Before the First Opioid: Genetic Risk Assessment and the Orthopaedic Patient

A patient recently presented for knee arthroplasty with a family history that made our routine discussion about postoperative pain anything but routine. Years earlier, her child sustained a major fracture that required operative fixation. Opioids were prescribed for postoperative pain, which is reasonable and common follow such an injury. Unfortunately, that initial exposure was followed by opioid dependence, and her child eventually fell into an inner-city opioid scene where the story ended with a fatal overdose. Now, the mother needed an operation of her own.

By the time I met her, I had started using AvertD, an FDA-approved genetic test that identifies patients who may have an elevated genetic risk for developing opioid use disorder (OUD). Of course, her result cannot tell us whether her child carried the same genetic risk, and it cannot establish that a different prescription after the fracture surgery would have changed what eventually happened. However, the genetic result gave us additional information before her own knee replacement, and we used it to construct a multimodal pain strategy that avoided opioids throughout her perioperative course.

My interest in AvertD arose organically. Minimizing opioid exposure and enhancing early recovery has been a major focus throughout my entire adult reconstruction practice. I was writing about multimodal perioperative pain management during my arthroplasty fellowship, and I more recently co-authored a real-world evidence protocol that explicitly collects information on acute pain control (Pepper et al. 2018; Mercuri and Wickline 2026). For roughly the past eight years, I have been asking a fairly simple question in my practice: How little opioid does a patient actually need after a hip or knee replacement?

Since I was a resident, orthopaedic surgeons have become much more sophisticated about multimodal pain control that decreases opioid consumption. However, before deciding whether a patient needs opioids, it is reasonable to ask how much we actually know about the patient who is filling the prescription. Brummett and colleagues reported new persistent opioid use in approximately 6% of previously opioid-naive adults after minor and major surgical procedures, compared with 0.4% in a nonsurgical comparison group (Brummett et al. 2017). Of course, persistent opioid use is not the same thing as OUD. Still, an acute medication intended to treat temporary surgical pain sometimes becomes a chronic dependency long after the procedure. For me, that is enough reason to care about the quality of the risk assessment that occurs before the first exposure.

There is an interesting imbalance in the way orthopedists have historically approached this problem. We spend considerable effort deciding whether a patient should receive 10 opioid tablets instead of 30, whether a regional block will reduce morphine equivalents, or whether a refill request on Friday afternoon is appropriate. Yet, the assessment we perform before the initial prescription may amount to only a few questions buried in a rushed outpatient visit. Current FDA labeling instructs prescribers to assess a patient’s risk for addiction, abuse, and misuse before prescribing opioids, as well as reassessment of that risk during treatment (US Food and Drug Administration 2023a). Historically, my team and I asked about prior opioid use and substance-use history, reviewed the electronic medical record (EMR), and queried the prescription drug monitoring program (PDMP).

Certainly, all of this is useful, but none of it is perfect. A patient may not remember a short prescription from years earlier, a medication list only reflects what was captured within our EMR, and the PDMP records what was prescribed but not what was actually consumed. Even a simple term such as “opioid-naive” becomes complex when one moves from a research protocol into a busy orthopaedic office. In a clinical trial, prior exposure is a variable that can be controlled. In an actual patient, establishing that exposure becomes extremely difficult. This matters because a test like AvertD inherits both the strengths and weaknesses of its design and approval process.

AvertD is a prescription genotyping test that evaluates 15 germline single-nucleotide polymorphisms. It then returns a qualitative result of either Elevated Genetic Risk or Non-Elevated Genetic Risk for OUD. The FDA approved the test through the premarket approval pathway in December 2023 for adults before a first prescription of oral opioids for acute pain (US Food and Drug Administration 2023b, 2023c). In April 2026, the FDA approved venous whole blood as an additional sample type (US Food and Drug Administration 2026b). In my practice, I discuss and order AvertD when the surgery is first indicated. Blood is collected with routine preadmission laboratory testing, and my trained PA-Cs review the result with the patient at the pre-operative history-and-physical visit. The result is therefore available before the final pain management plan is made.

The fact that AvertD measures germline variants is an important concept. Genetics are fixed at conception, and a prior Percocet prescription does not change a patient’s genotype. However, a classifying test like AvertD cannot be applied to every clinical population simply because the underlying genotype is fixed. Prior opioid exposure changes the clinical context, and it may change pretest probability. In other words, there are two different issues at play. What the test measures biologically is not the same as the population in which the test was performed. As a result, the most defensible place for orthopaedic adoption is the opiate-naive population for which FDA has approved the test.

