Exploring the Urgent Need for Diversity in Clinical Data

Redheads gave my anesthesiologist anxiety. Many years ago, during residency I was asked whether I had noticed anything particular about the last three patients on the day’s list. They were all redheads. “Redheads are a problem. I’ll need a bucket-load of anesthetic”, came the response. This may have been my first lesson in cohort-specific outcome diversity.

Anesthesiologists and dentists have long suspected issues around pain sensitivity associated with redheads, who seem to be more tolerant to sedation. With just 2% of the world’s population having red hair, understanding the mechanism has taken time.

Red hair is the phenotype for mutations of the melanocortin-1 receptor (MC1R) gene located on chromosome 16. The downstream effects of the mutation on a brain receptor related to pain sensitivity have now been published.

Importantly, this example is an intra-racial variation. Inter-racial differences may be more profound and more difficult to identify. Accordingly, the call for “diversity” in clinical data collection ought to be distinct from the political or social connotations. Data contributes fundamentally to the validity of observed patient outcomes and is essential to advancing orthopedic care for all members of society.

Notwithstanding what should happen, what is happening is different. For example, the topic of pulse-oxymetry inaccuracy in darker-skinned patients is a very live concern for the FDA. Recent testimony was provided to the FDA’s advisory panel spearheaded by state attorneys general and health equity consumer advocates. While there was a safety communication by the FDA in 2021, understandably in 2024 the discourse is frustrated.

In a broader sense, questions arise as to whether there is fettering of discretion when it comes to patient diversity in devices data. I use those the legalistic terms purposefully, as there is law covering this matter.

The National Institutes of Health Revitalization Act of 1993 aimed to address the historical underrepresentation of women and individuals from diverse racial and ethnic backgrounds in clinical research. While most federal health agencies embraced this policy, FDA did not immediately follow suit. The FDA Safety and Innovation Act then became law in 2012, encouraging greater diversity and subgroup analyses. Since then, FDA has regularly reiterated the importance of subgroup analyses, especially in applications for the highest‐risk medical devices.

At the upcoming AAOS Annual Meeting in San Francisco, an FDA-supported round table on the use of real-world evidence (RWE) will take place. I recently had the opportunity to contribute to the RWE working committee of the Orthopaedic Surgical Manufacturers Association. Clearly, it’s a hot topic.

Grasping the Importance

Diversity of data, especially when examining treatment outcomes, is crucial for several reasons.

• Patient-centered Care. Understanding how treatment options perform in diverse populations aligns with the goal of patient-centered care. Although patient-centered care is incredibly difficult to achieve due to competing realities, available granular data will help healthcare providers make informed decisions that consider individual patient characteristics. Such data has the potential to promote better outcomes for a broader range of patients and arguably optimize the benefit of medical devices in the market.
• Representation of the Population. Different demographic groups may respond differently to treatments. Including diverse data in clinical studies ensures that the findings apply to a broader population. If a registry only includes data from a specific demographic group, the results may not be generalizable to others.
• Understanding Variability. Genetic, environmental and lifestyle factors can influence how individuals respond to treatments. A diverse dataset helps capture this variability, allowing researchers to identify patterns and factors that contribute to treatment success or failure across different populations. In my opinion, herein lies another key variable: cultural differences influencing the interpretation of outcomes.
• Detection of Disparities. Examining data from diverse populations helps to identify potential healthcare disparities. Certain groups that consistently experience different outcomes could indicate issues related to access, quality of care or underlying health disparities that need to be addressed. Perhaps this is where the nocebo effect comes into play when the power of perception is informed by personal influencers.
• Precision Medicine. In the era of precision medicine, tailoring treatments to individual characteristics is essential. A diverse dataset allows for the identification of specific genetic or biomarker patterns associated with treatment response in different subpopulations, enabling more personalized and effective interventions. Enabling technologies and patient-matched implants could be the “precision surgery” equivalent.
• Avoiding Bias and Generalization Errors. Lack of diversity in data may lead to biased results and generalization errors. For instance, a treatment that appears highly effective in a homogeneous dataset may not perform as well when applied to a more diverse population.
• Robustness and Reliability of Results. A diverse dataset contributes to the robustness and reliability of research findings. Results obtained from a varied population are more likely to withstand scrutiny and be applicable across different contexts.
• Regulatory and Ethical Considerations. Regulatory agencies increasingly emphasize the importance of diverse participant representation in clinical trials and registries. Ethical considerations also highlight the need for inclusivity to ensure fair access to the benefits of medical research for all individuals. The latter is of course both complex and highly sensitive. It involves policy and political will to address and untangle.

Following Pharma’s Lead

The following examples from the field of pharmaceuticals highlight the real-world importance of diversity in patient data.

