion along with the other WDOC risk criteria increased the sensitivity of targeted testing to 92%. These findings may be informative to jurisdictions where universal HCV testing is not feasible because of resource limitations.

In 2010, an estimated 5 million people in the United States had hepatitis C virus (HCV) infection, the leading cause of liver failure and hepatocellular carcinoma.¹ People born during 1945–1965 have the largest share of chronic morbidity and mortality from HCV infection; three out of four people with HCV infection are in this birth cohort. A rise in the number of young adults with HCV infection during 2006–2012 was fueled by epidemics of injection drug use (IDU).^2–6 The prevalence of HCV infection in U.S. prisons ranges from 10% to 41%, up to 20 times higher than in the general population.^7–9 In 2012, liver disease was the third-ranking cause of death among the U.S. prison population, surpassed only by heart disease and cancer,¹⁰ and is associated with rising death rates after incarceration.¹¹ Increasing morbidity attributed to HCV infection among older adults and increasing prevalence of HCV infection among young adults have been noted in some prison populations.^4,12

One-third of people living with HCV infection are estimated to have passed through the U.S. correctional system,¹³ placing this system in a key position to identify infection; however, few correctional facilities conduct routine HCV testing.¹⁴ Targeted or risk-based testing, which focuses testing on those with the highest likelihood of HCV infection, is more common than routine testing. However, the reliability of self-reported IDU and the consistency with which the screening criteria are applied may decrease the ability of institutions with targeted screening to identify people with HCV infection.¹⁵ Although the Centers for Disease Control and Prevention and U.S. Preventive Services Task Force recommend one-time HCV testing for all adults born during 1945–1965, the impact and efficiency of this strategy in correctional facilities are unclear.^12,16

In Wisconsin during 2004–2014, an average of 2,600 people with HCV infection were reported to the Wisconsin Division of Public Health annually. The rate of reported HCV infection in Wisconsin among people aged 15–30 years increased 500% during that time period, from 13 to 78 cases per 100,000 population. Surveillance data indicate that most HCV infections in this age group were caused by recent IDU.^2,3

Twelve of the 59 deaths at Wisconsin Department of Corrections (WDOC) in 2014 were liver disease-related (Personal communication, James Greer, WDOC, June 2015), and the morbidity of untreated HCV infection is projected to escalate during the next decade. From 2001 through 2014, WDOC conducted targeted HCV testing in state correctional facilities based on medical history and self-reported risk factors. This strategy, also described in the Federal Bureau of Prisons Guidelines,¹⁷ is intended to maximize the efficiency of HCV testing where resources are limited. To estimate HCV infection prevalence at intake to state correctional facilities and to compare the performance of targeted testing with universal HCV testing, WDOC offered HCV testing to all adults at entry to its two primary prison intake facilities in Wisconsin during a three-month period. These data can be used to inform the development of strategies to improve detection of HCV infection in correctional institutions.

METHODS

Population

WDOC has two intake processing centers, one for men (aged ≥18 years) and another for women (aged ≥18 years), through which approximately 7,800 adult inmates pass each year. Upon admission to these facilities, each prisoner receives a clinical evaluation, which includes a physical examination, a test for hepatitis B virus core antibody (anti-HBc), and an offer for human immunodeficiency virus (HIV) testing. From 2001 through 2014, physicians at WDOC ordered HCV antibody (anti-HCV) and confirmatory HCV ribonucleic acid (HCV RNA) testing at intake according to risk-based criteria. Risk-based criteria included having a history of IDU, having elevated liver enzymes (alanine transaminase), being anti-HBc positive, being HIV positive, or having a history of liver disease. In 2013, the year preceding the evaluation, approximately 2,700 of 7,800 (35%) inmates entering the two adult intake facilities received HCV testing based on these criteria.

Data collection

We performed sample size calculations to determine how many inmates would need to be tested during intake to yield an estimate of the prevalence of anti-HCV within a 2% margin of error and with 95% confidence. To satisfy these numbers, HCV testing was added to the routine intake procedures for all entrants from November 3, 2014, through December 31, 2014, at the men's intake facility and from November 3, 2014, through January 31, 2015, at the women's intake facility. No other changes to intake procedures were made during the evaluation period.

Testing sequence for HCV

Laboratory methods to identify HCV infection have been established by the Centers for Disease Control and Prevention.¹⁸ Briefly, testing for HCV infection begins with an anti-HCV test. A person with a positive anti-HCV test may either (1) have chronic HCV infection or (2) have had HCV infection in the past (i.e., resolved or treated successfully). For those who test anti-HCV positive, a follow-up HCV RNA test is needed to distinguish chronic HCV infection (i.e., anti-HCV reactive and HCV RNA positive) from past infection that has resolved (i.e., anti-HCV reactive and HCV RNA negative).

