Objectives: Hepatitis C, an infection of high prevalence worldwide, is insidiously progressive in many. Reduction of person-to-person spread is possible, and treatment is possible for many, particularly if offered before cirrhosis develops. Screening for hepatitis C (HCV) would be appropriate if strategies could be developed to afford adequate sensitivity and specificity at reasonable cost. We evaluated the performance characteristics of several screening strategies to determine the best balance between cost and performance.
Methods: The database of a national hepatitis screening program was used to define risk factors for HCV. Features associated with increased risk for HCV by multivariable analysis were combined in various ways to construct HCV screening models. Screening Model 1 employed a mathematical model constructed to predict the probability of hepatitis C. Using this model, testing for HCV was done if the probability of HCV was determined to be higher than 7%. Models 2 and 3 called for HCV testing if certain risk factors, stratified as socially intrusive, or nonintrusive in nature, were present. Model 4 calls for testing for HCV only when ALT values are elevated. Costs per case discovered were calculated for each model.
Results: Nine thousand two-hundred sixty-nine individuals from a database of 13,997 has sufficient information to be included in the modeling studies. Risk factors considered socially intrusive were intravenous (i.v.) drug use and sex with an i.v. drug user. Risk factors considered not socially intrusive were: history of blood transfusion, age 30-49 yrs, and male gender. The sensitivity of Models 1-4 were 65%, 69%, 53%, and 63%, respectively. Specificities were 84%, 74%, 77%, and 92%, respectively. The cost per case detected was lowest when Models 1 or 2 were used ($357 and $439, respectively) and higher for models 3 and 4 ($487 and $1047, respectively).
Conclusions: The yield and cost of screening for HCV compares favorably with accepted current screening practices for other diseases. Models 1, 2, and 3 may be appropriate in certain clinical and epidemiological settings. Selective screening by a risk factor questionnaire (first three models) is more cost-effective than blood testing with ALT (Model 4).