Kamal and coworkers provide potentially valuable data on the treatment of patients with chronic hepatitis C infected with hepatitis C virus genotype 4.1 However, several confusing points should be clarified before their interpretations are accepted.
- 1The abstract states that 358 patients were randomized; in the text, the number is 378.
- 2The abstract states that 50 patients were randomized into the control group; in the text, the number is 60.
- 3In the abstract and text, the sustained virlogical response (SVR) rate in the controls is said to be 58%, but the numbers provided (30 of 50 patients) is 60%. Furthermore, in detailing analyses of responses by early virological response (EVR), only 27 patients had an SVR in the control group, and the remaining patients are reported to be nonresponders.
- 4In the results, the authors state that patients with a histologic stage of 0 or 1 at the baseline were “616 times more likely to achieve SVR” than those with later stages. Something is wrong here.
- 5This trial demonstrates some of the perils of randomization. All patients were treated in exactly the same manner for the first 12 weeks; yet the EVR rates were different between the two groups: nonresponse occurred in 32% (16 of 50 patients) of the fixed-duration versus 3% (10 of 318 patients) of the variable-duration group. Furthermore, all 10 patients who refused therapy were controls. These significant differences between the two groups were not mentioned or discussed.
- 6The authors claim that the “intention-to-treat principle was applied to all statistical analyses,” but this is clearly not the case. After randomization, 10 controls were excluded because they refused therapy. Furthermore, SVR rates in the variable group was reported as 68%, a proportion which includes only patients who achieved an EVR. Whereas, the calculated SVR rate in the control group included all treated patients, even those with a non-EVR at 12 weeks. A more accurate representation of the data is given in Table 1.
- 7The major conclusion of this paper was that monitoring hepatitis C virus RNA levels and varying the duration of therapy can save resources without sacrificing rates of SVR. The authors used the Blackwelder test to test the noninferiority of the variable-duration approach, stating that equivalence “was defined as SVR percentages within 1%”. This degree of accuracy is impossible in a study of this size. The authors claim that the groups were comparable (P = 0.024), but it is unclear how this P value was obtained or what statistical test was used. Table 1 demonstrates no differences in response rates (P = 0.24) and an overall higher response rate with fixed duration of treatment. However, the study was far too small to demonstrate noninferiority.
|Treatment Group||HCV RNA Negative at 4 Weeks||HCV RNA Positive at 4 but Negative at 12 Weeks||HCV RNA Decreased but Positive at 12 Weeks||Total of the 3 Groups|
|Fixed duration||88% (7/8)||87% (13/15)||64% (7/11)||79% (27/34)|
|Variable duration||86% (59/69)||76% (60/79)||56% (90/160)||68% (209/308)|
In summary, the failure of randomization to achieve comparable groups in this study renders it impossible to support its conclusions. The lack of consistency in presentation of results and statistical analyses and the numerous errors in addition and division do not build confidence in a careful reader.