Valid and Relevant Outcome Measures Are Critical for Objective Hypothesis-Testing

Abstract

To the Editor: Regarding Hofmann-Kiefer et al.'s article,1 we question the appropriateness of using the Constant-Murley Assessment score as the investigation’s primary endpoint. To test the study hypothesis, the Constant score was assessed the day before surgery and then 96 h postoperatively.1 However, the Constant score was originally designed “to assess function after injury”— not surgery2; and it has not been validated for patients in the immediate postoperative period.3 Among articles using the Constant score, none involved patients within 4 wk of surgery: five, including the initial report, included healthy volunteers and/or nonsurgical clinic patients3–7; another included postrotator cuff repair patients at least 1 mo postoperatively8; and the last studied preoperative patients.9 The authors of the current study indirectly acknowledge this lack of validation in the immediate postoperative period when they explain, “because most of our (postoperative) patients were not completely pain-free, even at rest, one item of the Constant score had to be adopted for our purposes [emphasis added].”1 Moreover, it is doubtful that the Constant score is even relevant to the hospitalized population studied by Hofmann-Kiefer and colleagues. Among the 100 possible points comprising the Constant score, maximum active shoulder abduction (10%), flexion (10%), external rotation (10%), strength (25%), and limb positioning (10%) comprise 65% of the total.3 Although there are no universally accepted post-open acromioplasty and rotator cuff repair rehabilitation guidelines, most patients receive passive range-of-motion within safe and pain-free ranges to prevent a loss of shoulder mobility, and progression to active exercises occurs only after “1–3 wk, but much more gradually after an open procedure.”10–12 This is to avoid injuring the supraspinatus and/or deltoid repairs. And, predictably, the authors found that “only 5.9% of the patients in the PCA group and 13.5% in the PCISB group were able to perform the strength test.”1 The majority of active range-of-motion and strength tests are simply inapplicable and irrelevant for this patient population in the immediate postoperative period. Eight additional Constant score points are determined by “activity level” during “work” and “recreation/sport,” which is, again, irrelevant in these hospitalized patients.3 Given that at least 73% of the Constant score is inapplicable or irrelevant to this hospitalized patient population, it is apparent why this measure has never been validated in the immediate postoperative period. We thus conclude that the Constant score was an inappropriate primary endpoint and that the study’s hypothesis was therefore inadequately tested. Drawing conclusions from the study results is a subjective exercise, and not an objective analysis. It would be unfortunate and counterproductive if the medical community assumed the Constant score is a valid, reproducible, and relevant assessment tool in the immediate postoperative period. We welcome a thoughtful discussion of this topic with our valued colleagues. Brian M. Ilfeld, MD, MS Department of Anesthesiology University of California San Diego San Diego, California Thomas W. Wright, MD Department of Orthopedics University of Florida Gainesville, Florida Daniel I. Sessler, MD Department of Outcomes Research The Cleveland Clinic Cleveland, Ohio Terese L. Chmielewski, PhD, PT Department of Physical Therapy University of Florida Gainesville, Florida