To Include or Exclude? That Is the Question for Clinical Researchers Investigating in the Current Opioid Epidemic

While randomized controlled trials are considered the gold standard in clinical research, a crucial limitation is their lack of generalizability, partly due to strict eligibility criteria. With the ongoing opioid epidemic, this seems particularly true for regional anesthesia-related randomized controlled trials because preoperative opioid use is a commonly applied exclusion criterion. Given the increasing number of opioid-tolerant surgical patients, such exclusion criteria may severely impede study enrollment. Our first-hand experience with 3 recent randomized controlled trials impacted by chronic opioid use among our population has resulted in this Open Mind, in which we will review the following: (1) the problems associated with restricting randomized controlled trials to opioid-naive populations and (2) possible solutions to improve recruitment and external validity. PROBLEM 1: SLOW RECRUITMENT OF AN OPIOID-NAIVE POPULATION The United States consumes 80% of the world’s opioid supply,1 and the amount of opioids prescribed has tripled from 1999 to 2015.2 Correspondingly, opioid tolerance is becoming increasingly common because previous retrospective investigations demonstrate a large percentage of patients reporting opioid use before orthopedic surgery. For example, Cheah et al3 reported a preoperative opioid use prevalence of 52% in patients undergoing total shoulder arthroplasty between 2012 and 2015, while this was 35% in patients undergoing primary total hip and knee arthroplasty between 2012 and 2014.4 These observations mirrored our own institutional experience; we therefore instituted less stringent exclusion criteria in 2014, permitting patient inclusion with some opioid exposure at baseline for our randomized controlled trials. We defined our preoperative opioid use limits, constituting grounds for study exclusion, as >30-mg oral morphine equivalents per day for <1 month or >5 mg of oral morphine equivalents per day for ≥1 month. Despite this, a review of our screening logs from our last 3 randomized controlled trials5,6 indicated that preoperative opioid use remained an important factor for slow trial accrual. Of all patients screened for eligibility, 12% (79 of 682 patients; September 2014 to December 2015),5 7% (52 of 754 patients; January 2015 to March 2016),6 and 22% (89 of 402 patients; November 2016 to February 2018; ClinicalTrials.gov, NCT02876055) represented exclusions based on preoperative opioid use in our studies evaluating patients undergoing total knee, hip, and shoulder arthroplasties, respectively. A brief review of previous literature demonstrates a substantial number of regional anesthesia-related randomized controlled trials not reporting the exact number of patients excluded due to preoperative opioid use, making it difficult for us to compare findings.Figure 1.: The number of patients excluded solely due to preoperative opioid use (gray) and patients excluded due to preoperative opioid use with coexisting study exclusions (blue) for each clinical trial. THA indicates total hip arthroplasty; TKA, total knee arthroplasty; TSA, total shoulder arthroplasty.Figure 2.: The number of patients enrolled and excluded solely due to preoperative opioid use per month during the study recruitment period for the total knee arthroplasty (A),5 total hip arthroplasty (B),6 and total shoulder arthroplasty (C) study (NCT02876055). The dashed line represents the theoretical stop date for enrollment if the patients excluded for opioid use only were enrolled into their respective clinical trials.A consequence of slow recruitment in randomized controlled trials is a delay in the expected completion date. We calculated theoretical stop dates of enrollment for the last 3 randomized controlled trials conducted at our institution, in case of (theoretical) elimination of the preoperative opioid use exclusion criterion. Assuming we enrolled all preoperative opioid use patients with no other coexisting study exclusions (ie, these patients would have otherwise fulfilled all aspects of eligibility criteria) (Figure 1), this analysis suggests that 1–4 months of additional recruitment time (over a total duration of 14–16 months) was needed due to exclusion of patients with preoperative opioid use (Figure 2). Previous literature suggests that eligibility criteria are crucial in determining effective study completion. In a review of 122 trials, Campbell et al7 reported that fewer than one-third of the trials were successful in recruiting 100% of their original target within a prespecified time frame, while most of the other trials requested a time extension and/or supplementary grant to complete the study. During the early recruitment process, “fewer eligible than expected” was the most common reason for slow enrollment (25%).7 Another investigation of close to 2600 phase II/III trials reported 19% of trials being terminated due to recruitment failure or failing to achieve ≥85% of expected sample size (up to 26% if using failure to achieve 100% of expected enrollment cutoff).8 Similarly, a “greater number of eligibility criteria” was identified as an implicating factor for poor recruitment. PROBLEM 2: PREOPERATIVE OPIOID USE IS A CONFOUNDING VARIABLE The main rationale for excluding patients with preoperative opioid use in regional anesthesia-related randomized controlled trials is that it may affect the perioperative analgesic protocol under study as well as the outcome (often pain or opioid use) and therefore can be defined as a confounding variable. Indeed, opioid use may negatively impact perioperative care due to its association with complications such as ileus, urinary retention, respiratory depression, and sedation.9 In turn, this may lead to higher costs,10 increased length of stay,10,11 and a higher probability of discharge to a rehabilitation facility.10,11 Furthermore, chronic