Cascade of Care for Hepatitis C Virus Infection Within the US Veterans Health Administration

Abstract

Objectives. We measured the quality of HCV care using a cascade of HCV care model.

Methods. We estimated the number of patients diagnosed with chronic HCV, linked to HCV care, treated with HCV antivirals, and having achieved a sustained virologic response (SVR) in the electronic medical record data from the Veterans Health Administration’s Corporate Data Warehouse and the HCV Clinical Case Registry in 2013.

Results. Of the estimated 233 898 patients with chronic HCV, 77% (181 168) were diagnosed, 69% (160 794) were linked to HCV care, 17% (39 388) were treated with HCV antivirals, and 7% (15 983) had achieved SVR.

Conclusions. This Cascade of HCV Care provides a clinically relevant model to measure the quality of HCV care within a health care system and to compare HCV care across health systems.

Hepatitis C virus infection is the most common blood-borne infection in the United States. An estimated 2.7 to 3.9 million Americans are chronically infected, representing 1.0% to 1.5% of the US population.^1,2 The Veterans Health Administration (VHA) is the single largest provider of HCV care in the United States. The burden of chronic HCV is known to be higher among patients seen in the VHA than among the general population.³

Prior to 2014, HCV antiviral therapies required long treatment durations, had numerous treatment-limiting side effects, and resulted in sustained virologic response (SVR) or cure rates of approximately 50% overall; cure rates were even lower among those with genotype 1 infection,^4,5 the most common genotype in the United States. New direct-acting antiviral agents require shorter durations of therapy, have favorable side effect profiles, and yield cure rates of greater than 90% in clinical trials, even among genotype 1–infected patients.^6,7 Curing chronic HCV is associated with reduced all-cause mortality.^8–10

With widespread use of more efficacious treatment regimens, it is helpful to use a population health approach to assess a health care system’s ability to identify individuals with chronic HCV, link them to care, provide antiviral treatment, and achieve clinical cure. Doing so allows for identification of gaps in care and of foci for improvement. Yehia et al. proposed a 7-step cascade of care methodology that estimated that among individuals in the United States with chronic HCV, 16% had received antiviral therapy and 9% had achieved SVR.¹¹

The cascade proposed by Yehia et al. may be limited in its applicability to large health care systems. For example, insurance status was used as a marker of access to outpatient care, which may not be applicable for a health care system trying to assess performance across the Cascade of Care among enrolled patients who, by definition, have health insurance. In addition, liver biopsy was included in their model, which in many cases may no longer be indicated. Finally, because their model concentrated on the entire US population, national seroprevalence surveys and patient self-reported survey data were used, which is not routinely available in a health care system.

We sought to create a cascade of HCV care that captured the current state of HCV care using steps applicable to a large health care system and that could be calculated with data generally available from an electronic medical record. We included an assessment of linkage to HCV-specific care because it is a more stringent measure than access to insurance or general medical care. We did not include patient awareness of infection, access to care, or liver biopsy as steps in our cascade. We developed and applied our cascade of HCV care in the VHA health care system.

METHODS

To create our cascade of HCV care, we determined the following:

1. number of patients with chronic HCV,

2. number of patients diagnosed with chronic HCV,

3. number of patients linked to HCV care,

4. number of patients treated with HCV antivirals, and

5. number of patients with an SVR.

For the first 2 steps, we used the VHA’s Corporate Data Warehouse, a national database that captures birth dates, gender, race, ethnicity, health care encounters, and laboratory tests from October 1, 1999, onward for patients who received VHA care. The analysis included all patients in VHA care in 2013, where “in care” was defined as having at least 1 VHA outpatient visit (including telehealth) in 2013. We required only 1 outpatient visit in 2013 to capture as broad a population as possible while still ensuring that the patients were using VHA health care to some extent. We identified all HCV antibody, viral load, and genotype tests that had been obtained between October 1, 1999, and December 31, 2013, for patients in care in 2013. We considered individual test results “informative” if the result could be categorized as negative or positive for antibody tests, detectable or undetectable for viral load tests, and detectable for genotype tests if a recognized HCV genotype was reported. For each patient in care in 2013, we determined if the patient ever had positive or negative antibody tests, and detectable or undetectable HCV RNA testing from all identified tests. We considered patients with a positive enzyme immunoassay HCV antibody test result who had a subsequent negative recombinant immunoblot assay test result to have negative antibody testing. For this analysis, we accessed Corporate Data Warehouse data on January 15, 2014.

