Rates and predictors of hepatitis C virus treatment in HCV–HIV-coinfected subjects


Dr A. A. Butt, 3601 Fifth Avenue, Suite 3A, Falk Medical Building, University of Pittsburgh Medical Center, Pittsburgh, PA 15213, USA.
E-mail: butta@dom.pitt.edu




True treatment rates and the impact of comorbidities on treatment rates for hepatitis C virus in the HCV–HIV-coinfected subjects are unknown.


To quantify the rates of treatment prescription and the effect of comorbidities on hepatitis C virus treatment rates in HCV–HIV-coinfected veterans.


The Veterans Affairs National Patient Care Database was used to identify all hepatitis C virus-infected subjects between 1999 and 2003 using ICD-9 codes. Demographics, comorbidities and pharmacy data were retrieved. We used logistic regression to compare the predictors of hepatitis C virus treatment in hepatitis C virus-monoinfected and HCV–HIV-coinfected subjects.


We identified 120 507 hepatitis C virus-infected subjects, of which 6502 were HIV coinfected. 12% of the hepatitis C virus-monoinfected and 7% of the -coinfected subjects were prescribed hepatitis C virus treatment (P < 0.0001). Those not prescribed treatment were older (48.6 years vs. 47.7 years, P = 0.007) and more likely to be black (52% vs. 32%, P < 0.0001). HIV coinfected was less likely to be prescribed hepatitis C virus treatment (OR 0.74, 95% CI: 0.67–0.82). Among the coinfected subjects, the following were associated with non-treatment (OR, 95% CI): black race (0.45, 0.35–0.57); Hispanic race (0.56, 0.38–0.82); drug use (0.68, 0.53–0.88); anaemia (0.17, 0.11–0.26); bipolar disorder (0.63, 0.40–0.99); major depression (0.72, 0.53–0.99); mild depression (0.47, 0.35–0.62).


A small number of HCV–HIV-coinfected veterans are prescribed treatment for hepatitis C virus. Non-treatment is associated with increasing age, minority race, drug use and psychiatric illness. Further studies are needed to determine the eligibility for treatment and reasons for non-treatment for hepatitis C virus.


Pharmacological treatment for hepatitis C viral infection (HCV) has evolved over the last 20 years. The current standard of care treatment is a combination of pegylated interferon-α and ribavirin, which leads to sustained HCV eradication in 54–56% of the patients in clinical trials.1–3 The sustained eradication rate is lower in HIV-infected subjects, and is reported to be 27–40% overall.4, 5 Despite such advances in pharmacotherapy of HCV, most HCV-infected subjects are not prescribed treatment. Reasons for non-treatment are poorly understood, but involve non-adherence to follow-up visits, medical and psychiatric comorbidities, and ongoing substance use.6–8

True treatment rates for HCV are unknown. Previous studies on treatment patterns, eligibility and reasons for non-treatment have been conducted on subjects referred to tertiary care centres, and geographically limited populations.6–8 Treatment rates are even lower in HIV-coinfected subjects.9 Not all chronically infected subjects are candidates for treatment, and such low treatment rates need to be interpreted with caution, because true eligibility is difficult to determine in observational cohort studies. However, knowledge of true rates and predictors of treatment in larger and more representative populations is important to design future intervention strategies. We undertook this study to determine the rate and predictors of treatment in a national cohort of HCV–HIV-coinfected subjects treated in the United States Department of Veterans Affairs (VA) healthcare system.


