Introduction

Opioid prescribing has become increasingly common over the last two decades. [1, 2] The harms associated with opioid use have reached epidemic proportions in countries like the United States, [3, 4] where approximately 170,000 people died from opioid-related overdoses between 2015 and 2019. [5] The prevalence of opioid use has been bolstered by steady growth in the number of patients who are prescribed long-term opioid therapy. [1, 6] The practice of prescribing opioids for chronic non-cancer pain has become commonplace despite limited evidence supporting its efficacy [7, 8] and recommendations against the use of opioids as a first-line treatment for conditions like osteoarthritis. [9,10,11] Moreover, a growing body of evidence now indicates that this practice places patients at an increased risk of serious harms such as dependence, overdose, hyperalgesia, and mortality. [12,13,14]

As a consequence of these prescribing practices, a growing number of patients are now presenting to surgery with a history of opioid use. [15••] The clinical impact of this trend is particularly pronounced among total joint replacement patients, as more than half of these patients are currently being prescribed opioids in the year prior to surgery. [15••] Recent studies indicate that preoperative opioid users experience increased rates of early revision and infection, [16•, 17•, 18••, 19••] worse safety-related outcomes, [19••] reduced improvements in pain and function, [20, 21] and an increased risk of long-term opioid use following total joint replacement. [22] Despite the breadth of these adverse findings, in many instances, the mechanisms underlying these relationships remain unclear. In this review, we provide an overview of these recent findings and highlight areas in which further research is needed to inform improved perioperative care for this high-risk surgical population.

Patterns and Prevalence of Preoperative Opioid Use

In the largest available study of opioid use prior to total joint replacement, Jin et al. [15••] examined 473,781 US Medicare enrollees who underwent total hip replacement or total knee replacement between 2010 and 2014. This study reported that 64% of total hip replacement and 58% of total knee replacement patients had been prescribed at least one opioid in the year prior to surgery. [15••] Among those who underwent total hip replacement, 8% filled a prescription for opioids in each of the 12 months leading up to their surgery (i.e., continuous use), and among total knee replacement patients, 7% demonstrated a similar pattern of continuous preoperative use. [15••] The odds of a patient presenting to surgery as a continuous opioid user increased by approximately 62% between 2010 and 2014. [15••] Being African American or female, having a history of drug abuse or back pain, and having been prescribed depression or anxiety-related medications were all factors that contributed to this increase. [15••] Benzodiazepine and anticonvulsant use was also far more prevalent among continuous opioid users, [15••] despite evidence that the concurrent use of opioids and these forms of medications may place patients at an increased risk of opioid-related overdose. [23,24,25] There was also evidence of significant regional variation in the prevalence of continuous opioid use, with patients from the Southern and Western parts of the United States having nearly two times greater odds of exhibiting such patterns compared to those in the North-East. [15••]

The regional variation reported by Jin et al. [15••] lends support to previous studies that have documented substantial regional and state-level variation in general opioid prescribing practices in the United States. [26, 27] It also highlights the possibility that differences in the prevalence of preoperative opioid use may be partially driven by differences in healthcare settings. Variation in patterns of preoperative opioid use is often particularly stark when comparisons are made between the United States and other nations. A recent Finnish study of 12,448 patients found that fewer than 18% of knee replacement patients and 23% of hip replacement patients used mild opioids in the year before surgery. [28] More striking still was the fact that fewer than 1% of these patients were prescribed strong opioids during this same period. [28] The prevalence of preoperative opioid use in Australia has been reported to be as high as 49% among total knee replacement patients and 54% among total hip replacement patients in the year prior to surgery. [29•] Almost identical rates have been reported in New Zealand, with 54% of total hip replacement patients and 47% of total knee replacement patients filling a prescription for opioids in the year leading up to surgery. [30•]