Number of Patients With Chronic HCV

The estimated total number of patients with chronic HCV in the population receiving VHA health care in 2013 is the sum of 3 subgroups: (1) those already identified with chronic HCV, (2) estimated additional cases from the projected prevalence among HCV antibody-positive patients who had not received RNA testing, and (3) estimated additional cases from the projected prevalence among the untested population.

Patients already identified with chronic HCV.

Patients were considered to have chronic HCV if they ever had positive RNA-based testing (a detectable viral load or genotype).

Estimated additional cases from the projected prevalence among HCV antibody-positive patients who had not received RNA testing.

We applied the observed prevalence of chronic HCV among HCV antibody-positive patients who had received RNA testing to the population that was HCV antibody positive but had not received RNA testing to estimate the prevalence of HCV in the latter population.

Estimated additional cases from the projected prevalence among the untested population.

To project the prevalence of HCV among untested individuals (i.e., individuals without any HCV antibody, viral load, or genotype tests), we first determined chronic HCV incident diagnosis rates per year of testing for those in care in 2013 (i.e., the number of people with their first detectable HCV RNA test result in the year divided by the total number of people with informative HCV testing in the year). Since chronic HCV prevalence is known to vary by gender, race/ethnicity, and birth cohort (born before 1945, born 1945–1965, and born after 1965),^12,13 we determined incident diagnosis rates separately among these demographic subgroups.

In addition, since incident diagnosis rates in the VHA are decreasing over time as more people undergo HCV testing, we approximated the estimated prevalence rate for the untested population as the incident diagnosis rate in 2013 multiplied by the ratio of the 2013:2012 rates.¹² For those subgroups for which the incident diagnosis rate in 2013 was actually higher than the rate in 2012, we used the 2013 rate as the estimated prevalence rate for the untested population. We calculated estimates of the projected prevalence of chronic HCV in the untested population by applying the projected prevalence rate to the people in the untested population. For example, among non-Hispanic White men born from 1945 to 1965 who were in care in 2013, 72 723 had their first informative HCV testing in 2012, of whom 3686 had detectable HCV RNA, for an incident diagnosis rate of 5.07; 77 083 had their first informative HCV testing in 2013, of whom 3304 had detectable HCV RNA, for an incident diagnosis rate of 4.29. Thus, the projected prevalence rate among those untested in this demographic cohort is the ratio of the 2013:2012 rates multiplied by the rate in 2013 ([4.29/5.07] × 4.29 = 3.62). Applying the projected prevalence rate of 3.62 to the 567 756 non-Hispanic White men born between 1945 and 1965 in care in 2013 and untested for HCV yields 20 553 additional cases of chronic HCV that would be found with complete testing of this demographic cohort.

Number of Patients Diagnosed With Chronic HCV

The number of patients in care in 2013 diagnosed with chronic HCV included those patients who had ever had a detectable viral load or genotype.

Number of Patients Linked to HCV Care

The VHA maintains a registry of HCV patients known as the HCV Clinical Case Registry, which was originally designed in part to ensure that patients with chronic HCV were linked to HCV care.¹⁴ The registry software identifies patients as potentially having HCV on the basis of International Classification of Diseases, Ninth Revision (ICD-9)¹⁵ codes or laboratory results and adds such patients to the list of “pending” patients. HCV Clinical Case Registry coordinators at each facility review the lists of pending patients to confirm the diagnosis of HCV. The HCV Clinical Case Registry also extracts information from the VHA electronic medical record, which includes laboratory results, pharmacy information, and ICD-9 diagnosis codes from inpatient hospitalizations, outpatient visits, and problem lists of HCV-infected patients seen at all VHA medical facilities. HCV Clinical Case Registry data for this analysis was available through December 31, 2013.

We considered a patient with an outpatient visit in 2013 to be linked to HCV care if the patient was entered into the HCV Clinical Case Registry and had HCV entered on his or her problem list in the electronic medical record.

Number of Patients Treated With HCV Antivirals

Among patients in the HCV Clinical Case Registry linked to HCV care in 2013, we determined the number who had ever received HCV antiviral medications, including boceprevir, consensus interferon, interferon, pegylated interferon, ribavirin, and telaprevir, from the VHA at any time up to and including December 31, 2013.