Serum samples were processed at the health services unit of each facility, and all HCV testing was performed off-site by the Wisconsin State Laboratory of Hygiene. Anti-HCV was detected by Architect Anti-HCV (Abbott Diagnostics, Abbott Park, Illinois) for the qualitative detection of immunoglobulin G and immunoglobulin M antibodies to HCV. Results of anti-HCV testing were reported as reactive-high, reactive-low, nonreactive, or indeterminate. A result of reactive-high or reactive-low indicated the presence of anti-HCV and was considered an HCV infection for this evaluation. If the anti-HCV result was positive, the intake serum sample was used for HCV RNA testing by polymerase chain reaction using the COBAS TaqMan Hepatitis C Virus Test, version 2.0 (Roche Molecular Systems, Inc., Indianapolis, Indiana). Results of HCV RNA testing were reported as HCV RNA detected or no HCV RNA detected. A result of HCV RNA detected was considered an HCV chronic infection for the evaluation. Results of anti-HCV and HCV RNA testing were reported to the individual by WDOC health-care staff members according to the standard protocol for notifying an inmate of infectious illness and post-counseling. Positive anti-HCV and HCV RNA results were reported to the Wisconsin Division of Public Health in accordance with Wisconsin state statute §252.05.¹⁹

Analysis

We calculated the number of HCV infections identified by individual risk factors to compare the effectiveness of WDOC risk-based testing with universal testing. We selected risk factors a priori and included clinical variables from the intake examination (i.e., physician note that the ALT level was elevated and the result of the anti-HBc test), demographic characteristics (i.e., birth year, sex, race, and ethnicity), and self-reported risk factors (i.e., previous incarceration, history of liver disease, HIV infection, and IDU).

We calculated the prevalence of HCV infection and HCV chronic infection as the proportion of prison entrants tested who had positive anti-HCV and HCV RNA results, respectively. We calculated prevalence estimates and 95% confidence intervals (CIs) for the overall population and for subgroups based on sex, age, race, ethnicity, and risk factor. We calculated the sensitivity of risk-based testing as the number of adults who had HCV infection and who met risk-based testing criteria divided by the overall number of adults infected. We calculated the specificity of risk-based testing as the number of adults who did not have HCV infection and who did not meet risk-based testing criteria divided by the number not infected. We calculated sensitivity and specificity of risk-based testing criteria for all prison entrants at intake and separately for entrants by sex. We conducted an extreme-value sensitivity analysis to account for the infection status of a small number of people for whom HCV testing was not performed at intake.

We calculated prevalence ratios (PRs) and 95% CIs to measure the relative prevalence of HCV infection among prison entrants who had a demographic or behavioral risk factor compared with prison entrants who did not have the risk factor. We used SAS^® version 9.3²⁰ to build univariate log-binomial regression models to estimate PRs. We included birth cohort, race, and ethnicity, three demographic variables found to be associated with HCV infection at a significance level of 0.05 in univariate analysis, as dichotomous variables in the multivariate model to determine if associations with risk factors were independent of these characteristics. We stratified models by facility to calculate PRs for men and women separately. We used two months of intake testing data to estimate the annual number of inmates who would receive an anti-HCV test based on WDOC risk criteria and 1945–1965 birth-cohort testing.

RESULTS

Prevalence of HCV infection at intake

During the evaluation period, 1,250 people were admitted to the two adult correctional intake facilities and 1,239 were tested for anti-HCV. We did not include 11 people without anti-HCV test results in the evaluation: seven did not have a serum sample collected at intake and four did not have anti-HCV testing indicated on the laboratory requisition order. Of the 1,239 prison entrants tested for anti-HCV, 155 (12.5%, 95% CI 10.7, 14.4) were anti-HCV positive and 110 (8.9%, 95% CI 7.3, 10.5) were HCV RNA positive. The HCV infection prevalence at intake to the women's facility (21.3%, 95% CI 15.4, 27.2) was almost twice as high as the HCV infection prevalence at the men's facility (11.0%, 95% CI 9.0, 12.9). Of those with evidence of HCV infection, we found no significant difference between men and women in the percentage who were HCV RNA positive: 75.0% (95% CI 67.0, 83.0) of men and 66.7% (95% CI 51.9, 81.5) of women had chronic infection (Table 1). Extreme value sensitivity analysis yielded similar results.

Table 1.

Hepatitis C virus (HCV) infection among all inmates entering men's and women's adult correctional facilities, Wisconsin, November 2014–January 2015^a

^aThe evaluation took place from November 3, 2014, through December 31, 2014, at the men's intake facility and from November 3, 2014, through January 31, 2015, at the women's intake facility.