The underlying evidence supports both the advocates and critics of AvertD. Donaldson and colleagues published the classifier validation in 2021 (Donaldson et al. 2021). The FDA Summary of Safety and Effectiveness Data provides a full regulatory summary (US Food and Drug Administration 2023c). A total of 812 subjects were enrolled, 385 were selected for analysis, and four samples were invalid, thus leaving 381 evaluable subjects. However, these subjects were not enrolled before an initial opioid prescription and followed prospectively. Instead, their 4- to 30-day index opioid exposure was based on self-reported consumption, with available medical records later reviewed for corroboration. OUD status was also assessed at enrollment using a DSM-5 clinical evaluation. In the final FDA analysis, AvertD demonstrated 82.8% sensitivity and 79.2% specificity, with a positive likelihood ratio of 3.98 and a negative likelihood ratio of 0.22 (US Food and Drug Administration 2023c). Donaldson reported nearly identical sensitivity and specificity (Donaldson et al. 2021). Therefore, Donaldson and the FDA analysis are essentially two descriptions of the same validation evidence.

AvertD’s operating characteristics are strong, but they still do not mean that a positive test predicts addiction. Predictive value depends heavily on the baseline prevalence of OUD, which the current provider education material emphasizes (SOLVD Health 2026). The study population also had demographic limitations. More than 92% of the final analysis group identified as White, with only 14 Black participants and two Asian/Pacific Islander participants (US Food and Drug Administration 2023c). Advocates for AvertD should not hide these limitations because the argument for the test is actually stronger when its limitations are acknowledged. The relevant question is not whether AvertD is a perfect predictor. The relevant question is whether the information adds something useful to the imperfect risk assessments that surgeons are already performing on a daily basis.

So, what should an orthopaedic surgeon actually do with the result? First, genetic predisposition is only one component of OUD risk. AvertD does not measure all of the variables that may influence behavior after opioid exposure such as psychiatric history, social environment, and personal substance-use history. An Elevated Genetic Risk result does not mean that a patient will develop OUD, and a Non-Elevated Genetic Risk result does not mean that a patient can consume opioids with reckless abandon. Similarly, hemoglobin A1c does not tell surgeons which patient will develop a periprosthetic joint infection, just as bone density does not identify who will sustain a fracture. Either way, surgeons still want to know these things because they change risk assessment and they change clinical practice.

In my early experience with AvertD, an elevated result has not created a mandatory pathway. For some patients, I pursue an opioid-free multimodal regimen. For others, a small prescription for severe breakthrough pain may be reasonable. In some, more substantial opioid exposure may be necessary. The point is not that the genetic result dictates one treatment. The point is that it changes my counseling, monitoring, and shared decision-making about opioids. Similarly, a non-elevated result changes nothing about my commitment to opioid stewardship. I do not prescribe more oxycodone because a patient is classified as Non-Elevated Risk. No single test can make these decisions for me, and the AvertD results are yet another piece of information that I use when designing a pain management plan.

Of course, there are scientific arguments against screening for genetic risk. In October 2022, an FDA advisory committee evaluating an earlier version of the test voted 2 to 11 that the benefits did not outweigh the risks (US Food and Drug Administration 2022). The panel raised concerns about false-positive and false-negative results, potential undertreatment of pain, limitations in the clinical evidence, and generalizability across populations. However, that vote was not the end of the regulatory process. The FDA subsequently worked with AutoGenomics as the test was modified, and the agency approved AvertD in December 2023 with safeguards that included mandatory provider training, patient and provider education, warnings against using the result alone, and a required post-approval study (US Food and Drug Administration 2023c). In short, the FDA ultimately approved the modified test after it was paired with specific mitigations and additional evidence generation.

The most substantial scientific challenge comes from Davis and colleagues who published an analysis in JAMA Network Open in 2025 using data from 452,664 opioid-exposed participants in the Million Veteran Program (Davis et al. 2025). They reported that the 15 variants used by AvertD collectively accounted for 0.40% of the variation in OUD case-control status, while an ensemble machine-learning model constructed from those variants classified case-control status with 52.83% accuracy in an independent testing sample (Davis et al. 2025). I take those findings seriously because they challenge the predictive value of the constituent variants.

At the same time, it is incorrect to describe the Davis study as though investigators simply tested 452,664 veterans with AvertD and found that the device performed like a coin flip. They did not have the proprietary AvertD test. Donaldson’s classifier used the 15 variants jointly in a blended machine-learning model, while Davis reported a regression-based variance estimate and constructed its own ensemble model because the proprietary classifier was unavailable (Donaldson et al. 2021; Davis et al. 2025). The Davis cohort was already opioid-exposed, had a mean age of approximately 61 years, and was more than 90% male. OUD was identified through diagnostic codes, and the investigators also examined patients with documented short-term opioid exposure. Finally, seven variants were imputed rather than directly genotyped which introduces ancestry-cofounding as a variable (Davis et al. 2025).

