Factors Associated With Outcome Response Frequency in a Knee Replacement Registry

Nonresponse can potentially introduce bias in arthroplasty registries, compromising confidence in outcome data, diminishing both internal validity and generalizability. Patient-reported outcome measures can be critical tools in evaluating clinical outcomes after total knee arthroplasty; however, patient-reported outcome measure data can be skewed when subsets of the population are nonresponsive. This study investigates sociodemographic and clinical factors associated with 12-month nonresponse after total knee arthroplasty and the effects of comprehensive multimodal follow-up methods.

A prospective cohort of 2,508 total knee arthroplasty patients enrolled in the Function and Outcomes Research for Comparative Effectiveness registry between 2018 and 2023 was analyzed. Sociodemographic and clinical data were collected preoperatively, and comprehensive multimodal follow-up methods were implemented. Hierarchical cluster analysis identified characteristics associated with nonresponse, and logistic regression was used to validate these findings.

At 12-month follow-up, 735 of 2,508 patients (29%; P < 0.0001) were nonresponsive. Nonresponders, represented by cluster 5, which had a 45.8% response rate, were more likely to be female (P < 0.0001), non-White or mixed race (P < 0.0001), Hispanic or Latino (P < 0.0001), have less than a college education (P < 0.0001), public insurance (P < 0.0001), greater comorbidity (P < 0.0001), and lower preoperative knee injury and osteoarthritis outcome scores (P < 0.0001). The highest response rate (76.9%) was found in cluster 1, which primarily included well-educated males (P < 0.0001), with private insurance (P < 0.0001), and a lower body mass index (P < 0.0001).

(1) Persistent and multimodal follow-up methods, through e-mail, paper mailings, and phone calls are needed to achieve high response rates above international registry standards of 60%. (2) Identifying patient characteristics linked to nonresponse provides an opportunity to help with targeted response strategies. These strategies may help reduce selection bias, improve data collection through improved response rates, and enhance the long-term utility of arthroplasty registries.