Even at the higher end, these international estimates remain marginally lower than those generally reported in the United States. [31,32,33,34,35, 15••] Among large studies that measured opioid exposure over the year prior to surgery, the lowest reported American estimates ranged from 54% among total knee replacement patients [33] to 60% among total hip replacement patients. [34] At the other extreme, one recent study reported that among a commercially insured cohort of total hip or knee replacement patients, 87% filled a prescription for opioids in the year prior to surgery. [35] At present, it is difficult to offer more precise estimates of the overall prevalence of preoperative opioid use, as there is no widely agreed upon approach to measuring or reporting this information. Although many studies report on opioid exposure during the 360–365 days immediately before surgery, [19••, 15••, 29•] this approach is far from universally accepted. Many other studies examine shorter timeframes, generally ranging from 90 [36] to 180 days [37] before surgery, and a small number also report on longer timeframes of up to 3 years. [30•] Further difficulties emerge when attempting to compare patterns of exposure such as chronic, intermittent, or acute use, as there is considerable variation in the ways that such patterns of use are defined. Not only does this hamper our ability to understand the scope of this issue across different healthcare settings, it also limits our ability to systematically compare and synthesize findings from studies that have investigated the relationship between preoperative opioid use and surgical outcomes. Going forward, there is a pressing need for standardization in the way that patterns of preoperative use are measured and reported.

Who Prescribes Opioids to Patients Prior to Total Joint Replacement?

To better understand what drives these varied patterns of use, it is necessary to clarify who is prescribing opioids to patients prior to total joint replacement. By drawing on data from the Kaiser Permanente Total Joint Replacement Registry, Namba et al. [38, 39] reported that general and internal medicine providers were responsible for more than 70% of opioid prescriptions written in the year prior to total hip or knee replacement. In a smaller, single-institution study of patients being screened for total hip or knee replacement, Calkins et al. [40] reported that primary care providers were responsible for 60% of opioid prescriptions written in the 6 months prior to patients presenting to a preadmission clinic. Rheumatologists were responsible for 4 of 116 (3.4%) cases of opioid prescribing identified in this cohort; [40] however, the small sample from which this finding was drawn highlights how little is known about preoperative opioid prescribing by subspecialties within internal medicine. Nonetheless, these findings suggest that interventions targeting internal medicine and primary care providers may be effective in reducing the prevalence of preoperative opioid use. In this context, lessons should be drawn from existing prescriber education interventions, which have previously been shown to be effective in reducing opioid prescribing in the acute pain setting. [41] The design of such programs is likely to benefit from additional research that characterizes these providers in greater detail and clarifies the clinical decision-making process that leads to high rates of opioid prescribing in the months immediately prior to surgery.

Preoperative Opioid Use and Total Joint Replacement Outcomes

Early Revision

Over the last 3 years, studies have started to report that chronic preoperative opioid use is associated with an increased risk of early revision. In one of the earliest studies to specifically examine this relationship, Ben-Ari et al. [16•] reported findings from a cohort of 32,636 patients who underwent total knee replacement within the US Veterans’ Affairs system. Those patients with a history of long-term (i.e., > 3 months) opioid therapy prior to total knee replacement had 40% greater odds (OR 1.40, 95% CI: 1.19–1.64) of undergoing a revision procedure within a year when compared to patients without a history of long-term use. [16•] Concerns about whether these findings could be generalized to non-veteran populations have now largely been allayed by subsequent studies of large cohorts consisting of both commercially insured patients and Medicare enrollees. Recent findings from a cohort of 316,593 US Medicare enrollees over the age of 65 indicate that chronic opioid use is associated with an increased risk of revision at 30 (HR 1.63, 95% CI: 1.15–2.32), 60 (HR 1.40, 95% CI: 1.05–1.88), and 90 days (HR 1.58, 95% CI: 1.21–2.05) following total knee replacement. [19••] Studies of commercially insured populations have reported similar findings among large cohorts of total knee, hip, and shoulder replacement patients, with follow-up ranging from 180 days to 3 years. [18••, 42, 36, 37]

Despite the consistency of these findings, important limitations persist within this body of evidence. While many key clinical and demographic covariates were controlled for in the models presented in these studies, other potential confounders were not addressed. Most notably, none of these models controlled for hospital or implant-related factors, despite such factors having previously been shown to be risk factors for revision. [43] Moreover, none of these studies adequately explored whether the increased risk of revision among chronic preoperative opioid users is due to differences in specific causes of revision. While Ben-Ari et al. initially reported that there was no significant association between preoperative opioid use and cause-specific revision, [16•] they later noted that this was likely due to reduced sample sizes in these sub-analyses. [44] Although it has yet to be tested, it is plausible to suggest that opioid-induced hyperalgesia [14, 45] or immunosuppression [46, 47] may explain higher rates of early revision due to pain and infection among preoperative opioid users. Gaining a clearer understanding of the mechanisms through which preoperative opioid use may result in higher rates of revision is a crucial step towards being able to offer appropriately tailored perioperative care to this high-risk patient population.

