We appreciate the comments by Dr. Boers about our article and his interest in our research. Although the original intent for our study of the outcomes following silicone metacarpophalangeal arthroplasty (SMPA) was to perform a randomized controlled trial (RCT), we found this approach was not feasible. Because of strong patient preferences, the likelihood of meeting our enrollment quota in an RCT would have been seriously compromised. Additionally, there is the ethical decision to consider when randomizing patients with rheumatoid arthritis who present with severe deformities but have a high level of function. Whether this type of rheumatoid arthritis patient should have surgery is questionable.
We agree that this cohort should be followed up for a longer period to document the outcomes over time. To facilitate this, we have received additional funding from the National Institute of Arthritis and Musculoskeletal and Skin Diseases and will follow this group for an additional 4 years, bringing our total followup time for this cohort to 7 years. The first 3 years of the study have provided valuable results and we expect that this further followup will do likewise.
Dr. Boers also indicated that it would be interesting to learn the results of surgery on the subgroup of patients who had hand function as good as the nonsurgery group at baseline. First, although the control patients showed higher function on average, for most measures, the ranges of the control patients were wider than the ranges of the SMPA patients. For example, the overall Michigan Hand Outcomes Questionnaire (MHQ) scores at baseline ranged from 3–96 in non-SMPA patients, whereas the scores ranged from 10–84 in SMPA patients after surgery. Similarly, the range for grip strength in the study hands was 0–34.7 in the non-SMPA group and 0–28.3 in SMPA group.
Second, our propensity-stratified estimate was based on a similar statistical approach to such a subgroup analysis in which we estimated between-group 3-year outcomes after trimming patients in nonoverlapping propensity regions. The propensities for receiving the operation were estimated using a logistic regression model with various baseline patient characteristics including baseline values of various outcome measures. Therefore, trimming patients in nonoverlapping regions of propensity essentially excluded patients who were not comparable at baseline. After trimming, this analysis used a smaller subset of 59 patients whose propensities for receiving the operation were comparable. As reported in the article, the propensity-stratified estimate for the 3-year overall summary score was 14.8 points higher in the SMPA patients than in non-SMPA patients, showing that in patients whose propensities for receiving the operation were similar, a significant improvement was expected in the SMPA group compared with the non-SMPA group based on MHQ scores.
Last, Dr. Boers commented on the data in Supplementary Table 1 of our article that compared the change in mean values for outcome variables in the surgical hand and the control hand. This within-person comparison controlled for the baseline difference noted in the surgical and nonsurgical groups to some extent. However, subjects could pick which hand would receive surgery and the study hand was often worse in appearance and function than the control hand. For example, overall mean baseline MHQ scores were 48 for unoperated hands and 37 for operated hands in the SMPA group and 61 for unoperated hands and 56 for operated hands in the non-SMPA group. Although not statistically significant, it is not clear why grip strength increased more in the control group than in the surgical group or, similarly, why grip strength increased more in control hands than in surgical hands in SMPA patients. However, these results highlight the importance of patient-reported outcomes over physical measures such as grip strength in assessing outcomes following surgery.