We thank Milam et al. for reading our publication and initiating discussions to promote future research.

First, the authors suggested that inadequate “fidelity” of evidence-based perioperative practice may partially explain persistent Black-White disparities observed in our study, i.e., a definition of evidence-based perioperative practice using an observational dataset that does not fully capture the intricacies of true evidence-based perioperative practice, or as Milam et al. state, “although documented as high-quality evidence-based perioperative practice, the actual care provided may not be clinically adequate to yield the same improvement in outcomes.” While this certainly is a possibility worthy of further investigation, we would like to emphasize what this reasoning would entail in clinical practice. The eight evidence-based perioperative practice components defined in our study were selected based on (1) common components of Enhanced Recovery After Surgery protocols, and (2) what was available in our data source (Premier Healthcare claims data, Charlotte, North Carolina).  These include mostly objective components, such as the yes/no use of multimodal analgesia or tranexamic acid, or yes/no avoidance of urinary catheters, wound drains, or patient-controlled analgesia. However, there might be partial evidence-based perioperative practice compliance with several but not all components applied to different patients, which would result in variation on outcome due to differences in effectiveness of various components. Our study on racial disparity would also benefit from greater granularity in assessment on how certain components are executed, such as details in multimodal analgesia (including variation in combinations of various medications, single shot vs. continuous peripheral nerve block, and so forth). Furthermore, some evidence-based perioperative practice components might be more sensitive to variations in quality, and thus the issue of fidelity may be more relevant for components such as physical therapy, with variations on exact timing of initiation of therapy, intensity and duration, measures of function, among others. While fidelity may play a role in our findings, we would like to emphasize that this would imply a (race- or other subgroup-specific) quality difference in the implementation, a grim assessment of potential mechanisms behind differences in quality of care and outcomes. Future studies are indicated to measure fidelity and study the outcome under various evidence-based perioperative practice combinations in greater detail. A traditional randomized controlled clinical trial would not be efficient to study such complexed system with so many components and variations. Such studies would benefit from machine learning or other artificial intelligence techniques with a much larger database.

We believe that another point mentioned by Milam et al. may be a more likely mechanism behind persistent disparities despite efforts to standardize care: racial heterogeneity in response to evidence-based perioperative practice and interventions. While a causal interaction between race and evidence-based perioperative practice cannot be concluded with our study design, our study did suggest that evidence-based perioperative practice might improve outcomes more among White patients. However, this causal interaction between race and evidence-based perioperative practice cannot be concluded with this analysis. This implies that standardization of care is not enough to reduce disparities; interventions might require subgroup-specific tweaking to result in similar effectiveness across patient subgroups. Findings by Green et al. support such an approach as they reported more severe pain among Black (vs. White) patients despite a standardized order set for postpartum pain management. We agree with the call by Milam et al. for subgroup and sensitivity analyses in effectiveness studies to “…to ensure the majority group is not responsible for improved outcomes with evidence-based perioperative practice…” Important complexities here will be to decide how to balance overall standardization against subgroup-specific personalization, and to avoid overt focus on race for a variety of reasons as highlighted by recent discussions on race in formulas to measure kidney function. 

Milam et al. also highlighted the urgency to understand why Black patients are less likely to receive evidence-based perioperative practice than White patients. This finding is indeed troubling, and we agree that further in-depth research is needed on mechanisms—for example, studies gathering detailed information on hospital-specific protocols and how these are applied in daily clinical practice. In addition, qualitative studies among perioperative care physicians and patients will further highlight sources behind variation in implementation of evidence-based perioperative practice as previous studies have identified patient and/or provider preferences as important sources of perioperative care variation. 

Last, Milam et al. suggested studying the role of socioeconomic status in the context of receipt of evidence-based perioperative practice and how it may influence the associations between race, evidence-based perioperative practice, and perioperative outcomes. We wholeheartedly agree and recommend additional research with data sources that have socioeconomic status data available. However, we do caution against a blanket statement as socioeconomic status is an extremely complex construct with substantial variation in definitions operationalized with often limited availability in datasets generally used for perioperative outcomes studies. We believe that a more important recommendation would be to a priori theorize which socioeconomic status components could impact evidence-based perioperative practice receipt and subsequent outcomes, and through what potential mechanism. For example, education is an important component of socioeconomic status and could influence patients’ comprehension of preoperative education, thus impacting engagement of Enhanced Recovery After Surgery protocols and subsequently influence perioperative outcomes.

The work on disparities in evidence-based perioperative practice and subsequent perioperative outcomes is far from complete, and we thank Milam et al. for their valuable and encouraging discussion.