Abstract
Introduction
Patients with a history of Opioid Use Disorder (OUD) have higher postoperative complication rates and mortality in many settings. Yet, it remains poorly understood how the opioid epidemic has affected patients undergoing major lower extremity amputation (LEA) and whether outcomes differ by OUD status.
Methods
We conducted a retrospective chart review of all 689 patients who underwent major LEA at a large tertiary referral center from 2015 to 2021. This study assessed patient characteristics and long-term postoperative outcomes for patients with preoperative OUD.
Results
133 (19.3%) patients had a lifetime history of preoperative OUD. Preoperative OUD was associated with key characteristics, comorbidities, and outcome measures. OUD was significantly associated with younger age (P < .001), black race (P = .026), single relationship status (P < .001), BMI <30 (P = .024), no primary care provider (P = .004), and Medicaid insurance (P < .001). Comorbidities significantly associated with OUD include current smoking (P < .001), Human Immunodeficiency Virus (HIV; P = .003), and history of osteomyelitis (P < .001). Preoperative OUD independently predicted lower rates of 30-60-day readmission (odds ratio [OR] .54, P = .018) and 1-12-month reamputation (OR .41, P = .006). There was no significant difference in long-term mortality and follow-up.
Conclusion
This study demonstrates the prevalence of OUD in patients undergoing major LEA and reports associations and long-term outcomes. Our findings highlight the importance of recognizing OUD and raise questions about the mechanisms underlying its relation to rates of postoperative readmission and reamputation.
Introduction
In the United States alone, more than 150,000 atraumatic major amputation operations are performed annually. 1 Lower extremity amputation (LEA) is deemed major when the level is proximal to mid-tarsal joints. Major LEAs, including above, below, and through the knee, are most commonly indicated when patients with peripheral vascular disease or diabetes progress to gangrene, osteomyelitis, or sepsis despite more conservative therapies. Major LEA is also indicated for patients with extremity trauma with significant contamination or soft tissue damage. While amputations aim to preserve as much salvageable tissue and function as possible, amputation remains a costly operation. The combined price of hospitalization, surgery, follow-up, complication management, and prostheses can amount to a lifetime cost of more than $500,000 for patients and their families. 2
Major LEA operations are associated with significant morbidity and mortality that have drawn attention to common postoperative complaints, such as phantom limb pain. This pain felt in the amputated limb causes significant suffering in an estimated 64% of patients. 3 Unfortunately, attempts to enhance recovery with innovative postoperative pain plans are complicated by increasing rates of chronic pain, opioid prescription use, and illicit opioid use. Several studies have shown that preoperative opioid use is associated with higher rates of revision after total knee arthroplasty 4 and mortality following a multitude of outpatient procedures. 5 Preoperative opioid use is also linked to a higher risk for surgical site infections following hernia repair 6 and 90-day readmission following arthroscopic 7 and general 8 surgeries. Yet, it remains poorly understood how these effects translate to the setting of amputation. To our knowledge, no large studies have examined the prevalence of opioid use disorder (OUD) in patients undergoing major LEA and outcomes for that patient population. Understanding this relationship may guide medical teams in optimizing each patient’s preoperative assessment, surgical intervention, and postoperative pain management strategies.
This study addresses this literature gap through several objectives. First, we aimed to describe the prevalence of preoperative OUD in our major LEA patient population. Then we identified associations between preoperative OUD and patient factors, surgical indications, comorbid conditions, surgical complications, and long-term outcomes. Finally, we aimed to quantify the independent contribution of preoperative OUD to outcomes. We hypothesized preoperative OUD would be associated with many factors and independently predict worse rates for readmission, reamputation, mortality, and follow-up.
Method
This retrospective study included every adult who underwent below-knee, through-knee, or above-knee amputations at a single large tertiary referral center from 2015 to 2021. Exclusion criteria were an age of less than 18 years at the time of surgery and a lack of documented social history in the electronic medical record (EMR). 771 major LEA cases were identified for a total of 689 patients. EMR review identified preoperative OUD diagnoses made with the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition. 9 Criteria include the presence of cravings and repeated use as well as core features such as the development of tolerance and withdrawal. Any method of obtainment (prescription or illicit) and administration (injection or oral) were included given sparse and inconsistent documentation. Chart review also yielded patient demographics, comorbid diagnoses, postoperative infection, return to OR during the admission, and long-term course. Survival time was the time elapsed between the index major LEA operation and the date of mortality by any cause. Primary outcome measures include rates of readmission, reamputation, mortality, and follow-up. Data collection concluded on 12.31.22, so every patient was followed for a minimum of 1 year.
