Our analysis has several limitations. First, our measure of the effectiveness of HCV testing captures only the identification of anti-HCV and patient notification and not the additional costs or benefits that follow, including actual diagnosis of current infection. As recommended by Haddix et al.14
and Gold et al.,15
cost-effectiveness analyses should ideally capture all future costs and health benefits associated with identifying anti-HCV and notifying clients, including the impacts on health-care utilization and quality and length of life. Accounting for these long-term impacts is challenging because there is no consensus on the expected impact of hepatitis C treatment on future morbidity and mortality.16
Our cost estimates are from the STD clinic perspective and do not include other societal costs, such as travel and time costs for clinic clients.
Second, we were unable to limit our analysis of anti-HCV prevalence to include only those people without known anti-HCV positive status because NHANES did not ask respondents whether they have ever been told by a doctor that they have HCV antibodies. To avoid wasting clinic resources, STD clinics should test only those clients without known anti-HCV positivity. To examine the extent to which this limitation affects our cost-effectiveness results, we excluded from our NHANES sample those people who reported that they had ever been told by a doctor that they had a liver condition. This question is likely to identify individuals with known HCV infection as well as those with other forms of liver disease. Approximately 38% of the IDUs who tested positive for anti-HCV responded “yes” to this question. Once those individuals were eliminated from our analysis, the estimated prevalence of anti-HCV among IDUs was 0.47, and the estimated cost per positive tester who returns was $64.60, as compared to the baseline estimate of $54.40.
For the other subpopulations, the impact of eliminating individuals with a known liver condition reduced anti-HCV prevalence by a small amount that did not substantially alter the cost-effectiveness of testing. For example, for men with more than 100 sex partners, eliminating those who reported a previous liver condition increased the cost-effectiveness from $179 to $186.
In summary, while we obviously do not recommend retesting patients who are aware of their positive anti-HCV status, removing these patients from the analysis is not likely to alter the decision to prioritize testing of those who disclose IDU behaviors but have not been tested before. We were unable to estimate the cost-effectiveness of retesting IDUs who have previously tested negative for anti-HCV because such a question was beyond the scope of our model. However, an evaluation of the cost-effectiveness of HIV screening found that recurrent screening was always less cost-effective than one-time screening but that it became more cost-effective as the annual disease incidence increased.17
Future studies should examine the impact of annual incidence on the cost-effectiveness of recurrent anti-HCV testing.
Third, our estimates of anti-HCV prevalence in each subgroup are from NHANES, which has several limitations that may affect how well the data represent similarly defined subgroups in STD clinics. NHANES excludes groups at especially high risk for HCV infection (e.g., homeless, incarcerated). Additionally, NHANES may understate the prevalence of anti-HCV in STD clinic clients if those clients are more likely than the general population to engage in behaviors that put them at risk for HCV infection. For NHANES respondents with more than 100 sex partners or who report injection drug use, the anti-HCV prevalence estimate may be representative of STD clinic clients who disclose those same behaviors. NHANES is most likely to understate anti-HCV prevalence for STD clinic clients who do not disclose risk behaviors, especially if those clients have riskier lifestyles than the general population.
Fourth, we did not consider females older than 40 with more than 100 sex partners for testing because the NHANES sample of this group was too small to estimate the prevalence of anti-HCV. Because sentinel surveillance information indicates that sexual transmission may be more common among women than men,18
clinics should use available information about risk factors for HCV infection to estimate anti-HCV prevalence among their female clients and develop testing policies accordingly.
Fifth, our estimates assume that the probability of returning for test results is the same for all subgroups. However, it is more realistic to expect return rates to vary by risk group. For example, because they know they are at increased risk for HCV infection, IDUs are generally more likely to return for test results than other targeted subgroups, suggesting that CTR may be more cost-effective for IDUs than our estimates suggest.
Sixth, our sensitivity analyses are limited in that they do not consider the impact of systematically examining the possible range of all values used in the cost-effectiveness model. For example, we did not examine the impact of variations in the sensitivity and specificity of the testing protocol or return rates for test results. Future research will incorporate additional sensitivity analyses, including probabilistic sensitivity analyses to examine which combinations of variables in the model have the greatest impact on cost-effectiveness findings.
Finally, our cost and return rate estimates are based on data from a small number of STD clinics, which may not be representative, and our cost-effectiveness estimates do not capture the challenges that STD clinics are likely to face in limiting testing to specific high-risk subgroups.