^bCalculated as the percentage of all adult prison inmates who were tested for anti-HCV who were anti-HCV positive

^cCalculated as the percentage of inmates testing anti-HCV positive at intake who were HCV RNA positive

^dFour inmates who were anti-HCV positive had insufficient quantity of sample for HCV RNA testing.

anti-HCV = HCV antibody

CI = confidence interval

RNA = ribonucleic acid

Performance of targeted testing

Of the 1,239 adults who entered prison during the evaluation period, 351 (28.3%) had at least one of the risk criteria used by WDOC, yielding a sensitivity of 88% (95% CI 83, 93). Of the 1,084 adults with no risk criteria, 869 were anti-HCV negative, yielding a specificity of 80% (95% CI 78, 83).

Nineteen people with HCV infection had none of the WDOC risk criteria. Of these, seven people belonged to the 1945–1965 birth cohort. The 1945–1965 birth-cohort criterion applied to 457 of 1,239 (36.9%) prison entrants during the evaluation period. Adding this criterion to the WDOC criteria identified seven additional HCV infections (143 of the total 155 HCV infections) and increased the sensitivity of the risk-based testing strategy to 92% (95% CI 88, 96) (Figure). Of the 1,084 people without HCV infection, 770 did not meet the 1945–1965 criterion, yielding a specificity of 71% (95% CI 68, 74).

Figure.

Percentage of prison inmates who met risk criteria for HCV testing at intake and percentage of HCV infection identified by risk-based testing strategy and universal testing, two adult intake facilities, Wisconsin, November 2014–January 2015^a,b

^aThe evaluation took place from November 3, 2014, through December 31, 2014, at the men's intake facility and from November 3, 2014, through January 31, 2015, at the women's intake facility.

^bError bars indicate 95% confidence intervals for the sensitivity calculation. Results of the extreme value sensitivity analysis were similar.

^cThe risk-based testing strategy included testing using the following risk-based criteria: reported injection drug use, liver disease, human immunodeficiency virus infection, elevated alanine transaminase noted by physician, and hepatitis B virus core antibody positive. “Birth cohort 1945–1965” and “Risk-based” are not mutually exclusive categories.

^dCalculated as the percentage of 1,239 adult prison entrants who met risk criteria for HCV testing.

^eCalculated as the percentage of 155 anti-HCV-positive adult prison entrants identified at intake among 1,239 entrants.

HCV = hepatitis C virus

WDOC risk-based criteria identified 36 of the 39 HCV infections in the women's facility, yielding an overall sensitivity of 92% (95% CI 84, 100) and specificity of 74% (95% CI 67, 81). The addition of the 1945–1965 birth-cohort criterion identified all 39 women with HCV infection at intake to prison (sensitivity: 100%, 95% CI 100, 100; specificity: 62%, 95% CI 54, 70). At the men's facility, risk-based criteria identified 100 of 116 HCV infections, yielding an overall sensitivity of 86% (95% CI 80, 92) and specificity of 81% (95% CI 79, 84). The addition of the 1945–1965 birth-cohort criterion identified 104 of 116 men infected with HCV at intake (sensitivity: 90%, 95% CI 84, 95; specificity: 72%, 95% CI 70, 75).

At both facilities, 12 prison entrants with HCV infection were not identified by WDOC risk-based criteria and were not in the 1945–1965 birth cohort, all of whom were men, nine of whom were 30–48 years of age, six of whom were non-Hispanic white, and six of whom had been previously incarcerated. Assuming that anti-HCV prevalence at intake is consistent throughout the year, an estimated 72 HCV infections would be missed at intake under a highly sensitive risk-based testing strategy.

Based on two months of testing data, we estimated that adding 1945–1965 birth-cohort testing at intake would result in testing 600 additional inmates annually who have no other reported risk factor and would not otherwise be tested using the WDOC criteria. HCV testing of these inmates was estimated to identify 36 additional anti-HCV-positive inmates annually, of which, based on a rate of 33% of people in the 1945–1965 birth cohort without other reported risk factors who were anti-HCV positive, 12 were estimated to be HCV RNA positive.

Prevalence of HCV infection by risk factor

Prevalence of HCV infection was highest among entrants born during 1945–1965 (24.8%, 95% CI 18.0, 31.6) and was 2.9 times higher among this birth cohort than among those born in 1985 and later (p<0.001). Prevalence was highest among non-Hispanic white prison entrants (18.0%, 95% CI 15.1, 21.0) and was 4.8 times higher among this racial/ethnic group compared with non-Hispanic black entrants (p<0.001). The difference in prevalence by birth cohort, race, and ethnicity was not significant among women. The prevalence of HCV infection among prison entrants with liver disease (PR=9.7, 95% CI 7.8, 12.0), IDU (PR=9.9, 95% CI 7.4, 13.2)), or elevated ALT (PR=3.6, 95% CI 2.7, 4.9) was higher than among prison entrants without these risk factors (Table 2). In a multivariate model, these risk factors were associated with HCV infection at intake, independent of race, ethnicity, and birth cohort.