We assembled a national cohort of HCV-infected subjects from the VA National Patient Care Database (NPCD), and the VA Pharmacy Benefits Management (PBM) database. The NPCD contains hospitalization records and discharge diagnoses from 1970 onwards, which are coded according to the Clinical Modification of International Classification of Diseases, 9th revision (ICD-9-CM). Starting in 1997, the NPCD also contains out-patient visit records, including diagnoses and clinic visits. The PBM database contains records of all medications prescribed to patients by any VA pharmacy, including the dose and quantity prescribed. The utility and accuracy of the VA administrative data and ICD-9 codes extracted from the NPCD has been previously reported by our group and others.10–14 Diagnostic code agreement and κ-values for HCV were validated in the 3-Site Veterans Aging Cohort Study (VACS 3).14 Compared with the HCV antibody test, the sensitivity and specificity of the codes were 74% and 98%, respectively, and positive and negative predictive values were 96% and 84%. The agreement in these readings was 88% and the κ-value was 0.74. Kramer et al. also reported the positive predictive value of the VA administrative ICD-9 diagnosis to be 94% and the negative predictive value of 90%.11 In the VACS 3 cohort, 880 of the 881 participants were correctly identified as having HIV when compared with chart abstractions and antibody test, providing a very high level of sensitivity.9

Hepatitis C viral infection was defined as having one in-patient or two out-patient ICD-9 codes for HCV on two separate visits. All subjects seen in the VA Healthcare system between fiscal years (FY) 1999–2003 were included. Demographic information including date of birth, recorded race and sex, and information on medical and psychiatric comorbidities and drug and alcohol use was retrieved. Medical and psychiatric comorbidities were also retrieved using ICD-9 codes, and a comorbid condition was determined to be present if at least one in-patient or two out-patient codes were identified. Prescription of interferon-α, pegylated interferon-α, ribavirin and combinations of either type of interferon with ribavirin, including dates of prescription were retrieved from the PBM database. The doses of interferon generally used in the treatment of HCV were retrieved. High-dose interferon-α therapy (>5 million units per dose given ≥3 times per week) was excluded as such higher doses of interferon are not the standard doses used to treat HCV infection.

Prescription for HCV was defined as having received interferon-α, pegylated interferon-α or a combination of either with ribavirin for any duration of time. If multiple courses of treatment were prescribed, only the first prescription was counted. Whether the subjects actually took the treatment as prescribed was not ascertained. As depression and anaemia are well known side-effects of interferon and ribavirin therapy, respectively, we excluded those diagnoses if they were documented after initiation of HCV therapy. We compared the subjects who received treatment for HCV with those who did not receive treatment. The demographic characteristics were compared using the chi-square or the t-test as appropriate. We used univariable and multivariable logistic regression analysis to determine the predictors of treatment for HCV. We used stata (version 8.2, College Station, TX, USA) for statistical analyses.

We determined the frequency of HCV diagnoses and treatment rates by FY. Each subject was counted once for HCV diagnosis in the year the first such diagnosis was recorded. The HIV-coinfected subjects were those among the HCV infected for whom the first HIV diagnosis was recorded in that particular year. Treatment prescription was counted for the FY it was recorded. Multiple courses were counted in each FY they were prescribed.


We identified 120 507 subjects with HCV infection, of which 6502 (5%) were coinfected with HIV. The HCV–HIV-coinfected subjects were younger, more likely to be black race and male (Table 1). The HCV–HIV-coinfected subjects were less frequently diagnosed with coronary artery disease, peripheral vascular disease, hypertension, diabetes, cirrhosis and post-traumatic stress disorder than subjects with HCV infection alone (P < 0.0001 for all comparisons, see Table 1). The HCV–HIV-coinfected subjects were more frequently diagnosed with anaemia, hepatitis B, major depression, mild depression, schizophrenia, alcohol use and drug use (P < 0.0001 for all comparisons). HCV–HIV-coinfected subjects were less likely to have been prescribed treatment for HCV (7% vs. 12%, P < 0.0001; Table 1).

Table 1.   Demographic characteristics and comorbidities in hepatitis C virus (HCV)-monoinfected and HCV–HIV-coinfected veterans
  HCV monoinfected (n = 114 005)HCV–HIV coinfected (n = 6502) P-value
  1. Values are expressed as percentages.

