Abstract
Objective
Endovascular techniques continue to be increasingly utilized to treat vascular disease, but the effect of these minimally invasive techniques on opioid use following surgery is not known.
Methods
Using Medicare data, we identified opioid-naive patients undergoing vascular procedures between 2009 and 2017. We selected patients ≥65 years old with continuous enrollment 12 months before and 6 months after surgery and had no additional operations. We defined new persistent opioid use (NPOU) as one or more opioid prescription fills both between 4–90 and 91–180 days postoperatively. Multivariable regression was performed for risk adjustment, and frequencies of NPOU were estimated between endovascular and open techniques to compare surgical approach.
Results
A total of 77,767 patients were identified, with 2.6% of all patients developing new persistent use. In addition to the identification of several risk factors for new persistent use, patients undergoing endovascular carotid or vertebral interventions were found to have higher adjusted frequencies of persistent use compared to those undergoing open interventions (3.0% vs. 1.8%, p < 0.001) as did those undergoing endovenous compared to open vein procedures (2.2%, vs. 1.6%, p = 0.019). We found no difference for peripheral vascular or aortic/iliac procedures.
Conclusions
Patients undergoing vascular surgery are at high risk for new persistent use. Undergoing endovascular carotid or venous surgery was associated with an increased risk of NPOU, whereas no differences were found between endovascular and open approaches for peripheral arterial or aortic disease.
Keywords
Introduction
The opioid epidemic continues in the United States, with more than 130 deaths per day from opioid-related drug overdoses in 2017 and the number of individuals dying from opioid-related causes now surpassing deaths from motor vehicle collisions and gun violence.1,2 Opioids are effective for acute pain management and are commonly provided following surgical procedures; however, there is mounting evidence that a proportion of patients may transition to chronic opioid use and continue to fill prescriptions for opioid medications past the period in which acute pain would typically resolve.3–5
Prior studies have described opioid utilization following major and minor operations including total knee and hip arthroplasty, pediatric surgery, and curative intent oncologic surgery, among others.3,6–8 These prior studies have suggested several important risk factors including prescription size, presence of medical comorbidities, mental health disorders, and prior pain disorders as increasing the probability of new persistent opioid use (NPOU). Patients undergoing vascular surgical procedures represent a unique population with high rates of medical comorbidities as well as pain disorders, but patterns of opioid use following vascular surgical operations have not been described.
Previous attempts to investigate the effect of degree of invasiveness on frequency of new persistent use have been limited to laparoscopic versus open general surgical operations, which have been limited in their ability to separate the effect of the disease process and prescribing habits from the effect of pain experienced from the surgical procedure.9–11 Within the context of these limitations, these studies collectively have not shown a significant decrease in risk of new persistent use for patients undergoing less invasive procedures. The increasing use of endovascular procedures for treatment of vascular diseases represents a unique opportunity to decouple the effect of invasiveness and pain from the underlying disease process or patient characteristics.
In this study, we investigate the frequency of NPOU among patients undergoing a variety of vascular surgical procedures. We hypothesized that patients undergoing vascular surgical procedures would have high rates of NPOU postoperatively. Furthermore, we sought to determine what effect, if any, surgical approach (i.e., endovascular vs. open) had on the frequency of NPOU among vascular patients.
Materials and methods
Data source and study population
We used a random 20% sample of Medicare fee-for-service claims data from 1 January 2009 through 30 June 2017. The Centers for Medicare & Medicaid Services maintains this database using claims submitted by hospitals where Medicare beneficiaries receive care. This database includes patient information regarding age, sex, race, and comorbidities. It also includes procedural codes and information on outcomes, resource utilization, and costs. For this study, we selected patients who underwent open or endovascular abdominal aortic aneurysm or iliac artery aneurysm repair, open or endovascular carotid or vertebral artery surgery, open or endovascular treatment of varicose veins, and open or endovascular peripheral vascular intervention/lower extremity bypass. A complete list of Current Procedural Terminology codes used to identify and categorize patients is included as Supplementary Table 1.
We excluded patients less than 65 years of age. Furthermore, we excluded patients who discharged to a location other than home, had a length of stay greater than 30 days, or had discontinuous enrollment in Medicare Parts A, B, and D 1 year before admission and 6 months after discharge. We also excluded patients who underwent a subsequent surgical procedure within 180 days of discharge from their index surgical hospitalization. To investigate new opioid users, we additionally excluded patients who filled at least one opioid prescription between 365 and 31 days prior to their surgical admission. Patient enrollment is reported graphically in Figure 1. Enrollment schema. Final cohort size = 77,767.