Number of Patients With an SVR

We evaluated SVR rates using all HCV RNA results available after the end of treatment for those patients linked to HCV care in 2013 who had a calculated end of treatment included in the available data. We calculated the end of treatment as the last day covered by any HCV antiviral medication, using the cumulative number of days of medication supplied and the dispensed dates. We considered that patients with an undetectable HCV RNA on all tests after the end of treatment, including at least 1 test 12 weeks or more after the end of treatment, had achieved SVR; patients with a detectable HCV RNA after the end of treatment were considered not to have SVR. SVR status could not be definitively determined for patients who were still on treatment at the end of the available date, who had not had HCV RNA testing after the end of treatment, or who had an undetectable HCV RNA result after the end of treatment but no test 12 weeks or more after that date. We then applied the SVR rate among those evaluable for SVR to all those who had started HCV antiviral treatment to determine the number of patients in care in 2013 who would be expected to achieve SVR.

RESULTS

Results for the 5 steps of our cascade of HCV care appear in Figure 1. Comparisons to national estimates for these same 5 steps appear in Figure 2.

FIGURE 1—

Cascade of HCV care in the Veterans Health Administration (VHA) in 2013 (n = 233 898).

Note. SVR = sustained virologic response. The proportion of patients in each step of the cascade from the patients in the preceding step is presented in the arrows between each bar.

^aChronic HCV was estimated from the sum of those already identified with chronic HCV plus estimated additional cases from projected prevalence among HCV antibody-positive patients who had not had RNA testing plus estimated additional cases from projected prevalence among the untested population.

^b“Diagnosed with chronic HCV” was defined as ever had a detectable HCV RNA or genotype.

^c“Linked to HCV care” required an outpatient visit in 2013, entry in the VHA’s HCV registry, and HCV entered on the patient’s medical record problem list.

^d“Treated with HCV antivirals” was defined as ever received HCV antivirals from the VHA as of December 31, 2013.

^e“Achieved SVR” was defined as undetectable HCV RNA on all tests after end of treatment, including at least 1 test at least 12 weeks after the end of treatment, with the SVR rate among those evaluable for SVR applied to those without definitive SVR status.

FIGURE 2—

Cascade of HCV care within the Veterans Health Administration (VHA) and relative to national US estimates in 2013.

Note. SVR = sustained virologic response. Total estimated chronic HCV population in the VHA in 2013 was 233 898.

Source. National US estimates were derived from the following sources: diagnosed with chronic HCV,¹⁶ linked to HCV care,¹⁷ treated with HCV antivirals and achieved SVR.¹¹ VHA estimates were derived from the VHA Corporate Data Warehouse and the HCV Clinical Case Registry.

^a“Diagnosed with chronic HCV” was defined as ever had a detectable HCV RNA or genotype.

^b“Linked to HCV care” required an outpatient visit in 2013, entry in the VHA’s HCV registry, and HCV entered on the patient’s medical record problem list.

^c“Treated with HCV antivirals” was defined as ever received HCV antivirals from the VHA as of December 31, 2013.

^d“Achieved SVR” was defined as undetectable HCV RNA on all tests after end of treatment, including at least 1 test at least 12 weeks after the end of treatment, with the SVR rate among those evaluable for SVR applied to those without definitive SVR status.

*P < .001.

Number of Patients With Chronic HCV

Overall, 5 596 178 patients had a VHA outpatient visit in 2013 and 3 135 396 (56.0%) had ever had any HCV testing (HCV antibody, viral load, or genotype) in the VHA; 3 117 014 (99.4% of those with any testing) had HCV antibody testing, 502 378 (16.0%) had viral load testing, and 191 552 (6.1%) had genotype testing.

Patients already identified with chronic HCV.

Of 218 111 patients with positive antibody tests, 208 893 (95.8%) had RNA-based testing for chronic HCV. Of those with positive antibody tests who received RNA-based testing, 161 044 (77.1%) had positive RNA test results and therefore had chronic HCV. We identified an additional 20 124 patients with chronic HCV from RNA testing without preceding antibody testing. Overall, we identified 181 168 individuals who had an outpatient visit in 2013 as ever having chronic HCV, for a prevalence of 5.8% among those with informative testing.

Estimated additional cases from the projected prevalence among HCV antibody-positive patients who had not received RNA testing.

Among those with positive antibody results, 9218 patients did not have RNA-based testing. Applying the observed prevalence among HCV antibody-positive individuals who did have RNA-based testing (77.1%) to the 9218 who did not have RNA-based testing, we estimated that an additional 7107 individuals would be identified with chronic HCV by RNA-based testing of these individuals.

Estimated additional cases from the projected prevalence among the untested population.

