Increasingly, efforts are being directed at measuring quality in the delivery of health care. Given the economic burden of providing health care in the U.S., high-volume procedures are being targeted for assessing the quality and benefits of the intervention. Hip and knee arthroplasty are procedures that are ripe for analysis. A common proxy for quality has been volume. However, true quality measures, unlike volume, are exceedingly difficult to measure because of numerous confounding factors. Therefore, risk adjustment becomes paramount so that valid comparisons can be made, thereby leading to accurate conclusions. However, in the rush to measure quality, risk adjustment frequently falls by the wayside. One major problem in performing risk adjustment has been a dearth of information regarding factors that independently predict outcome. This problem is especially true for hip and knee arthroplasty. For this reason, this article by Hooper and his associates that analyzes preoperative medical status and outcomes following hip or knee arthroplasty could not be timelier.
The paradigm shift of paying for performance is in its infancy. To date, initial attempts to implement this strategy have used adherence to process guidelines (e.g., timing of antibiotic prophylaxis) instead of evidence-based performance measures1. Although it is intuitive that baseline medical status and patient comorbidities would impact procedural outcomes, the data to support this hypothesis have been lacking. The study by Hooper et al. provides some data analyzing the relationship between the American Society of Anesthesiologists (ASA) physical status classification system and three end points (mortality, physical function, and implant survivorship). As expected, patients with a higher complexity of comorbidities (i.e., higher ASA class, independent of age and sex) had higher six-month mortality rates. Somewhat surprisingly, a higher ASA class was found to predict lower Oxford scores. Higher ASA class did correlate with a higher implant revision rate for hips but not for knees. These findings would suggest that worse baseline physical status, a non-surgeon-related factor, is not only related to worse outcomes, but may play a greater role in predicting poorer function than has been traditionally believed. It would be valuable to analyze the ASA class relative to change in function over baseline preoperative score, rather than absolute functional score, as this may be a purer measure of the benefit of arthroplasty; however, the authors did not have these data and could not study this end point.
The choice to study ASA class undoubtedly is related to its availability in the New Zealand Joint Registry. ASA class has been criticized for its mediocre validity2 and other comorbidity indexes may be preferable3, especially when functional outcomes are the end point4. Moreover, comorbidity indexes are a compilation of variables, and individual variables instead of a compilation of variables may be preferable for risk adjustment purposes.
As the push for quality measurement and reporting evolves, orthopaedic surgeons need to be engaged in their development and implementation as pay-for-performance initiatives are rolled out in health policy initiatives. The experience with public reporting of cardiac surgery outcome data has shown that there may be unintended consequences such as risk avoidance behaviors by surgeons and decreased access for minority patients5-7. In order to avoid these inadvertent effects, risk adjustment must be performed and is dependent on the identification of independent factors, both surgeon and non-surgeon-related, that are predictive of functional outcome. The study by Hooper et al. represents a small but important step forward in this effort.