Given the use of retrospective data, this cohort study was deemed exempt from approval and informed consent by the Institutional Review Board at the University of Michigan, Ann Arbor. This report follows the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline for cohort studies. 12
Outcomes and variable definitions
This study was designed to identify the relationship between vascular surgical operations and the development of the primary outcome, NPOU. This was defined as any patient that filled at least two opioid prescriptions, with at least one prescription within each of the following time periods: a prescription within 90 days following discharge, and a prescription within 91–180 days following discharge.
Age, sex, race (white, Black, Hispanic, and other), geographic region, and urban/rural residence as well as whether or not the surgery was elective were included as covariates. Comorbidity was quantified using Charlson Comorbidity Index (CCI) score. The CCI scores were divided into four categories, none (CCI 0), mild (CCI 1 or 2), moderate (CCI 3 or 4), or severe (CCI ≥5). Additional variables investigated include whether the patient filled a perioperative opioid prescription (within 30 days prior to admission or within 3 days after discharge) as well as the size of the perioperative prescription converted to number of 5 mg oxycodone tablets. As complications can be pain generating and affect the trajectory of recovery/opioid utilization, we included the occurrence of postoperative complications as an independent covariate in the model. Specific postoperative complications included were myocardial infarction, deep vein thrombosis/pulmonary embolism, respiratory failure, reintubation, aspiration pneumonia, atrial fibrillation/flutter, incomplete heart block, complete heart block, ventricular fibrillation/flutter, delayed gastric emptying, gastrointestinal bleeding, ileus, intestinal obstruction, cardiac arrest, clostridium difficile infection, pneumonia, surgical site infection, urinary tract infection, sepsis, acute renal failure, urinary retention requiring catheter placement, and acute limb ischemia/thrombosis. Additional covariates include mental health disorders, categorized into the following groups: anxiety, mood, schizophrenia, other mental health disorders, and drug/substance abuse. A history of pain disorder was included as a covariate in the following categories: arthritis, back, and other pain disorder. Concomitant perioperative fill of any prescription for benzodiazepines, sedatives, hypnotics, anxiolytics, or gabapentinoids was also included as a categorical yes/no variable. Dual eligibility (Medicare and Medicaid) was also assessed and included as a covariate. Finally, because of changing patterns of opioid prescriptions over time, year of surgery was included as a covariate.
Statistical analysis
We report descriptive statistics to detail the clinical and demographic characteristics of the study cohort. Chi-square tests were used to assess unadjusted associations between patient/procedure characteristics and NPOU with the exception of age, which was analyzed using Student’s t-test, and prescription size, which was assessed with the Mann–Whitney U test as it was a non-normally distributed.
We then utilized multilevel logistic regression to identify patient and procedural factors that were associated with NPOU in the context of some patients clustering within surgeons. Covariates included in this model were patient demographics such as age, sex, race, and geographic region, comorbidities (defined as above) mental health/substance abuse disorders, perioperative prescription fill as well as size, pain disorders, dual eligibility, surgical year, complication occurrence, elective/non-elective surgery status, and surgical procedure. Adjusted frequencies of new persistent use were calculated for each surgical procedure, and pair-wise comparisons were made between open and endovascular counterparts using chi-square tests. Marginal frequencies of NPOU were calculated as a function of prescription size from our model. Statistical significance was established at an α of 0.05 for all comparisons unless noted otherwise. All analyses were completed with SAS 9.4 (SAS Institute, Cary, NC). All statistical analyses were carried out between 5 January and 18 October 2020.
Additional analyses
Given the substantially different rates of postoperative readmission among patients with different vascular diseases and the fact that many open procedures have higher rates of readmission compared to their endovascular counterparts, we conducted a separate sensitivity analyses in which we excluded all patients who suffered a postoperative readmission within 180 days to investigate the degree to which the frequency of new persistent use was effected by postoperative readmission.
Results
Cohort characteristics.
NPOU, new persistent opioid use; SD, standard deviation; CCI, Charlson Comorbidity Index.
Univariate analysis
Patients with NPOU suffered postoperative complications at a higher frequency (5.2% vs. 2.5%, p < 0.001) and were more likely to fill an opioid prescription from 30 days preoperatively to 3 days postoperatively (51.6% vs. 24.1%, p < 0.001). Similarly, patients with NPOU underwent elective operations at a lower frequency compared to those who did not develop NPOU (82.9% vs. 90.3%, p < 0.001). Opioid prescriptions sizes (reported as equivalent number of 5 mg oxycodone tablets) were also higher among those developing NPOU (mean 23.4 ± 43.0 vs. 7.2 ± 22.0 tablets, p < 0.001). Mean age for the cohort was 75.0 ± 6.6 years and was comprised of 54.7% women. Significant differences were found in proportions of patients within racial groups, with Black patients comprising 11.1% of the NPOU group compared to only 5.3% of the group without NPOU (p < 0.001 for the comparison across all racial groups). New persistent users also had more medical comorbidities as well as mental health disorders such as anxiety (p < 0.001), mood (p < 0.001), or drug/substance use disorders (p = 0.008).