Age, median (mean)50 (51.1)48 (48.5)<0.0001
Gender (male)96.998.2<0.0001
Coronary artery disease13.59.0<0.0001
Peripheral vascular disease4.53.60.001
Pulmonary disease19.319.90.2
Hepatitis B3.09.1<0.0001
Major depression18.423.0<0.0001
Mild depression32.435.4<0.0001
Bipolar disorder10.410.70.5
Post-traumatic stress disorder19.716.5<0.0001
Alcohol use44.348.1<0.0001
Drug use38.956.2<0.0001
Received HCV treatment11.87.2<0.0001

Of the 6502 HCV–HIV-coinfected subjects, 469 (7%) were prescribed treatment for HCV. The HCV–HIV-coinfected subjects who were not prescribed treatment were older (mean age 48.6 years vs. 47.7 years, P = 0.007) and more likely to be black race (52% vs. 32%, P < 0.0001). The following comorbidities were less frequently diagnosed among the HCV–HIV-coinfected subjects who were prescribed treatment: coronary artery disease; pulmonary disease; anaemia; hepatitis B; major depression; mild depression; schizophrenia; bipolar disorder; alcohol use; and drug use (Table 2).

Table 2.   Comparison of hepatitis C virus (HCV)–HIV-coinfected veterans prescribed treatment for HCV with those not prescribed treatment
 Not prescribed treatment (n = 6033)Prescribed treatment (n = 469) P-value
  1. Values are expressed as percentages.

Age, median (mean)48 (48.6)48 (47.7)0.007
Gender (male)98.397.40.2
Coronary artery disease9.26.40.04
Peripheral vascular disease3.72.10.08
Pulmonary disease20.215.40.01
Hepatitis B9.45.10.002
Major depression23.813.0<0.0001
Mild depression36.718.1<0.0001
Bipolar disorder11.14.9<0.0001
Post-traumatic stress disorder16.516.00.8
Alcohol use49.234.5<0.0001
Drug use57.638.4<0.0001

HIV coinfection was associated with a lower odds of being prescribed treatment for HCV in univariable analysis (OR 0.58, 95% CI: 0.53–0.64) as well as multivariable logistic regression analysis (OR 0.74, 95% CI: 0.67–0.82). The predictors of treatment prescription for HCV in the HCV–HIV-coinfected subjects are presented in Table 3. In multivariable logistic regression analysis, the following were predictors of non-treatment for HCV: increasing age (OR per 5 years 0.86, 95% CI: 0.80–0.94); black race (OR 0.45, 95% CI: 0.35–0.57); Hispanic race (OR 0.56, 95% CI: 0.38–0.82); drug use (OR 0.68, 95% CI: 0.53–0.88); anaemia (OR 0.17, 95% CI: 0.11–0.26); bipolar disorder (OR 0.63, 95% CI: 0.40–0.99); major depression (OR 0.72, 95% CI: 0.53–0.99); mild depression (OR 0.47, 95% CI: 0.35–0.62; Table 3). We repeated our analysis with the year of diagnosis as a covariate in the multivariable model and it did not significantly alter our results.

