Author’s Reply to Letter to the Editor: Medicaid Payer Status Is a Predictor of Early Postoperative Pain Following Upper Extremity Procedures

Although the methodology in any scientific paper can always be improved, the correlation and statistical significance identified between patients undergoing upper extremity surgery, their insurance status, and their pain are likely real. Although I agree that we would ideally control for type of procedure, disease severity, time to presentation, lower socioeconomic status, and psychologically associated diagnoses, this is near impossible in Medicaid patients. In essence, this was my point in writing the study. Medicaid to me is much more than just an insurance diagnosis as it often is correlated with that already mentioned, such as delay in treatment/diagnosis, issues with access to care, and mental health disorders.^1,2 In fact, I have previously brought this to light in the spine literature where we called 234 spine practices and not one offered an appointment to Medicaid patients. ² The insurance status of Medicaid varies by state and often is based on income and family size alone. It is therefore exceedingly difficult to tease out differences in lower socioeconomic populations and those in Medicaid patients as the two are linked. I am going to expand on my rationale as to why the study was conducted the way it was published.

First, as the readers have mentioned, ideally it would have been possible to have controlled for disease severity. However, we chose upper extremity procedures whereby there is little correlation between severity and clinical relevance (eg, carpal tunnel, carpometacarpal arthritis, cubital tunnel). Recently as part of a Quality Collaborative Initiative on carpal tunnel, my coauthors and I asked the question of whether clinical disease severity correlated with electrodiagnostic study severity, and it did not. ³ In short, disease severity as classified using radiographs or objective measures rarely correlates with true clinical findings and therefore is not a true measure of severity. As such, we did not feel as though this would have changed the results because pain is so subjective and not necessarily correlated to severity in many of the pathologies studied here. Similarly, time from onset of disease/symptoms to presentation is difficult to measure for several of the pathologies studied, including thumb arthritis and cubital tunnel, as these pathologies often go underdiagnosed or are identified through coincidence (eg, thumb arthritis identified in patients at the time of distal radius fractures). ⁴ Although studying a single disease pathology would remove confounding data, it would also significantly limit power enough that the study would not be feasible. As such, I agree the data should be interpreted with caution, but would also point out that the study would have taken substantially longer to accomplish if patients were excluded by diagnosis and categorized by severity.

In addition to the above, the readers inferred that “racial and ethnic differences by insurance class may contribute to their findings.” It is actually the contrary, which is that there are clear racial and ethnic differences within insurance classes (eg, access to medical care may be different by race). The advent of Medicaid has actually helped, in some instances, reduce the differences as evidenced in the study by Winkelman et al. ⁵ However, my coauthors and I have recently looked at trends within the last decade, and there are still substantial differences that exist, at least within shoulder arthroplasty. ⁶ The current paper clearly was not meant to look at racial or ethnicity disparities, nor how it relates to pain. The statement within the paper merely just states that it is possible that there could be other confounding variables.

The readers also quote Seth Leopold in which he states that Medicaid alone cannot cause surgical complications. This essentially embodies my paper. In my opinion, using insurance status as a predictor in any paper is using all of the inherent conditions that correlate with that insurance. In the case of Medicaid, it is all that is already said earlier: access to care, poor coping skills, affiliated psychological diagnoses, and lower socioeconomic status. Whether we like it or not, these issues negatively impact our ability to treat patients and for them to gain a similar outcome as those without these issues. The body of literature regarding this specific topic is vast, and when writing this reply, I encountered hundreds of articles, but these few will prove my point.^{1,7 -12}

Finally, as one of the few providers that routinely provides care to Medicaid patients, I am acutely aware that these patients have little avenues for care. The purpose of this paper was not to dissuade providers or health care institutions from providing care, but rather to show that despite one’s best efforts to treat Medicaid patients well, the results are still not as good as we would have hoped.