Bivariate analyses were with Chi-Square contingency tables. Age was distributed non-normally, so non-parametric one-way ANOVA (Kruskal-Wallis) tests were used to compare means by OUD status. For multivariable analysis, we incorporated variables with statistical significance in bivariate analysis in a backward-stepwise elimination multivariable logistic regression model for each primary outcome measure. All tests were completed with Jamovi v2.3.21.0 and IBM SPSS Statistics v29.0.1.0 with two-tailed with a significance threshold of P < .05.
Results
Patient Characteristics.
Statistical significance noted by *P < .05, **P < .01, and ***P < .001.
Patient Comorbidities.
Statistical significance noted by *P < .05, **P < .01, and ***P < .001.
COPD, Chronic Obstructive Pulmonary Disease; PAD, Peripheral Arterial Disease; CAD, Coronary Artery Disease; CKD, Chronic Kidney Disease.
Patient Outcomes Stratified Opioid Use Disorder.
Statistical significance noted by *P < .05, **P < .01, and ***P < .001.
Binomial Regressions for Opioid Use Dependence.
The following variables were included in the multivariable models: Age, race, relationship status, Body Mass Index ≥ 30, primary insurance, primary care, smoking status, and history of opioid use disorder, diabetes mellitus, hyperlipidemia, peripheral arterial disease, chronic kidney disease, malignancy, osteomyelitis, and immunocompromise. Statistical significance noted by *P < .05 and **P < .01.
Discussion
Preoperative OUD was highly prevalent and associated with many patient characteristics. Most notably, patients with a history of OUD were younger and more likely to be single, African American, non-obese, enrolled in Medicaid, and without a PCP. Our data also suggest history of OUD co-occurs more often with current smoking, chronic kidney disease, osteomyelitis, and HIV. In contrast, our patients with a history of OUD were less likely to have diabetes mellitus, hyperlipidemia, peripheral or coronary artery disease, or malignancy. These findings are reasonable given confounding by age, method of administration, and opioid source, none of which are accounted for in these bivariate analyses. The literature is mixed given differences in patient populations. For example, a sample of older patients prescribed different doses of opioids before total shoulder arthroplasty was found to have significantly more comorbidities than our patient population using opioids preoperatively. 10 Still, other socioeconomic factors such as income and housing may cloud the link between race, relationship status, primary care utilization, and insurance type given the overlap with baseline risk factors for OUD in Baltimore City. 11
However, the multivariable model controlled for the significant bivariable analyses and demonstrated that preoperative OUD was an independent predictor of significantly decreased rates of readmission and reamputation postoperatively. We found the effect for readmission in the interval of 30 to 90 days following the major LEA operation, which is consistent with the timeframe implicated following arthroscopic knee surgery. 7 Yet, differing surgical contexts as well as biologic or behavioral contributions may explain the opposite direction of the effect. We also found OUD predicted lower rates of reamputation from 30 days to 1 year following the major LEA operation. While Belkin et al. created a multifactorial model predicting 1-year readmission rates following transmetatarsal amputation, preoperative opioid use was not measured. Thus, our novel finding is difficult to interpret without rich literature to draw from. Nevertheless, we propose that the younger age and better average renal function in patients with OUD may account for the group’s lower risk for readmission and reamputation postoperatively. Our multivariable analysis is a separate study of the same cohort which showed that CKD and age predict greater 5-year mortality following major LEA. 12 However, the finding could also be confounded by factors our retrospective analysis did not measure.
This study should be interpreted with its strengths and weaknesses in mind. The largest advantage is the large, diverse sample. It draws from both the urban area immediately surrounding an academic tertiary care center as well as the suburban and rural areas in its catchment. Further, the study’s inclusion criteria cover every indication and major amputation level, limiting bias by etiology or surgeon preference. Still, the study relies on retrospective analysis of a heterogenous group, which limits the capacity to form causal inferences from the results. Further, although we aimed to complete comprehensive data collection, we cannot complete subgroup analysis without a more detailed characterization of patients’ OUD history. Additionally, we cannot control for every potential confounding variable, such as socioeconomic factors or surgical service-specific preferences.
Ultimately, the study aims to better understand the implications of the opioid epidemic in the context of an incredibly common set of surgical procedures. Our results demonstrate the prevalence of OUD and its link with numerous factors including an influence on readmission and reamputation rates independent of many confounders we controlled for. These findings highlight the importance of identifying OUD and understanding its link to postoperative outcomes. Other health care systems may benefit from replicating this analysis. Addressing this research gap is essential for formulating evidence-informed interventions that improve postoperative outcomes for patients of all backgrounds. Future research could further evaluate these relationships as well as the proposed explanation that OUD may be a proxy for the absence of strong risk factors for postoperative morbidity and mortality, old age, and chronic kidney disease. In particular, well-powered prospective studies would be better equipped to identify any causal relationships between OUD and outcomes following major LEA operations.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.