Algorithm Versus Expert: Machine Learning Versus Surgeon-Predicted Symptom Improvement After Carpal Tunnel Release.

BACKGROUND AND OBJECTIVES: Surgeons rely on clinical experience when making predictions about treatment effects. Incorporating algorithm-based predictions of symptom improvement after carpal tunnel release (CTR) could support medical decision-making. However, these algorithm-based predictions need to outperform predictions made by surgeons to add value. We compared predictions of a validated prediction model for symptom improvement after CTR with predictions made by surgeons. METHODS: This cohort study included 97 patients scheduled for CTR. Preoperatively, surgeons estimated each patient's probability of improvement 6 months after surgery, defined as reaching the minimally clinically important difference on the Boston Carpal Tunnel Syndrome Symptom Severity Score. We assessed model and surgeon performance using calibration (calibration belts), discrimination (area under the curve [AUC]), sensitivity, and specificity. In addition, we assessed the net benefit of decision-making based on the prediction model's estimates vs the surgeon's judgement. RESULTS: The surgeon predictions had poor calibration and suboptimal discrimination (AUC 0.62, 95%-CI 0.49-0.74), while the prediction model showed good calibration and appropriate discrimination (AUC 0.77, 95%-CI 0.66-0.89, P = .05). The accuracy of surgeon predictions was 0.65 (95%-CI 0.37-0.78) vs 0.78 (95%-CI 0.67-0.89) for the prediction model ( P = .03). The sensitivity of surgeon predictions and the prediction model was 0.72 (95%-CI 0.15-0.96) and 0.85 (95%-CI 0.62-0.97), respectively ( P = .04). The specificity of the surgeon predictions was similar to the model's specificity ( P = .25). The net benefit analysis showed better decision-making based on the prediction model compared with the surgeons' decision-making (ie, more correctly predicted improvements and/or fewer incorrectly predicted improvements). CONCLUSION: The prediction model outperformed surgeon predictions of improvement after CTR in terms of calibration, accuracy, and sensitivity. Furthermore, the net benefit analysis indicated that using the prediction model instead of relying solely on surgeon decision-making increases the number of patients who will improve after CTR, without increasing the number of unnecessary surgeries.

Involvement in a Personal Injury Claim Is Associated With More Pain and Delayed Return to Work After Elective Nonsurgical or Surgical Treatment for Hand or Wrist Disorders: A Propensity Score-matched Comparative Study.