Infection

Mounting evidence now suggests that opioid use is associated with an increased risk of infection. [47] In the first published study examining this relationship in a cohort of total joint replacement patients, Bell et al. [17•] found that a history of opioid use was an independent risk factor (OR 1.53, 95% CI, 1.14–2.05) for developing a periprosthetic infection in the 2 years following total hip or knee replacement. Unfortunately, this study’s reliance on patients self-reporting their opioid usage limited the authors’ ability to control for the dose, duration, or recency of prior opioid therapy. This may also have resulted in the true rate of preoperative use being under-reported. [48] Although no other studies have specifically explored links between preoperative opioid use and periprosthetic infection, a recent study by Blevins Peratikos et al. [18••] reported that rates of surgical site infection were 35% higher among total hip, knee, and shoulder replacement patients with a history of opioid use in the 90 days prior to surgery (HR 1.35, 95% CI 1.14 to 1.59) when compared to opioid-naïve patients. As this study relied upon comprehensive administrative claims data rather than patient recall, it was also able to document a dose-response relationship between preoperative opioid use and surgical site infection. [18••]

Establishing a causal link between preoperative opioid use and infections following total joint replacement is likely to be hampered by the rarity of surgical infections, which limits the feasibility of conducting sufficiently powered randomized trials to examine this relationship. In the absence of evidence from randomized trials, our understanding of this relationship may still be improved by a combination of preclinical and observational research. To date, preclinical research has shown morphine to have immunosuppressive properties; however, not all opioids impact immune function and the immunosuppressive effects of exposure to certain opioids has yet to be adequately researched. [46, 47] Supplementing additional preclinical research with high-quality observational studies is likely to improve our understanding of this relationship. Findings from such research may also provide support for novel patient optimization strategies, such as transitioning patients to non-immunosuppressive opioids prior to surgery.

Short-Term Complications and Safety-Related Outcomes

In a recent study of 316,593 US Medicare enrollees who underwent total knee replacement, Kim et al. [19••] found that patients who filled an opioid prescription in each of the 12 months leading up to surgery experienced worse clinical and safety-related outcomes. After adjusting for key demographic covariates including age, sex, race, and region of residence, Kim et al. [19••] reported that chronic opioid users were at an increased risk of all-cause mortality, hospital readmission, opioid overdose, revision operation, vertebral and non-vertebral fractures, pneumonia, and bowel obstruction at 30, 60, and 90 days following surgery. However, after adjusting for clinical factors including comorbidity score, frailty score, and the number of unique medications used before surgery, this relationship was only statistically significant at all time points examined for opioid overdose (30-day HR 4.82, 95% CI: 1.36–17.1; 60-day HR 7.91, CI 2.50–25.0; 90-day HR 13.6 95% CI: 4.70–39.6), vertebral fractures (30-day HR 2.37, 95% CI: 1.37–4.09; 60-day HR 2.42, 95% CI: 1.70–3.44; 90-day HR 2.36, 95% CI: 1.79–3.13), and revision operation (30-day HR 1.63, 95% CI: 1.15–2.32; 60-day HR 1.40, 95% CI: 1.05–1.88; 90-day HR 1.58 95% CI: 1.21–2.05). [19••] The relationship with readmission (60-day HR 1.12, 95% CI: 1.04–1.20; 90-day HR 1.19, 95% CI: 1.12–1.27) and non-vertebral fractures (90-day HR 1.75, 95% CI: 1.11–2.77)) remained significant, though not at all time points. [19••]

These findings indicate that differences in baseline clinical risk profiles play a prominent role in the relatively high rates of adverse outcomes among chronic opioid users. This, in turn, suggests that we are limited in the conclusions that can be drawn from much of the existing research examining rates of re-admission [37, 42] and risk of short-term complications [42, 49] among preoperative opioid users. Such caution is necessary as few of these studies have controlled for key clinical characteristics other than broad measures of comorbidity. Importantly, the models utilized by Kim et al. [19••] failed to control for some potentially pertinent covariates, including the severity of the underlying disease and socio-economic status, alongside a range of hospital or surgeon-related factors. Even with this being the case, the approach employed by Kim et al. [19••] highlights the need to carefully address a broad range of risk factors when preparing preoperative opioid users for surgery. Put simply, focusing only on addressing a patient’s history of opioid use is unlikely to be the most appropriate means of ensuring that they have the best possible outcome from surgery.