Of the patients in VHA care in 2013, 2 460 782 had not been tested for HCV (excluding the 552 patients without gender [n = 26] or birth-date information [n = 526]). If the projected prevalence rates in each demographic cohort were applied, an estimated additional 45 623 patients with chronic HCV would be identified among the untested population in VHA care (Table 1).

TABLE 1—

Projected Prevalence of HCV in 2013 Among Veterans Health Administration Patients Who Were Previously Untested, by Sociodemographic Characteristics: United States

	Born Before 1945			Born Between 1945–1965			Born After 1965
Characteristic	No.	Projected Prevalence,a %	Estimated No. of Cases	No.	Projected Prevalence,a %	Estimated No. of Cases	No.	Projected Prevalence,a %	Estimated No. of Cases
Women
American Indian/Alaska Native	122	0.00	0	618	0.84	5	819	0.00	0
Asian	99	0.00	0	505	0.90	5	1 447	0.00	0
Black	1 184	0.00	0	16 940	1.18	200	19 323	0.13	25
White	15 190	0.05	8	34 294	1.04	357	33 605	0.54	181
Hispanic	390	0.00	0	2 103	0.64	13	6 136	0.13	8
Native Hawaiian/Pacific Islander	148	0.00	0	443	0.50	2	687	0.00	0
Mixed, other, or unknown	2 675	1.52	41	8 514	1.92	163	10 896	0.22	24
Subtotal			49			745			238
Men
American Indian/Alaska Native	4 497	1.71	77	6 379	5.87	374	3 185	0.26	8
Asian	6 157	0.58	36	6 734	0.72	48	5 728	0.11	6
Black	66 442	1.86	1 236	122 370	6.85	8 382	53 566	0.27	145
White	820 785	0.45	3 694	567 756	3.62	20 553	198 289	0.80	1 586
Hispanic	29 245	1.57	459	35 512	4.84	1 719	28 687	0.41	118
Native Hawaiian/Pacific Islander	6 445	0.67	43	5 529	1.70	94	2 811	0.23	6
Mixed, other, or unknown	169 155	0.86	1 455	113 287	3.90	4 418	51 533	0.26	134
Subtotal			7 000			35 588			2 003

Note. The total estimated number of HCV cases among those untested was 45 623.

^aProjected prevalence is based on the incident diagnosis rate in those first tested in 2013 multiplied by the ratio of the 2013:2012 incident diagnosis rates.

The estimated total number of patients with chronic HCV (n = 233 898) thus reflects the sum of (1) those already identified with chronic HCV (n = 181 168), (2) estimated additional cases among those known to have positive HCV antibody without RNA testing (n = 7107), and (3) estimated additional cases among untested patients (n = 45 623).

Number of Patients Diagnosed With Chronic HCV

As noted in the previous paragraph, 181 168 (79.9%) of an estimated 233 898 patients in VHA care in 2013 with chronic HCV had been diagnosed with chronic HCV.

Number of Patients Linked to HCV Care

In 2013, 172 857 patients in the HCV Clinical Case Registry had at least 1 outpatient visit in the year. Of these, 160 794 had HCV entered on their problem list. By this definition, 88.8% of all patients in VHA care in 2013 with diagnosed chronic HCV had been linked to HCV care.

Number of Patients Treated With HCV Antivirals

Of the 160 794 patients linked to HCV care, 39 388 (24.5%) had ever received VHA HCV antiviral therapy as of December 31, 2013.

Number of Patients With an SVR

Of the 39 388 patients in care in 2013 who had ever received HCV antivirals from the VHA, SVR status could be determined for 37 069, of whom 15 041 (40.6%) achieved SVR. Conservatively, applying this SVR rate to the 2319 who did not have definitive SVR status and adding it to those known to have SVR, the projected number with SVR among those in care in 2013 who had received HCV antivirals was 15 983, or 6.8% of the entire population projected to have chronic HCV.