The Davis paper should not be dismissed because of these limitations. It leaves AvertD with some important questions to answer, but it does not tell surgeons how the approved test actually performs when used prospectively in the population for which it was approved. In fact, the FDA wants to see that type of performance data, so it required a prospective post-approval study of at least 3,000 patients tested before a first oral opioid prescription, with assessment of real-world device performance, demographic subgroups, prescribing behavior, and patient and provider comprehension (US Food and Drug Administration 2023c). As of September 2026, the study is listed as delayed, although the FDA accepted a revised protocol in May 2026 (US Food and Drug Administration 2026a). I also want to see those data, and I think the orthopaedic community should push for them.

However, the need for more data does not answer the clinical question in front of me when speaking to patients. I still have to decide how to manage pain and whether to expose that patient to an opioid. If an elevated result causes a surgeon to label a patient as a future addict and withhold adequate analgesia, then the test has been misused. Similarly, false-positive results are not harmless. Anxiety, stigma, and undertreatment of pain are legitimate concerns. However, those concerns do not justify ignoring potentially useful information when correctly interpreted. A high-risk patient should receive a more detailed conversation, greater use of multimodal agents, smaller quantities of breakthrough medication, closer observation, and potentially opioid avoidance.

For these reasons, I believe that AvertD should become a routine component of preoperative OUD risk assessment for appropriately selected orthopaedic patients. Surgeons still need to ask about prior substance use, family history, psychiatric disease, medications, and social circumstances. We still need the PDMP and the medical record. We still need to understand the operation, the expected pain, and the alternatives available to that particular patient. AvertD relieves surgeons of none of these responsibilities. It simply gives us one additional piece of patient-specific information while there is still time to change the treatment plan.

Orthopaedic surgeons already accept this same logic throughout the perioperative period. We measure hemoglobin because anemia increases risk, and we care about nicotine because wound healing matters. Surgeons collect this information when changes the probability of an outcome that matters for the patient, and it creates the opportunity to modify the risk factor before the complication occurs. There is no persuasive reason to treat OUD differently simply because its biology is complicated or because discussing addiction makes surgeons and patients uncomfortable

Orthopaedic surgeons should also help determine the ultimate value of AvertD rather than waiting for that question to be answered entirely by regulators, geneticists, addiction specialists, or the manufacturer. Adult reconstruction is particularly well suited to studying this topic because the procedures are highly planned, perioperative pathways are standardized, and opioid consumption can be closely measured. The questions that matter to me are practical: Can opioid exposure be reduced or eliminated without worsening pain or recovery? How often does an elevated risk actually change my prescribing? How does elevated risk relate to persistent opioid use? How does the test perform across my patient population? Do patients understand what elevated genetic risk actually means for them?

My own experience with roughly 50 to 100 tested adult reconstruction patients is too small and uncontrolled to answer any of those questions. However, it is enough to convince me that the information can change real conversations with real patients, and those conversations can change my pain management plan. This is why I think careful clinical adoption of AvertD and rigorous evidence generation on its performance can occur at the same time. Surgeons should use the test within the population for which it was approved, document what they do with the result, and contribute to the evidence that will ultimately determine whether my present enthusiasm is justified.

I keep returning to the mother in my office because her situation illustrates both the promise and the limits of AvertD. Her elevated result cannot tell us what her child’s genotype was, and it cannot establish that avoiding opioids after the fracture surgery would have changed her child’s trajectory. However, before her own knee replacement, she and her surgical team had information about her genetic susceptibility that was not previously available. That information was placed alongside all the other factors that influence OUD, and we used it to plan an opioid-free pain control strategy. The patient valued knowing the information for her own health history, and I valued knowing the information to change my decision-making while I still had time.

Discovering a risk factor before surgery allows orthopaedic surgeons to do something differently before the complication occurs. Surgeons will never eliminate all of the uncertainty surrounding opioids, and a 15-variant genetic classifier is certainly not going to solve the country’s opioid epidemic. However, the first opioid prescription is one clinical decision with incredibly important consequences. I see no persuasive reasons to make that decision without as much relevant information as possible. AvertD provides another piece of the puzzle, and I believe that it belongs in the preoperative assessment.

Disclosure

The author reports no financial or research relationships with SOLVD Health or AutoGenomics, Inc.