use of opioids produces deleterious effects to the central nervous system such as increased central sensitization and development for opioid-induced hyperalgesia.12 Opioid use has also been strongly linked to psychological disorders (eg, anxiety and depression).13 Consequently, preoperative opioid use has been associated with higher pain scores, worse functional outcomes, and higher opioid consumption after orthopedic surgical procedures.3,4,14 Therefore, to estimate an unbiased effect of an analgesic protocol under study one has to take into account preoperative opioid use when applicable. Theoretically, randomization would minimize the confounding risk of opioid use through equal allocation because randomizing treatment assignments balances known and unknown variables among the intervention groups. However, including opioid-tolerant patients in an randomized controlled trial would not only increase the variance in results but is also more likely to result in suboptimal adherence to protocols under study. Increased variance would require larger sample sizes with negative trade-off consequences such as effort, direct costs, study fatigue, and delayed reporting of results.15 PROBLEM 3: EXTERNAL VALIDITY Randomized controlled trial patient enrollment can be highly selective due to stringent eligibility criteria, which tend to reduce study variability, reduce sample size, and narrow the spread of potential confounders.15 However, it is imperative to weigh these advantages against the generation of a homogenous group which may not represent daily clinical practice. The trade-off between internal validity (ie, whether study results are valid for the original study population) and external validity (generalization to the real-world population) represents one of the main challenges in interpreting results from randomized controlled trials.16 Because randomized controlled trials tend to generate a homogenous cohort, clinicians are left stranded to extrapolate the results to real-world clinical practice. In the context of the opioid epidemic, this appears particularly true for regional anesthesia-related randomized controlled trials, which provide results that may not be generalizable because a growing number of patients are opioid tolerant at the point of care. Importantly, principles of pain management strategies for opioid-tolerant patients are commonly derived from randomized controlled trials which exclude such patients. This may be troublesome because multimodal analgesic protocols, which incorporate regional anesthesia techniques, may provide desirable outcomes for opioid-naive patients but fail for opioid-tolerant patients.3,4 Because the beneficial effects of interventions can be dependent on patient characteristics,16 evaluating the external validity of published regional anesthesia-related randomized controlled trials to the (growing) opioid-tolerant population would be valuable. Importantly, randomized controlled trials should maintain transparency regarding eligibility criteria by utilizing a standardized reporting tool such as the Consolidated Standards of Reporting Trials guidelines17 and consider adopting a “To whom do these results apply?” section, as suggested by Rothwell.16 This will greatly assist fellow clinicians to determine the applicability of study results to their patients. SOLUTION 1: LIBERALIZATION OF PREOPERATIVE OPIOID USE THRESHOLDSTable.: Preoperative Opioid Use Thresholds Used in Regional Anesthesia-Related Randomized Controlled TrialsAn intuitive approach to address the increasing number of opioid-tolerant patients in regional anesthesia-related randomized controlled trials is to include them in studies. Several strategies may be followed, from including all patients with preoperative opioid use to only those meeting prespecified cutoffs regarding daily dose used or duration of opioid use. While this may greatly improve recruitment and external validity of study results, there currently is no standard and various—seemingly random—strategies have been applied in recent studies (Table). Given the known potential effect of preoperative opioid use on outcomes such as pain scores and perioperative opioid use, 1 analytic approach could be to adjust for this variable in subsequent analysis.18 Of note, the shortcomings of subgroup analyses should be acknowledged or addressed.19 It will be important to detail these strategic and analytic choices a priori in the randomized controlled trial design phase and analysis plans. Given the significance of randomized controlled trial eligibility criteria in general and the increasing concern of preoperative opioid use due to the opioid epidemic, we encourage researchers to reevaluate their eligibility criteria and consider initiating a pilot study assessing the baseline demographics of their patient population. Pilot studies are particularly useful for evaluating the feasibility of recruitment with the proposed eligibility criteria. Information gathered from pilot studies would provide researchers the opportunity to refine the study methodology and help support the design of a future definitive randomized controlled trial.20 However, pilot studies may require a significant amount of time to perform. SOLUTION 2: RANDOMIZED CONTROLLED TRIALS INCLUDING ONLY OPIOID-TOLERANT PATIENTS Given the growing number of opioid-tolerant patients, another (sparsely used) approach would be to perform a trial exclusively targeting opioid-tolerant patients only. With the lack of randomized controlled trials evaluating analgesic protocols in opioid-tolerant patients, such studies would provide clinically valuable information. The main drawback of this approach would be, again, generalizability; in this case, however, generalizability to opioid-naive patients. In addition, opioid-tolerant patients may be heterogeneous in terms of opioid type, dose used, duration of use, and route of administration. With respect to preoperative opioid