• Warfarin is an anticoagulant used to prevent blood clots. Studies have shown that individuals from different ethnic backgrounds may metabolize warfarin differently. For example, some genetic variants associated with drug metabolism and response can vary among populations. Without diverse patient data, it would be challenging to identify these variations, leading to potential underdosing or overdosing in certain groups.
• Trastuzumab is a targeted therapy used in the treatment of HER2-positive breast cancer. Research has indicated that the response to trastuzumab may differ among different ethnic groups. Without diverse patient data, it would be difficult to identify these variations in treatment response, potentially leading to suboptimal outcomes for certain populations.
• Albuterol is a widely used bronchodilator to treat asthma. Studies have suggested that genetic factors can influence how individuals respond to the drug. Variations in drug metabolism and receptor sensitivity may exist across different ethnic groups. Without diverse patient data, it would be challenging to uncover these genetic factors, potentially impacting the effectiveness of asthma treatment for certain populations.

The effectiveness and safety of the drugs can be influenced by genetic and demographic factors. Without a diverse representation of patients in clinical trials and registries, these factors might go unnoticed, leading to a lack of understanding about how specific drugs perform in different populations.

In terms of orthopedic devices, it is common ground that the success of joint replacement surgeries can be influenced by factors such as bone density, body weight, and activity levels, which may vary among different demographic groups. Interestingly, these biometrics do not directly represent the zeitgeist of published data on arthroplasty and lack of diversity.

Unlike in pharma, there may well be relevant barriers and nuances in medical devices that make enforcement of the 2012 Act a challenge. Perhaps this explains FDA’s apparent workaround: “If a subgroup is known to have a significantly different response than the rest of the population, or if a specific claim is sought in a certain subgroup, additional analyses may be needed.”  Incorporating a “safety valve” to legislative interpretation has value.

The challenge revolves around the condition set forth by FDA — that additional analyses are required if a subgroup is “known” to have a significantly different response. Pending a determination of “significantly different,” it can be problematic or even impossible to establish such knowledge without conducting subgroup analyses in the first place. This presupposes information that may not be readily available.

Subsequently, and unsurprisingly, study authors have reported on how the lack of patient diversity and the absence of publicly available data often hinder clinicians and patients from determining the safety and effectiveness of specific devices for different demographic groups.

A Nuanced Issue

Whatever I write here will not do justice to the subject matter nor will it allow me to explore the nuances in an accessible and fair manner. Nevertheless, and grateful for the prereview of this manuscript and encouragement of colleagues from diverse backgrounds, the discussion must endure and be promoted for the betterment of all our patients.

First, understanding any clinical outcome requires skills akin to the critical appraisal of a research paper. That appraisal is likely to require subject matter expertise to understand the issues in a clinical context and appreciate the relevance and validity of the data in a real-world application. Many experts claim to have such skills, but very few execute them well or certainly don’t do so consistently.

Have you ever wondered how a paper got through the peer-review process in a respected journal? It will become apparent that understanding the word “outcome” is not easy. The concept of outcome is as powerful as it is limited by its context and definitional requirements. It is driven by subjectivity as much as it is by collective perspectives and policy.

In a scoping review of the available literature, multiple narratives touch on diversity. For example, there is evidence of less favorable patient-reported outcome measures (PROMs) for arthroplasty in non-Caucasians. Study authors have noted that these disparities may be racial and ethnic. Even inter-racial inconsistencies have been described as barriers to arthroplasty for some but not others.

Whereas poorer PROMs in one racial group have not been shown to pass the threshold of minimum clinically important difference (MCID), other studies have shown a clear level of dissatisfaction with arthroplasty surgery and even increased complication rates in that same subgroup. Reconciling the published data is appreciably difficult, especially where “dissatisfaction” is of clear significance with certain races but is not seen by others.

Post-op complications, such as increased length of post-op stay, seen with significant frequency are not related to device failure or revision surgery. However, when these “adverse” events are correlated with increased healthcare costs and do not impact a device’s safety or performance profile, they are remarkable and challenging to digest and tackle. Additionally, PROMs, considered fundamental in modern arthroplasty metrics, may not align across racially diverse voices on perceived results of their surgery.

Race and ethnicity aside, the measurements available to us contain limitations in more narrowly defined cohorts. There is plenty of evidence of the effect of gender on PROMs scores.

Outcome variability as described in the contemporaneous growing volume of published literature on medical devices does not necessarily fit with the “redhead” model I opened with. Whereas one uncovers biological determinants of response to an intervention that once understood can be accounted for, the publications on diversity-related outcome discrepancy are seemingly different.

Call to Action

Where does this leave medical device industry professionals engaged with clinical data, regulators who are mandated to receive diversity in those data sets and the process of RWE collection?

RWE has obvious underrepresentation. But what is the effect of underrepresentation on understanding the outcomes of medical devices? Outcomes mean different things to different people. PROMs, with all their limitations with respect to joint replacement surgery, are a surrogate for pain relief and joint function, which includes subjective measures of satisfaction that have become a benchmark.

Implant survivorship, a mixed reality of performance and safety, is a hard outcome to measure. From a surgeon’s perspective, a device that lasts 10 years and provides the expected range of joint movement backed by sublime repeat radiographs will not be seen as a negative outlier in current registry data — even if the patient is unhappy with the outcome of surgery.