Table 3.   Factors predicting hepatitis C virus (HCV) treatment prescription in the HCV–HIV-coinfected veterans (logistic regression analyses)
 Univariable [odds ratio (95% CI)]Multivariable [odds ratio (95% CI)]
Age, per 5 years0.91 (0.84–0.97)0.86 (0.80–0.94)
 Black0.43 (0.34–0.55)0.45 (0.35–0.57)
 Hispanic0.57 (0.40–0.83)0.56 (0.38–0.82)
 Other/unknown1.2 (0.94–1.6)0.91 (0.70–1.2)
Gender (male)0.67 (0.37–1.2)0.71 (0.37–1.4)
Alcohol use0.54 (0.45–0.66)0.92 (0.71–1.2)
Drug use0.46 (0.38–0.56)0.68 (0.53–0.88)
Anaemia0.18 (0.12–0.26)0.17 (0.11–0.26)
Hepatitis B0.52 (0.34–0.79)0.72 (0.45–1.1)
Cirrhosis1.5 (1.1–2.1)1.9 (1.4–2.7)
Diabetes1.1 (0.89–1.4)1.3 (0.99–1.7)
Coronary artery disease0.68 (0.46–0.99)0.86 (0.57–1.3)
Stroke0.69 (0.42–1.1)0.95 (0.56–1.6)
Pulmonary disease0.72 (0.55–0.93)0.92 (0.70–1.2)
Bipolar disorder0.41 (0.27–0.63)0.63 (0.40–0.99)
Major depression0.48 (0.36–0.63)0.72 (0.53–0.99)
Mild depression0.38 (0.30–0.48)0.47 (0.35–0.62)
Schizophrenia0.50 (0.35–0.73)0.81 (0.54–1.2)
Post-traumatic stress disorder0.96 (0.75–1.2)1.6 (1.2–2.2)
Peripheral vascular disease0.57 (0.30–1.1)0.62 (0.32–1.2)
Hypertension1.0 (0.86–1.3)1.4 (1.1–1.7)

The number of unique HCV-infected subjects increased over 1999–2001, but decreased in 2002 and 2003. The HIV-coinfected subjects among the HCV infected showed the same trend (Figure 1). The number of HCV–HIV-coinfected subjects who were prescribed treatment for HCV for the first time increased over 1999–2003 (Figure 2). This paralleled an increase in treatment for the HCV monoinfected. The proportion of HCV–HIV-coinfected subjects treated for the first time from among the newly diagnosed subjects increased from a low of 0.42% in 1999 to 17% in 2003. The number of new HCV diagnoses, which were HIV coinfected, remained steady from 1999 through 2001 and then decreased in 2002 and 2003. The largest number diagnosed were in 2001 (1531 subjects) and the smallest number in 2003 (766 subjects).

Figure 1.

 Number of unique subjects diagnosed with hepatitis C virus (HCV) per Veterans Affairs (VA) fiscal year (FY), and the number of HIV-coinfected subjects among the HCV infected. The diagnosis of HCV is incident for the FY it was first recorded. The HIV-coinfected subjects are those with a first diagnosis in that FY among the HCV-infected subjects.

Figure 2.

 Number of unique subjects who were prescribed treatment by Veterans Affairs (VA) fiscal year (FY). Treatment prescription is incident for the FY it was prescribed.


The optimal treatment strategy for HCV-infected subjects continues to evolve. While significant improvements in response to pharmacological therapy have been made, not all HCV-infected subjects are prescribed therapy. About 15–25% of HCV-infected patients do not have evidence of ongoing HCV viral replication and do not require treatment. Toxicity of interferon-α and ribavirin precludes therapy in an unknown number of otherwise eligible subjects. Treatment has been shown to reduce long-term complications, including mortality, liver failure and hepatocellular carcinoma.15–18 Hence, it is important to understand the reasons for non-treatment and address any modifiable factors that may contribute to non-treatment in otherwise eligible HCV-infected subjects. While HIV coinfection has been associated with a lower response rate to treatment for HCV, it is not a contraindication to treatment.

We found that the HCV–HIV-coinfected subjects were treated less frequently for HCV than HCV-monoinfected subjects. Even after adjusting for demographic factors, medical and psychiatric comorbidities and drug and alcohol use, the HCV–HIV-coinfected subjects were less likely to have been prescribed treatment for HCV. The reasons for this are unknown. Comparatively less favourable response to therapy, severity of HIV disease, complicated antiretroviral regimens with potential drug interactions and provider attitudes may all contribute to this lower likelihood of treatment in this group. While HIV-infected subjects have higher rates of psychiatric illness and drug and alcohol use, the low treatment rates cannot be explained by these factors alone.