Pain, Function, and Health Economic Outcomes

Over the last decade, several studies have examined the relationship between preoperative opioid use and patient-reported outcomes following total joint replacement. [50,51,52,53,54,55] At 6 months following total knee replacement, Smith et al. [50] found that those who had been prescribed opioids prior to surgery experienced lower levels of improvement in pain when compared to a clinically similar opioid-naïve patients. An earlier study by Zywiel et al. [52] reported similar findings at a mean follow-up of 3 years following total knee replacement. Pivec et al. [51] reported that, when compared to a matched cohort of opioid-naïve patients, preoperative opioid users experienced fewer postoperative improvements in pain and function following total hip replacement. Despite consistent results in studies of hip and knee replacement patients, [50,51,52,53] these findings appear to conflict with available studies examining total shoulder replacement. Morris et al. [55] found that preoperative opioid users and opioid-naïve patients experienced almost identical improvements in pain and function at an average of 3 years following reverse shoulder replacement. A subsequent study by Morris et al. [54], which examined a cohort of anatomic shoulder replacement patients, reported comparable findings. However, these two studies did not attempt to control for differences in baseline pain and function through matching or other statistical methods, which may explain this apparent conflict with evidence from matched cohorts of hip and knee replacement patients.

Although it has yet to be tested directly, lower levels of improvement in pain and function among preoperative opioid users may be the result of opioid-induced hyperalgesia. [52] If this is the case, opioid tapering prior to surgery may reduce the adverse effects of prior opioid exposure. [45] Recent observational research has lent support to this hypothesis. [53] By examining 41 chronic opioid users who successfully reduced their opioid consumption by at least 50% prior to total hip or knee arthroplasty, Nguyen et al. [53] found that these patients had similar improvements in pain and function to opioid-naïve patients. Unfortunately, the retrospective design of this study leaves open the possibility that successful tapering acted as a proxy for other unmeasured patient characteristics, such as greater willingness to comply with care or higher levels of social support. To determine if preoperative opioid tapering interventions are a safe and effective means of improving patient-reported outcomes following total joint replacement, there is a pressing need for high-quality randomized trials to be conducted. Prospective trials of preoperative opioid tapering interventions are potentially relevant to all other outcomes discussed in this review. However, the relative rarity of many of these serious adverse outcomes may limit the feasibility of trials in which the primary outcome is, say, periprosthetic infection or early revision.

Despite reporting that opioid users may experience reduced postoperative improvements in pain and function relative to opioid-naïve patients, each of these studies still found that opioid users experience meaningful improvements in their symptoms following surgery. The degree to which these patients experience symptomatic improvement has important implications for the cost-effectiveness of both surgical and non-surgical care of patients with end-stage arthritis. To date, available health economic models have primarily been concerned with the impact of prescribing opioids to patients who are likely to require surgery at some point in the future. [56, 57] These models indicated that opioids such as tramadol or oxycodone may delay or reduce the utilization of total knee arthroplasty. [56, 57] Nonetheless, this approach was found to be cost-ineffective over the longer term, as these models accounted for the higher rates of early revision and reduced improvements in postoperative quality of life that are associated with preoperative opioid use. [56, 57] While these assumptions about the impact of preoperative opioid use on surgical outcomes were based upon a study that is now almost a decade old, [52] they have largely been supported by more recent studies. [50, 51, 53, 19••] With that being said, further health economic studies are still necessary to establish if these findings can be generalized to other healthcare settings or to procedures other than total knee replacement.

Importantly, studies have yet to directly examine the cost effectiveness of offering total joint replacement to patients who have already been initiated onto opioids. At present, the cost-based data needed to address this question are sparse, indirect, and largely drawn from a single study. In this study, Blevins Peratikos et al. [18••] found that the median level of healthcare spending on preoperative opioid users was $964 higher than spending on opioid-naïve patients over the year following surgery. However, due to lower levels of spending in the year prior to surgery, opioid-naïve patients appear to have experienced a greater median increase in annual medical spending following surgery. This demonstrates that the relationships between opioid use, postoperative quality of life, and the cost-effectiveness of total joint replacement are far from straightforward. Careful examination of these relationships is needed to allow health-care funders to make informed decisions about financing total joint replacement for patients who exhibit high-risk patterns of preoperative opioid use. The findings of such research may also impact upon funders’ willingness to finance targeted interventions that aim to improve post-operative outcomes for preoperative opioid users.