use, researchers should evaluate for confounding and interaction (effect modification).21 Smaller clinical trials are vulnerable to baseline heterogeneity15; therefore, confounding may occur if there is an imbalance in preoperative opioid dose between randomized groups and preoperative opioid dose impacts outcome. Interaction occurs when the magnitude of the effect of an intervention is different across varying levels of preoperative opioid use.21 As such, assessing interaction would help identify subgroups of opioid-tolerant patients (based on different amounts of opioid use) who may benefit the most and least from an intervention. SOLUTION 3: SHIFT TOWARD PRAGMATIC-ORIENTED TRIALS Briefly, pragmatic trials are designed to include clinically relevant interventions in a diverse cohort with the intention of maximizing the applicability of results to everyday practice.22 Patients are not burdened with substantial data collection, and clinicians are not constrained to strict protocols; in fact, clinicians are provided the freedom to treat as they normally would, thus reflecting their daily practice.22 While pragmatic trials are seen as a feasible solution in the trade-off between internal and external validity, they are often poorly received by the research community who perhaps seem to be unfamiliar with the methodology or misinterpret the scientific rigor of pragmatic trials.23 The major limitation is reduced statistical precision due to a broad selection of patients, individual practice variability, and nonstringent data collection.22 Moreover, to improve reliability in study findings and reduce confounding, larger sample sizes are required.23 With regard to the opioid epidemic, including all opioid-tolerant patients into a clinical trial and providing clinicians more flexibility in managing acute pain would be more representative of real-world practice, and therefore may impact decision-making. However, additional analyses would be required to determine the value of interventions in different subgroups (ie, opioid-naive versus opioid-tolerant or different opioid-tolerant groups). Another approach would be to perform an explanatory randomized controlled trial to determine the efficacy of a specific intervention followed by a pragmatic-oriented trial, or even a large cohort study, to evaluate its (real-world) effectiveness.23 Prospective observational studies can complement data from an randomized controlled trial by confirming and extending the results obtained by randomized controlled trials to a sizeable, diverse patient population.24 More importantly, observational studies allow researchers to explore the effectiveness of an intervention over a longer follow-up period and between various relevant subgroups, capture data on treatment compliance and persistence in a real clinical setting, and examine long-term treatment safety and tolerability.24 SOLUTION 4: MULTICENTER TRIALS Partnering for multicenter trials with other institutions and enrolling all opioid-naive and opioid-tolerant patients could speed enrollment, produce a larger sample size, and improve generalizability by combining diverse patient populations and evaluating treatment efficacy from different geographical locations.25,26 Furthermore, assessing interaction between varying amounts of preoperative opioid use (eg, none, low, moderate, high) and treatment effect may also be implemented here. Although this appears to be the optimal solution, conducting multicenter trials is a difficult undertaking and entails a complex organizational process. Establishing a steering committee, obtaining institutional review board approval for each research site, substantial funding, authorship agreement, and proper training are only a few of the many obstacles that need to be established before beginning a multicenter trial.25,26 As such, months to years in extensive preparation can delay study commencement. FUTURE DIRECTION The recent designation of the opioid epidemic as a national public health emergency emphasizes its national health implications while also recognizing the challenges in research ahead. Increased public awareness has generated the development of several novel strategies, such as preoperative patient education on opioids and alternative pain management techniques,27 preoperative opioid weaning programs,28 and multidisciplinary perioperative pain programs.29 Important future avenues for research that will improve these strategies and may fill pertinent evidence gaps include studies evaluating clinically meaningful cutoffs for preoperative opioid use to be included in randomized controlled trials, and optimal analgesic strategies to be applied in opioid-tolerant patients. In addition, the perioperative surgical home practice model can play a pivotal role to improve pain management and curtail opioid prescriptions in the chronic pain population.30 SUMMARY The opioid epidemic has adversely affected patient recruitment and consequently prolonged time until study completion for our recent randomized controlled trials. Opioid-tolerant surgical patients create a dilemma between internal and external validity in regional anesthesia-related randomized controlled trials. We have highlighted several problems performing experimental trial research in the opioid epidemic while also detailing potential solutions. DISCLOSURES Name: Jason K. Panchamia, DO. Contribution: This author helped write and revise the manuscript. Name: Adam W. Amundson, MD. Contribution: This author helped write and revise the manuscript. Name: Jashvant Poeran, MD, PhD. Contribution: This author helped write and revise the manuscript. Name: Ngoc Tram V. Nguyen, BA. Contribution: This author helped write and revise the manuscript. Name: Sandra L. Kopp, MD. Contribution: This author helped write and revise the manuscript. Name: Rebecca L. Johnson, MD. Contribution: This author helped write and revise the manuscript. This manuscript was handled by: Richard Brull, MD, FRCPC.