We found that minority race was associated with a lower likelihood of receiving treatment for HCV. Among the HCV–HIV-coinfected subjects, black subjects were 55% less likely to be prescribed treatment even after adjusting for comorbidities and drug and alcohol use. Minority race has been associated with poorer access to care, less prescription of aggressive interventions and higher mortality for other conditions.19–22 Whether access to care or poorer socioeconomic status is a cause of lower treatment rates for HCV is unknown. Mistrust of authority on behalf of patients and a perception of poorer adherence on part of the providers may enhance the treatment gap, compounded by the lower treatment response rates seen in black subjects in clinical trials.

Anaemia and depression were associated with a lower likelihood of treatment prescription in the multivariable model. Ribavirin therapy is associated with a dose-dependent haemolytic anaemia and interferon therapy may induce new-onset depression or exacerbate pre-existing depression. Hence, these are relative contraindications to treatment. Advanced HIV disease, certain opportunistic infections and some antiretrovirals used for treatment of HIV may also cause anaemia, which may partly explain the higher prevalence and the lower treatment prescription associated with anaemia in the HCV–HIV-coinfected subjects. In our study, subjects with mild depression were less likely to be treated than those with major depression. We believe that this is due to an overlap of diagnosis between the two conditions. An interaction term for the two diagnoses was not significant in our analysis.

Subjects with HIV infection have a high prevalence of medical and psychiatric comorbidities, substance use and a disproportionate number of black subjects are infected with HIV.23, 24 These factors likely play a part in the evaluation of these subjects when treatment for HCV is being contemplated. Smaller studies in tertiary care centres have shown that healthcare providers consider these factors in deciding when not to treat HCV-infected subjects.6 It is likely that these factors are also important in determining suitability for treatment for HCV in the HCV–HIV-coinfected subjects.

The number of new HCV diagnoses in our study population increased between 1999 and 2001. The number of subjects with HIV coinfection also increased from 1999 to 2000, and then declined through 2003. The earlier increase is likely related to the VA campaign to increase awareness and testing for HCV, and the subsequent drop may indicate ‘testing fatigue’ by the providers or the fact that most infected subjects had already been diagnosed. The number of subjects receiving HCV treatment for the first time increased throughout 1999–2003. This is likely attributable to increased awareness and designating HCV a national priority within the VA, as well as increasing comfort of the providers with more data now available regarding efficacy of treatment and associated side-effects, as well as expansion of guidelines to include more groups to be considered for treatment. Availability of pegylated interferons, which need to be injected once a week compared with the older standard interferon, may also have led to the increasing number of infected veteran being treated. Other possibilities are that the subjects most willing to receive treatment presented as soon as the treatment became available, or the patients thought most likely to benefit from treatment were prescribed treatment earlier on.

Certain limitations to this study should be considered when interpreting the results. This is an analysis of administrative databases and diagnoses were extracted using ICD-9 codes. Treatment prescription was determined from pharmacy records and whether those prescribed treatment actually took any of the drugs prescribed is unknown. Some subjects may have received treatment outside a VA facility or while participating in a clinical trial. We did not determine how many of the untreated subjects were actually eligible for treatment based on lack of indications or the presence of contraindications according to the current treatment guidelines. The severity of HIV infection was not determined. Subjects with advanced HIV disease with low CD4 counts are poor candidates for treatment, as interferon therapy is associated with a transient drop in CD4 counts.

There are several strengths to this study. It provides robust estimates of treatment rates and predictors of treatment in a large national population, eliminating the bias of geographically limited or convenience samples. Availability of HCV treatment to all veterans regardless of their ability to pay further strengthens our results by eliminating another bias.

In conclusion, HCV–HIV-coinfected subjects are prescribed HCV treatment less frequently than those with HCV monoinfection. Non-treatment is associated with increasing age, minority race, drug use, anaemia and psychiatric comorbid illnesses. Reasons that affect patients’ and providers’ decisions regarding HCV treatment in the HCV–HIV-coinfected subjects need further study.


This study was funded by National Institutes of Health/National Institute on Drug Abuse (DA016175-01A1, Dr Butt).