Objectives: From the literature on the hepatitis C virus, the existence of a gap between a sustained virologic response (SVR) attainable in randomized clinical trials (RCTs) versus routine practice is not clear. Further, in terms of the pretreatment prediction of SVR, to date, studies have focused only on reporting the magnitude of association (MOA) between each predictor and an SVR. They fail to acknowledge that a predictor with a large MOA is of little value to clinicians if it has low variability in the treatment population.
Methods: Hepatitis C virus clinical databases were used to derive a large, representative cohort of Scottish pegylated interferon and ribavirin initiates.
Results: Overall, 39% [123/315, 95% confidence interval (CI) 34-45%] of genotype 1 and 70% (414/594, 95% CI 66-73%) of genotype 2/3 patients achieved an SVR; this compares with the pooled estimates of 47% for genotype 1 (95% CI 41-52%) and 80% for genotype 2/3 (95% CI 75-85%) RCT participants. Significant predictors of SVR identified from logistic regression were ranked on the basis of the akaike information criteria (reflecting an approach that will account for each predictor's MOA and variability) as follows: (i) genotype, % increase in akaike information criteria of the final model when variables are excluded, 58.49%; (ii) γ-glutamyl transferase, 18.64%; (iii) platelet count, 6.48%; (iv) alanine aminotransferase quotient, 4.63%; (v) ever infected with hepatitis B virus, 4.31% and (vi) sex, 3.10%.
Conclusion: (i) The proportion of patients attaining an SVR in Scottish routine practice is marginally lower than in RCTs and (ii) other than genotype, γ-glutamyl transferase emerges as a valuable predictor of an SVR in routine practice. Further, we demonstrate an approach to more clearly discern the predictive value of response predictors.