Background: The impact of systematic reviews depends on their usefulness in guiding practice. A focus on surrogate outcomes substantially weakens the applicability of evidence because it does not permit a tradeoff between patient-important desirable and undesirable consequences of alternative interventions. Objectives: To estimate the treatment effect on development of cirrhosis in a systematic review of combination therapy of lamivudine plus interferon-α (ie, experimental) versus lamivudine (ie, control) for chronic hepatitis B (CHB), in which only surrogates were measured. Methods: We pooled trials reporting HBeAg seroconversion, and generated pooled HBeAg seroconversion rates in experimental and control groups. Cohort studies provided estimates of the risk of developing cirrhosis in CHB patients with and without HBeAg seroconversion. Assuming similarity of patient characteristics in cohort studies and trials, we applied estimates of cirrhosis rate from cohort studies to estimates of proportion of patients with and without HBeAg seroconversion from the meta-analysis to estimate the risks of developing cirrhosis in experimental and control groups. We calculated the absolute risk reduction (ARR) based on the estimated risks. We incorporated variances from the trial and cohort study data to generate 95% confidence intervals of ARR. Results: Pooling estimates from 5 trials involving 1095 patients suggested that combination therapy had a higher HBeAg seroconversion rate than lamivudine (31.8% versus 18.2%; RR 1.76, 95% CI 1.29 to 2.41; ARR 13.6%, 95% CI 5.9% to 21.2%). Modeling suggested a small and uncertain absolute difference in the development of cirrhosis between combination therapy and lamivudine alone (ARR 0.90%, 95% CI −0.15% to 3.37%). Conclusions: Our results suggest that combination therapy, relative to lamivudine, improves HBeAg seroconversion. However, its effect on the development of cirrhosis, if it exists at all, is small. The modeling provides a novel approach that may provide important insights that could influence the advisability of administering toxic and expensive therapy.