Multivariable analysis
Multivariable analysis of risk factors for new persistent use.
aOR, adjusted odds ratio; CI, confidence interval; CCI, Charlson Comorbidity Index.
Frequency of NPOU by surgical procedure
Patients undergoing peripheral arterial surgery, either endovascularly or via an open approach, had the highest frequency of NPOU of the procedures investigated, with an adjusted frequency of NPOU of 3.8% for endovascular peripheral vascular intervention and 3.5% for open infrainguinal arterial interventions. This contrasted with patients undergoing treatment for varicose vein disease, for which adjusted frequencies of NPOU were found to be 2.2% for endovenous procedures and 1.6% for open varicose vein procedures. When compared pair-wise, patients undergoing endovascular carotid or vertebral procedures had higher adjusted frequencies of NPOU than open carotid or vertebral operations (3.0%, 95% CI 2.4–3.6%, vs. 1.8%, 95% CI 1.6–2.0%, p < 0.001) as did patients undergoing endovenous procedures compared to patients undergoing open vein procedures (2.2%, 95% CI 2.0–2.4%, vs. 1.6%, 95% CI 1.1–2.1%, p = 0.019). These results are shown graphically in Figure 2. Adjusted frequencies of new persistent use by procedure and operative approach. Data represent adjusted rate with error bars representing 95% confidence intervals.
In an effort to investigate the degree to which our results were sensitive to readmissions in the postoperative period, we conducted sensitivity analyses excluding any patient who suffered a readmission within 180 days. Marginal frequencies of NPOU by procedure are shown graphically in Supplementary Figure 1. Of note, endovascular carotid/vertebral procedures continued to be significantly associated with NPOU compared to an open approach (p < 0.001), whereas no difference was found between endovascular and open approached for the treatment of aortic/iliac disease, infrainguinal arterial disease, or varicose veins.
Effect of prescription size
We calculated marginal frequencies of NPOU as a function of initial prescription size, which are reported in Figure 3. For patients receiving the equivalent of less than 10 tablets of 5 mg oxycodone, the adjusted frequency of NPOU was found to be 2.5% (95% CI 2.4–2.6%). Increasing prescription size was associated with higher frequencies of NPOU, with patients given a prescription equivalent to 100 tablets of 5 mg oxycodone having an adjusted NPOU frequency of 3.6% (95% CI 3.1–4.0%) and 200 tablets of 5 mg oxycodone an adjusted NPOU frequency of 5.0% (95% CI 3.8–6.3%). Adjusted frequency of new persistent use by initial prescription size. Data represent adjusted rate with error bars representing 95% confidence intervals.
Discussion
In this cohort study of Medicare beneficiaries, we found that patients undergoing vascular surgical procedures are at high risk of NPOU. Patients undergoing endovascular peripheral vascular interventions or open infrainguinal arterial interventions were at significantly higher risk compared to other vascular procedures, with those undergoing open varicose vein procedures being at lowest risk of the vascular procedures investigated. Treatment via an endovascular approach was not associated with a lower risk of NPOU relative to open surgical procedures but rather endovascular treatment of carotid/vertebral or varicose veins was associated with an increased risk of NPOU. Filling of a perioperative opioid prescription was found to be associated with significantly increased odds of NPOU development, and furthermore, prescription size showed a dose-response relationship, independent of other factors investigated. Importantly, frequencies of NPOU decreased over the study period. These results suggest that treatment with an endovascular approach may not result in substantial improvement in the risk of NPOU and furthermore suggests that other factors including the presence of wounds, underlying comorbidities, prescription patterns/sizes, or other unmeasured pain management strategies may be more important in the development of persistent opioid use disorders rather than surgical invasiveness.
A considerable amount of work continues to be done to address the substantial public health burden that the opioid epidemic poses to society, for which post-surgical prescription opioids remain an important aspect. These studies have suggested interventions such as prescribing recommendations and multimodal analgesic regimens, but the risk of development of NPOU following vascular procedures was previously not known.13–15 Given the variability of prescription sizes and the concern that patients treated with less invasive techniques may be lower risk, there is utility in understanding the burden of risk among vascular surgical patients specifically as well as the relationship of operative technique to risk of NPOU development.