DISCUSSION

The high prevalence of chronic HCV, the rapidly changing treatment climate, and the high cost of antiviral therapies all warrant the use of a population health approach to HCV care. Our cascade of HCV care demonstrates that the VHA performs very well in the initial steps of the cascade and outperforms national US estimates (Figure 2). For example, the VHA has diagnosed an estimated 77.5% of its patients in care with chronic HCV. This compares with national US estimates from the Chronic Hepatitis Cohort Study that suggest that only 57% of those with chronic HCV have been diagnosed.¹⁶ The VHA also performs well in linking patients to HCV care: 69% of patients with chronic HCV (89% of those with diagnosed chronic HCV) have been linked to HCV care. National US estimates indicate that only 38% of patients with chronic HCV (77% of those with diagnosed chronic HCV) have been linked to HCV care, when linkage to care is defined as self-reporting a visit with a health care provider after receiving a positive HCV test.¹⁷ We posit that the decrement between being diagnosed with chronic HCV and being linked to HCV care is smaller within the VHA (a decrement from 77% to 69% within VHA, vs 57% to 38% within the United States; Figure 2), in part because of higher HCV diagnosis rates within VHA, as well as differences in definitions of “linked to care.” A clinically meaningful, readily accessible measure of “linked to care” is difficult to define; we believe our definition is most appropriate given the availability of the Clinical Case Registry as a resource, as well as the relative ease of using electronic medical records to extract problem list entries.

This cascade illuminates important opportunities to improve care within a health care system. Within the VHA, as of the end of 2013 before the availability of interferon-free regimens, only 17% of patients with chronic HCV (24% of those linked to HCV care) had received antiviral treatment. This figure is similar to that of Yehia et al., who estimated that 16% of HCV-infected individuals have received treatment in the United States.¹¹ The model also documents the historically low SVR rates from prior interferon-based HCV antiviral regimens. In the VHA, approximately 41% of the patients who had ever been treated with HCV antivirals through Dec 31, 2013—and 7% of all patients with chronic HCV—achieved SVR, comparable to the estimated national SVR rates of 9%.¹¹ Although all steps of the cascade require attention, within the VHA health care system increasing treatment initiation will have the largest impact on increasing cure rates. Both the treatment rates and SVR rates have already improved markedly with the availability of shorter-duration, highly efficacious treatment regimens.

This cascade provides a health care system with a method to assess and monitor over time the quality of its HCV care and to identify performance metrics that require improvement. The information this model generates is useful across a broad range of reporting units, so that a large health care system may identify variation between regions or facilities and appropriate targets for performance improvement interventions. The cascade uses data that are typically available within a health care system’s electronic medical record, thereby encouraging health care systems to use this model to monitor their own HCV care and enabling them to compare the quality of their HCV care with that of other systems. Additionally, since it mirrors the flow of clinical care from diagnosis to cure, it provides relevant information to clinicians and administrators: How many patients do we have with chronic hepatitis C? How many have been treated? How many have been cured?

Although our model offers many benefits, there are limitations to this approach. Even after incorporating the decline in incident diagnosis rates from 2012 to 2013 and the variability by gender, race/ethnicity, and birth cohort, the projected prevalence rates for the untested likely still overestimate the number of additional cases that would be identified with complete population testing, particularly in the birth cohorts born before 1945 and after 1965. For these 2 cohorts, risk-based screening is recommended.¹⁸ Thus, the observed incident diagnosis rates are likely higher for these patients who have been identified as at risk and tested compared with the rate that would be observed among those who are untested and theoretically not at risk. Even with this overestimation, however, only 9290 (20.4%) of the estimated 45 623 additional cases that would be found with testing the untested occur among those born outside the 1945 to 1965 birth cohort. Although changes in the estimates of the undiagnosed would change the absolute percentages, the percentage changes from preceding steps along the cascade would not change. The percentages of each preceding step provide useful information for identifying gaps in care—for example, regardless of the estimate of the undiagnosed, only 24% of those linked to HCV care have received antivirals.

Another limitation is that our definition of “linked to care” relies partly on the entry of patients with chronic HCV into the HCV Clinical Case Registry, a VHA registry of individuals with chronic HCV that is not available in other health care systems. However, other health care systems may have similar ways of tracking HCV among their patient populations. Finally, the SVR estimates provided here reflect a static number that relies heavily on patients who received therapy during a period of less efficacious interferon-based treatments. Although SVR rates are increasing with the rapid uptake of more efficacious regimens, there will be a delay in observing the impact of this increased effectiveness at the population level. Moreover, our SVR estimates would not capture patients who received HCV antiviral therapy and achieved SVR outside the Department of Veterans Affairs health care system.

Our cascade of HCV care offers health care systems the ability to readily measure and monitor performance during a period of rapid change in the field of HCV treatment and care. The dramatic improvements in HCV treatment effectiveness pose parallel challenges to public and private health care systems grappling with high treatment costs and inadequate numbers of trained providers to meet the demand for HCV care. This cascade is a useful tool for systems aimed at maximizing provider and fiscal resources to improve the overall quality of HCV care.