[1] Unité d'Hépatogastroentérologie, Centre hospitalier, 64046 Pau Cedex

[2] Département d'Information Médicale, Centre hospitalier, 64046 Pau Cedex

[3] Unité d'Hépatologie, Hôpital Laënnec, 60109 Creil.

Patient characteristics (table I) were comparable to those in the Poynard et al. trial [3] . Patient gender, mode of contamination, and genotype were similar. Our patient population was older and had a lower body weight, but viral load was elevated less often and mean histology score showed lower activity; conversely fibrosis was more pronounced and cirrhosis was more frequent. Ninety-three of our patients had a comorbid condition (table II) which was an exclusion factor in the Poynard et el. trial. In addition, 27 patients had hematological diseases (anemia, n = 7; neutropenia, n = 1; thrombopenia, n = 19) which would have excluded them from the Poynard et al. trial [3] .

Considering the entire population, the rate of sustained virological response was 6% lower in our Observatory than in the Poynard et al. trial (table III) . This difference was not significant (95%CI: ? 18% to + 6%). Likewise the difference in sustained virological response between the subgroups designated by genotype, viral load, gender, and fibrosis score (table IV) was not significant.

Moreover, sustained virological response by age did not show any difference for the age groups (= 40 yrs versus > 40 years) used in the Poynard et al. study (34% versus 39%, NS). When selecting a median age of 46 years for our patients, the sustained virological response was not significvantly higher in young patients (40% vs 33%). Expressed by response factors (genotype 2 or 3, low viremia, age < 41 years, fibrosis score < F3, female gender), (table V) the results we obtained were equivalent to those reported by Poynard et al. Response rate in patients with comorbid conditions (table VI) was significantly lower (25/93, 27%) than in those free of such conditions (71/116, 43%; p = 0.02). Similarly, drug doses had to be reduced more often in patients with comorbidity (35/93, 38% versus 30/116, 18%; p = 0.0005) with no impact on duration of treatment. Advanced fibrosis was also more frequent (50/92, 54% versus 46/165, 28%; p < 0.001).

At logistic regression analysis, factors found to be significant and independent variables predictive of sustained response were, in decreasing order: genotype 2 or 3 (p < 0.001), low fibrosis score (p = 0.01), and absence of comorbidity (p = 0.026). The following factors were not found to be significant independent predictive factors: young age (p = 0.53), female gender (p = 0.30), low viral load (p = 0.3), and need to lower drug doses (p = 0.17). The model established with the three significant factors explained 11% of the variability.

Mean decrease in hemoglobinemia was 2.9 ± 1.4 g/dl and 22 patients (8%) had a hemoglobinemia < 10 g/dl. Neutropenia (< 1000/µl) was observed in 36 patients (14.5%) including 16 (6%) with a white cell count below 750/µl. Thrombopenia below 100,000/µl was noted in 26 patients (10) and below 50,000/µl in 4 (1.5%). Drug doses were reduced in 65 patients (25%): interferon in 20 (7.5 %) and ribavirin in 60 (23%).

Treatment was discontinued before the planned term in 100 patients (38%) for inefficacy at the 6^th month (n = 41), in patients lost to follow-up (n = 14), and for adverse effects (n = 45), the most frequent being depression in 10 (4%) and weight loss in 8 (3%). Undesirable effects were noted more often in patients with comorbid conditions (91% versus 68%, p < 0.001). There was one death from an accidental cause during the study period.

Our data were obtained in a perfectly independent manner. The study was conducted without material support (which was not solicited) or outside influence concerning study design, implementation, interpretation, or reporting and was not supervised by a clinical research physician working in the framework of an investigating unit. The investigators were entirely responsible for data collection and quality of data. This setup inevitably led to missing data since the study was conducted under the conditions of routine practice free of any mandatory prescriptions for complementary examinations (with the exception of inclusion criteria and the main outcome criterion). The number of missing data was nevertheless relatively low reaching only 2% for mode of contamination, 8% for genotype and liver biopsy, and 16% for viral load. The influence of missing data was thus minimal.

We compared our findings with those obtained in the best arm of the Poynard et al. study [3] (alpha-2b interferon 3MU three times a week plus ribavirin 1000-1200 mg/d depending on body weight for 48 weeks) for the following reasons: 1) the Poynard et al. study basically included European patients generally comparable with our patients for contamination mode and body weight;2) that study also involved the analysis of detailed subgroups;3) no difference in treatment efficacy was demonstrated between six and twelve weeks for patients with genotype 2 and 3 and results showed that treating beyond the 6^th week is not indicated in patients with genotype 1 who are non-responders at six weeks, the approach adopted by the Observatory investigators.

Globally, the results of our study confirm the efficacy of the bitherapy regimen in patients with chronic hepatitis C. The rate of sustained response was 6% lower (95%CI: ? 18% to + 6%) but the difference was not significant. The intrinsic efficacy was undoubtedly the same, the difference probably resulting from more severe disease in ?real life? populations. This hypothesis is supported by several observations: a) our patients were older, had more severe fibrosis, and 17% had cirrhosis;b) comparing results by response factors as described by Poynard et al. [3] shows that the rates of response were nearly identical;c) patients would have been excluded from the Poynard et al. study because of comorbidity had a lower overall rate of sustained response. Logistic regression analysis showed that presence of comorbid conditions as an independent negative factor predictive of sustained response. In this situation, adverse effects and dose reductions were frequent.

The adverse hematological effects observed in our population were similar in intensity and frequency. Treatment was discontinued because of adverse effects in more patients in our population (21%) than in the Poynard et al. population (8% after six months treatment and 19% after 12 months).

In conclusion, the rates of sustained response obtained directly from the clinical trials conducted for marketing approval purposes probably overestimated the efficacy of the alpha-2 inteferon plus ribavirin combination for the treatment of chronic hepatitis C. This overestimation, which could be a source of deception and loss of confidence for patients, probably resulted from the fact that patients included in the therapeutic trials had less severe disease. The results of treatment must be individualized based on the criteria of response used by Poynard et al. [3] . Moreover, public health decision-makers should interpret the rate of sustained response calculated from the therapeutic trials with caution, particularly when developing medico-economic models. Independent structures such as our Observatory should be supported by public funds provided in an unconditional non-bureaucratic manner in order to pursue useful evaluations of treatment efficacy after marketing approval.