BACKGROUND: A small proportion of patients treated for a hand or wrist condition are also involved in a personal injury claim that may or may not be related to the reason for seeking treatment. There are already indications that patients involved in a personal injury claim have more severe symptoms preoperatively and worse surgical outcomes. However, for nonsurgical treatment, it is unknown whether involvement in a personal injury claim affects treatment outcomes. Similarly, it is unknown whether treatment invasiveness affects the association between involvement in a personal injury claim and the outcomes of nonsurgical treatment. Finally, most studies did not take preoperative differences into account. QUESTIONS/PURPOSES: (1) Do patients with a claim have more pain during loading, less function, and longer time to return to work after nonsurgical treatment than matched patients without a personal injury claim? (2) Do patients with a personal injury claim have more pain, less function, and longer time to return to work after minor surgery than matched patients without a personal injury claim? (3) Do patients with a personal injury claim have more pain, less function, and longer time to return to work after major surgery than matched patients without a personal injury claim? METHODS: We used data from a longitudinally maintained database of patients treated for hand or wrist disorders in the Netherlands between December 2012 and May 2020. During the study period, 35,749 patients for whom involvement in a personal injury claim was known were treated nonsurgically or surgically for hand or wrist disorders. All patients were invited to complete the VAS (scores range from 0 to 100) for pain and hand function before treatment and at follow-up. We excluded patients who did not complete the VAS on pain and hand function before treatment and those who received a rare treatment, which we defined as fewer than 20 occurrences in our dataset, resulting in 29,101 patients who were eligible for evaluation in this study. Employed patients (66% [19,134 of 29,101]) were also asked to complete a questionnaire regarding return to work. We distinguished among nonsurgical treatment (follow-up at 3 months), minor surgery (such as trigger finger release, with follow-up of 3 months), and major surgery (such as trapeziectomy, with follow-up at 12 months). The mean age was 53 ± 15 years, 64% (18,695 of 29,101) were women, and 2% (651 of 29,101) of all patients were involved in a personal injury claim. For each outcome and treatment type, patients with a personal injury claim were matched to similar patients without a personal injury claim using 1:2 propensity score matching to account for differences in patient characteristics and baseline pain and hand function. For nonsurgical treatment VAS analysis, there were 115 personal injury claim patients and 230 matched control patients, and for return to work analysis, there were 83 claim and 166 control patients. For minor surgery VAS analysis, there were 172 personal injury claim patients and 344 matched control patients, and for return to work analysis, there were 108 claim and 216 control patients. For major surgery VAS analysis, there were 129 personal injury claim patients and 258 matched control patients, and for return to work analysis, there were 117 claim and 234 control patients. RESULTS: For patients treated nonsurgically, those with a claim had more pain during load at 3 months than matched patients without a personal injury claim (49 ± 30 versus 39 ± 30, adjusted mean difference 9 [95% confidence interval (CI) 2 to 15]; p = 0.008), but there was no difference in hand function (61 ± 27 versus 66 ± 28, adjusted mean difference -5 [95% CI -11 to 1]; p = 0.11). Each week, patients with a personal injury claim had a 39% lower probability of returning to work than patients without a claim (HR 0.61 [95% CI 0.45 to 0.84]; p = 0.002). For patients with an injury claim at 3 months after minor surgery, there was more pain (44 ± 30 versus 34 ± 29, adjusted mean difference 10 [95% CI 5 to 15]; p < 0.001), lower function (60 ± 28 versus 69 ± 28, adjusted mean difference -9 [95% CI -14 to -4]; p = 0.001), and 32% lower probability of returning to work each week (HR 0.68 [95% CI 0.52 to 0.89]; p = 0.005). For patients with an injury claim at 1 year after major surgery, there was more pain (36 ± 29 versus 27 ± 27, adjusted mean difference 9 [95% CI 4 to 15]; p = 0.002), worse hand function (66 ± 28 versus 76 ± 26, adjusted mean difference -9 [95% CI -15 to -4]; p = 0.001), and a 45% lower probability of returning to work each week (HR 0.55 [95% CI 0.42 to 0.73]; p < 0.001). CONCLUSION: Personal injury claim involvement was associated with more posttreatment pain and a longer time to return to work for patients treated for hand or wrist disorders, regardless of treatment invasiveness. Patients with a personal injury claim who underwent surgery also rated their postoperative hand function as worse than similar patients who did not have a claim. Depending on treatment invasiveness, only 42% to 55% of the personal injury claim patients experienced a clinically relevant improvement in pain. We recommend that clinicians extensively discuss the expected treatment outcomes and the low probability of a clinically relevant improvement in pain with their personal injury claim patients and that they broach the possibility of postponing treatment. LEVEL OF EVIDENCE: Level III, therapeutic study.

Predictors of wrist function and health status after operative treatment of fractures of the distal radius.

PURPOSE: To identify the most important determinants of physician-based and patient-based scoring systems for the wrist and upper extremity after operative treatment of a fracture of the distal radius, with the hypothesis that pain is the strongest determinant of both types of scores. METHODS: Eighty-four patients were evaluated a minimum of 6 months after operative fixation of an unstable distal radius fracture using 2 physician-based evaluation instruments (the Mayo Wrist Score and the Gartland and Werley Score) and an upper extremity-specific health status questionnaire (Disabilities of the Arm, Shoulder, and Hand; DASH). Multivariate analysis of variance and multiple linear regression modeling were used to identify the degree to which various factors affect variability in the scores derived with these measures. RESULTS: The physician-based scoring systems showed moderate correlation with each other and with DASH scores. The results of multiple linear regression modeling were as follows (percent variability accounted for by the best fit model/model with top factor alone): Mayo: 54% grip and flexion arc/47% grip alone; Gartland and Werley: 70% pain, flexion arc, radiocarpal arthritis, and duration of follow-up/53% pain alone; DASH: 71% pain, forearm arc, and type of fracture/65% pain alone. CONCLUSIONS: At early follow-up, pain dominates the patient's perception of function after recovery from an operatively treated distal radius fracture as measured by the DASH score and the physician-based rating according to the system of Gartland and Werley. The Mayo Wrist Score is determined primarily by grip strength rather than pain. Because perception of pain and strength of grip have been shown to be influenced by psychosocial factors in some individuals, both patient-based and physician-based measures of wrist function may be vulnerable to illness behavior. TYPE OF STUDY/LEVEL OF EVIDENCE: Prognostic III.