So how do you optimize surgery for patient satisfaction when dissatisfaction may not be clinically significant and certainly not consistent across the diversity of your patient population? I have no idea whether FDA or other regulatory agencies have looked at the cohort data from this perspective. They may not need to. I suspect, given the inherent lack of data coming back from the field, that they have introduced flexibility into their expectations.

Rather than any fettering of discretion, I think this position gives industry a pragmatic approach that honors the mandate but doesn’t stifle innovation. The existence of legislation and a continued regulatory emphasis keep the conversation alive and relevant.

This is important, as there are real-world variations in RWE that we do not at present fully understand. As is seemingly become the case in medicine and pharma, that the misunderstood and unexplained becomes evident and categorizable with the correct biomarker and understanding of genetics. I have no idea if one day shotgun genomics will be used to optimize patient-specific outcomes for patient-specific devices. So much is discovered every day.

Until then, we are duty-bound to collect and analyze data. Despite the challenges in collecting relevant and reliable clinical data, I respect the FDA and other regulatory bastions in maintaining the delivery pressure. I hope we will collectively rise to that challenge, and in any event, I know we will keep trying as it matters. Why would we strive for anything less?

Dr. Erturan is the Medical Director – Global Lead Medical Affairs at Orthofix/SeaSpine.

References:

Liem EB, Lin CM, Suleman MI, Doufas AG, Gregg RG, Veauthier JM, Loyd G, Sessler DI. Anesthetic requirement is increased in redheads. Anesthesiology. 2004 Aug;101(2):279-83.

Robinson KC, Kemény LV, Fell GL, Hermann AL, Allouche J, Ding W, Yekkirala A, Hsiao JJ, Su MY, Theodosakis N, Kozak G, Takeuchi Y, Shen S, Berenyi A, Mao J, Woolf CJ, Fisher DE. Reduced MC4R signalling alters nociceptive thresholds associated with red hair. Sci Adv. 2021 Apr 2;7(14):eabd1310.

Fox-Rawlings SR, Gottschalk LB, Doamekpor LA, Zuckerman DM. Diversity in Medical Device Clinical Trials: Do We Know What Works for Which Patients? Milbank Q. 2018 Sep;96(3):499-529.

Singh V, Realyvasquez J, Kugelman DN, Aggarwal VK, Long WJ, Schwarzkopf R. Does racial background influence outcomes following total joint arthroplasty? J Clin Orthop Trauma. 2021 May 21;19:139-146.

Rudisill SS, Varady NH, Birir A, Goodman SM, Parks ML, Amen TB. Racial and Ethnic Disparities in Total Joint Arthroplasty Care: A Contemporary Systematic Review and Meta-Analysis. J Arthroplasty. 2023 Jan;38(1):171-187.e18.

Ling K, Leatherwood W, Fassler R, Burgan J, Komatsu DE, Wang ED. Disparities in postoperative total shoulder arthroplasty outcomes between Black and White patients. JSES Int. 2023 May 26;7(5):842-847.

Patel AP, Gronbeck C, Chambers M, Harrington MA, Halawi MJ. Gender and Total Joint Arthroplasty: Variable Outcomes by Procedure Type. Arthroplast Today. 2020 Jul 22;6(3):517-520.

Vij N, Bingham J, Chen A, Irwin C, Leber C, Schwartz K, Schmidt K. Race and Sex Disparities in Lower Extremity Total Joint Arthroplasty: A Retrospective Database Study. Cureus. 2023 Jul 26;15(7):e42485.

Koettnitz J, Migliorini F, Peterlein CD, Götze C. Same-gender differences in perioperative complications and transfusion management for lower limb arthroplasty. BMC Musculoskelet Disord. 2023 Aug 16;24(1):653.

Goodman SM, Mannstadt I, Gibbons JAB, Rajan M, Bass A, Russell L, Mehta B, Figgie M, Parks ML, Venkatachalam S, Nowell WB, Brantner C, Lui G, Card A, Leung P, Tischler H, Young SR, Navarro-Millán I. Healthcare disparities: patients’ perspectives on barriers to joint replacement. BMC Musculoskelet Disord. 2023 Dec 18;24(1):976.

Burke OC Jr, Gibbons JAB, Do HT, Y Lai E, Bradford L, Bass AR, Amen TB, Russell LA, Mehta B, Parks M, Figgie M, Goodman S. Racial Differences in Patient Satisfaction With the Hospital Experience Undergoing Primary Unilateral Hip and Knee Arthroplasty: A Retrospective Study. Arthroplast Today. 2023 Sep 16;23:101212.

Aggarwal VA, Sohn G, Walton S, Sambandam SN, Wukich DK. Racial variations in complications and costs following total knee arthroplasty: a retrospective matched cohort study. Arch Orthop Trauma Surg. 2024 Jan;144(1):405-416.

Bischof AY, Steinbeck V, Kuklinski D, Marques CJ, Bohlen K, Westphal KC, Lampe F, Geissler A. What is the association between gender and self-perceived health status when controlling for disease-specific conditions? A retrospective data analysis of pre- and post-operative EQ-5D-5L differences in total hip and knee arthroplasty. BMC Musculoskelet Disord. 2023 Nov 27;24(1):914. Save Supplier