Long-Term Postoperative Opioid Use

A significant proportion of total joint replacement patients continue to use opioids for a prolonged period following surgery. In a recent systematic review, Kent et al. [22] reported on 17 studies that examined the rate of prolonged opioid use following total hip or knee replacement in the United States and Canada. In those studies deemed to be of at least moderate quality by Kent et al., reported rates of prolonged postoperative use ranged from 0.6 to 4% in opioid-naïve patients, to between 35 and 68% in patients with a prior history of opioid use. [22] These findings are broadly consistent with studies examining other forms of total joint replacement (i.e., total shoulder replacement [58] and total ankle replacement [59]), studies from outside North America (i.e., Australia [60, 61] and New Zealand [30•]), and studies published after the literature search conducted by Kent et al. [31, 62, 63] Commonly reported risk factors for prolonged opioid use following total joint replacement include anxiety, depression, substance abuse, and chronic pain conditions. [22] Each of these risk factors are potentially important targets for interventions that aim to lower rates of prolonged postoperative opioid in both opioid-naïve patients and those with a prior history of opioid use. To reduce the harms associated with postoperative opioid use most effectively, it is important to also target the practice of co-prescribing opioids alongside gabapentinoids or benzodiazepines, which has been shown to place patients at a heightened risk of opioid-related overdose. [23,24,25] While such co-prescribing has become more common among the general population in recent years, [24, 64] such trends have yet to be directly quantified in the patients who have recently undergone surgery.

Over and above these other factors, preoperative opioid use has consistently been found to be the strongest predictor of long-term use of opioids following total joint replacement. [22] The drastically different levels of postoperative opioid consumption between preoperative opioid users and opioid-naïve patients raise an important and as yet unanswered question: To what extent is the relationship between preoperative opioid use and adverse surgical outcomes mediated by higher rates of postoperative opioid consumption? One plausible understanding of the relationship between preoperative opioid use and adverse outcomes is that preoperative use (i) directly affects adverse outcomes immediately following surgery through mechanisms such as hyperalgesia or immunosuppression and/or (ii) indirectly affects on adverse outcomes through increasing post-operative opioid consumption over the mid-to-long term. Better understanding the degree to which these relationships are mediated by postoperative opioid consumption may highlight new avenues for interventions that target preoperative opioid users in the weeks and months following surgery.

Finally, it is important to note that there is substantial variation in the ways that prolonged postoperative opioid use is defined in the current literature. [22] In a recent study of opioid-naïve patients undergoing elective surgery, Thiels et al. [65] were the first to specifically investigate the impact that divergent definitions can have on reported rates of prolonged post-operative opioid use. Thiels et al. [65] found that 7.1% of their cohort satisfied one widely-employed definition of prolonged postoperative opioid use, which requires patients to have filled at least one opioid prescription between 90 and 180 days following surgery. In this same cohort, only 0.46% of patients met the widely used CONSORT definition, [66] which requires patients to have an episode of opioid use starting in the 180 days after surgery that also spans ≥ 90 days and includes either ≥ 10 opioid fills or ≥ 120 days’ supply of opioids. [65] The magnitude of the difference reported simply as a result of employing different definitions, once again highlights the need for a standardized approach to measuring and categorizing patterns of opioid use both prior to and following total joint replacement.

Conclusion

Total joint replacement patients with a recent history of opioid use generally experience higher rates of infection, revision, short-term complications, and prolonged postoperative opioid use, along with fewer improvements in pain and function following surgery. These risks are particularly pronounced among chronic opioid users. However, certain limitations in the available evidence constrain our ability to draw important conclusions about how to best care for these patients (see: Table 1). As available evidence is drawn largely from observational studies, the extent to which preoperative opioid use directly affects these outcomes remains unclear. Nevertheless, it is plausible that certain outcomes are impacted by opioid use before and after surgery through mechanisms such as opioid-induced hyperalgesia and immunosuppression. By gaining a clearer understanding of these mechanisms, it may be possible to design and implement targeted perioperative interventions for those with a history of opioid use. One such intervention, which has shown early promise in observational research, involves carefully tapering or discontinuing patients from opioids prior to surgery. Going forward, the degree to which the risks associated with preoperative opioid use can be mitigated through preoperative opioid tapering—or other forms of intervention—should be evaluated through randomized trials whenever doing so if feasible. When sufficiently powered trials are infeasible due to the infrequent nature of the outcome being studied (e.g., revision, periprosthetic infection), carefully designed observational studies may also be necessary to inform tailored care for this high-risk population.

Table 1 Five key recommendations
Full size table