Table 2.

Characteristics and hepatitis C virus (HCV) infection among inmates entering adult men's and women's correctional facilities, Wisconsin, November 2014–January 2015^a

^aThe evaluation took place from November 3, 2014, through December 31, 2014, at the men's intake facility and from November 3, 2014, through January 31, 2015, at the women's intake facility.

^bPercentage of all inmates entering prison with the characteristic

^cPercentage of all anti-HCV-positive inmates with the characteristic

^dAnti-HCV prevalence was calculated as the percentage of inmates in each characteristic or risk group who were anti-HCV positive at intake. 95% CIs are for the binomial proportion.

^ePrevalence ratios and 95% CIs were calculated to measure the relative prevalence of HCV infection among inmates entering prison with a demographic or behavioral risk factor compared with those without the variable of interest, indicated in this table as the reference group.

^fOther race for males includes Asian American/Pacific Islander and multiracial. Number not shown because of small sample size.

^gRisk factor was significant in a multivariate model accounting for effects of race and age group. Prevalence ratio shown is not adjusted for the effect of race and age group.

^hOther race for females includes American Indian/Alaska Native, Asian American/Pacific Islander, and Hispanic. Number not shown because of small sample size.

anti-HCV = HCV antibody

CI = confidence interval

Ref. = reference group

NA = not applicable

HBV = hepatitis B virus

ALT = alanine transaminase

LESSONS LEARNED

We determined that WDOC risk criteria for HCV testing applied to fewer than one-third of prison entrants and detected 88% of HCV infections. We also determined that including the 1945–1965 birth cohort as a criterion along with other WDOC risk criteria increased the sensitivity of testing to 92%. Adding the 1945–1965 birth cohort rather than other birth cohorts increased sensitivity because it captured data on a group of people with HCV infection who did not have other clinical or behavioral risks. These findings may be informative to jurisdictions where universal opt-out HCV testing is not feasible because of resource limitations.

Even with a highly sensitive risk-based testing approach at WDOC, we estimated that 72 HCV infections annually would go undetected at intake. An epidemic of HCV infection among young adults who inject drugs is occurring, particularly in rural areas of the United States. At least one correctional facility has reported an increased HCV infection rate at intake in recent years.⁴ A testing strategy that fails to identify HCV infection in people who inject drugs misses an opportunity for HCV care and prevention and may ultimately fuel continued HCV transmission, both in prison and at release. Periodic evaluations of prevalence such as this one are important to monitor risk-based approaches and, if missed cases among young adults are found, universal HCV testing in younger age groups might be considered if universal HCV testing for all inmates at intake is not practical.

This evaluation estimated that 11% of men and 21% of women had evidence of HCV infection at intake. Rates in the women's facility were substantially higher than rates in the men's facility and correspond with the frequency with which incarcerated women, particularly young women, report IDU.^4,12,21 Knowledge of this disparity can be a focus for strategies in HCV care and management.

Our prevalence estimate of 12.5% for anti-HCV at intake at both facilities combined is similar to the estimate of 13.5% in a cross-sectional sample of Wisconsin inmates at intake in 1999²² and is low compared with the rates for entrants in other state prison systems (e.g., 30% in Maryland, 24% in Iowa, 23% in Washington, 18% in Pennsylvania, and 14% in Michigan).^8,23 Data from state prison systems that conduct routine HCV testing indicate a decline in HCV seroprevalence, which is attributed to a smaller proportion of individuals entering prison who were born during 1945–1965, the birth cohort with the highest HCV infection prevalence.⁸ As such, targeted testing of the 1945–1965 birth cohort alone is expected to identify a minority of newly incarcerated cases and a decreasing proportion of cases over time.^12,16

Limitations

This study had two limitations. First, analyses were conducted using routinely collected data for a limited and relatively brief period of time at two correctional facilities. The average number of intakes to the Wisconsin prison during the evaluation period was about 10% lower than the average number admitted during the rest of the year. Although the trend in admissions is relatively stable, we do not know the effect that seasonality may have on HCV prevalence or the performance of screening criteria. Second, the performance of targeted testing may have been inflated if data on risk factors were collected more completely during rather than before the evaluation period. Intake nursing staff members, physicians, and laboratory staff members were asked to continue the intake screening examination as usual during the evaluation period.

CONCLUSION

Findings from this evaluation have been incorporated into testing practices by WDOC, which continues to provide targeted testing and community standard-of-care medical management for HCV infection. The use of short-duration, high-efficacy therapies holds promise of a cure for HCV infection, and treatment in the prison population has been demonstrated as cost effective to society.²⁴ For those who do not qualify for treatment while incarcerated, evidence-based educational programs for inmates, including alcohol avoidance and safer injection practices for people who inject drugs, can prevent transmission of HCV at release.