Socioeconomic factors and total joint replacement: Comment on the article by Mnatzaganian et al

Abstract

the data are so highly skewed, it would be difficult for a general program such as Review Manager to produce a valid P value. In addition, their combined analysis of GO-BEFORE and GO-FORWARD trial data to evaluate the golimumab treatment effect is problematic given the heterogeneity between the 2 studies, including differences in patient populations and in criteria for inclusion in the MTX control arm.We acknowledge that the statistical methods were not described in detail in the original article because of space limitations, and given this, some of the questions Drs. Graudal and Jurgens pose are understandable. The reason the change from baseline to week 52 in erosion scores and joint space narrowing (JSN) scores do not sum to be similar to the change from baseline to week 52 in total SHS in the GO-FORWARD study is that, for patients who changed treatment or discontinued, linear extrapolation was used to calculate the change in total SHS but not the changes in erosion or JSN scores. Observed data were used for changes in erosion score and JSN score at week 52.Drs. Graudal and Jurgens are correct in pointing out that there is an error in Table 3 of our report, and we appreciate their bringing this to our attention. For the GO-FORWARD study, the values for mean change at week 52 for both erosion score and JSN score in the last 2 columns of the table (i.e., the golimumab 100 mg plus MTX group and the combined groups 3 and 4) were inadvertently transposed. Thus, in the GO-FORWARD study, the mean Ϯ SD change in erosion score from baseline to week 52 should be Ϫ0.04 Ϯ 1.09 in the golimumab 100 mg plus MTX group and 0.14 Ϯ 1.41 in the combined groups 3 and 4; the mean Ϯ SD change in JSN score from baseline to week 52 should be 0.24 Ϯ 0.97 in the golimumab 100 mg plus MTX group and 0.23 Ϯ 0.94 in the combined groups 3 and 4. We have submitted a separate erratum for this correction.In their concluding paragraph, Drs. Graudal and Jurgens state "it is difficult to understand how applying 2 different statistical tests to the same data should yield such disparate results." As discussed above, a key point in understanding the difference is that the same data were not used in their analysis and ours. Our analysis utilized the entire raw data set from the GO-BEFORE and GO-FORWARD studies, while they utilized only the means and standard deviations. Their results are flawed by inappropriate methodology given the highly skewed nature of the radiographic data and the differences between the 2 studies, making their combined analysis problematic.