This study expands on prior work in several ways. Previous studies have investigated postoperative NPOU risk among patients undergoing high volume, mainly outpatient operations with a relatively low burden of medical comorbidity and have almost exclusively investigated non-vascular patients. The results presented here describe NPOU risk among a population of patients with high baseline medical comorbidity and variable postoperative prescribing practices. Additionally, the availability of two treatment strategies with such discrepant levels of invasiveness for many vascular diseases provides a unique opportunity to evaluate invasiveness and its contribution to risk of opioid use disorders more broadly, which previously has only been investigated in the context of open versus laparoscopic surgery. Our data suggest that procedural invasiveness is not directly correlated with risk of NPOU but rather that patients undergoing some less invasive procedural modalities may actually have higher rates of NPOU. Risk of NPOU may not be particularly dependent on pain perceived around the time of the operation but rather more dependent on patient characteristics or other factors associated with treatment of postoperative pain. Furthermore, many endovascular procedures allow patients to stay in the hospital for a shorter period of time or not at all, which may cause patients to be discharged earlier in their pain recovery trajectory and result in lack of multimodal pain treatment options or in higher initial prescription sizes than they would have required later in their recovery. Additionally, non-steroidal anti-inflammatory medications may be contraindicated in a higher proportion of patients undergoing endovascular procedures which could limit non-narcotic options for pain control in this population. Finally, prior work has often focused on younger patients as they are represented in commercial claims data, whereas our study represents one of the first applications of these methods to an elderly population of Medicare patients.
The results of this study should be interpreted within the context of certain limitations. The primary limitation of this analysis lies in the relatively poor ability to control for certain clinical characteristics given the Medicare claims dataset. While we adjusted for medical comorbidities using the CCI, it may be that patients selected for endovascular procedures represent a population that is sicker at baseline in a way that is not reflected by comorbidities and that this may confound the relationship between surgical approach and NPOU. Furthermore, anatomic or physiologic characteristics of the patients’ vascular disease may be associated with both the way they utilize/fill opioid prescriptions postoperatively as well as the surgical approach. Unfortunately, these data are not available through traditional claims and therefore cannot be assessed in this analysis. Our data do not contain information on opioid consumption, but rather prescription fills, which may not directly correlate with consumption. That said, we utilized a definition for NPOU that includes prescription fills at multiple and varying timepoints throughout the postoperative course. We would not expect a substantial number of patients to refill multiple prescriptions if they were not consuming some if not most of the pills from the prior prescription. The requirement for a prescription fill more than 3 months from the date of discharge is well beyond when we would expect most patients to have had decreases in postoperative pain and therefore should minimize the risk of incorrectly classifying a patient as a new persistent user when they are rather simply treating ongoing post-surgical pain with opioid analgesics. While it may be possible that some prescriptions in the 180 days postoperatively are related to ongoing surgical pain or were given for the development of a postoperative complication, the results from our sensitivity analysis suggest that opioid prescriptions provided at the time of readmission likely do not represent a substantial proportion of opioid prescriptions within the postoperative period for those found to have NPOU. This study also utilizes data from Medicare fee-for-service beneficiaries and therefore does not represent patients who have alternative coverage or are too young to qualify for Medicare. In this way, the generalizability of these results to younger or non-Medicare populations is not established. We limited our analysis to only patients who had continuous Medicare Part D coverage, but it remains possible that a portion of these patients had alternative prescription drug coverage that would result in our analysis only capturing a fraction of opioid prescription fills.
Conclusions
In this cohort study, we found that NPOU was common following a variety of vascular surgical procedures and furthermore that less invasive operative techniques were not associated with a decreased risk but rather, in the case of carotid/vertebral or varicose vein procedures, was associated with an increased risk. Higher postoperative prescription size was associated with an increased risk of NPOU development. Leveraging these data, in coordination with patient reported outcomes data, may allow the development of postoperative prescribing guidelines as well as pain management strategies that minimize the risk of NPOU development while maintaining high patient satisfaction in this particularly high-risk group of patients undergoing vascular surgery procedures.
Supplemental Material
sj-pdf-1-vas-10.1177_17085381211024514 – Supplemental Material for Endovascular surgery is not protective against new persistent opioid use development compared to open vascular surgery
Supplemental Material, sj-pdf-1-vas-10.1177_17085381211024514 for Endovascular surgery is not protective against new persistent opioid use development compared to open vascular surgery by Craig S Brown, Nicholas H Osborne, Hsou M Hu, Dawn Coleman, Michael J Englesbe, Jennifer F Waljee, Chad M Brummett and Chandu Vemuri in Vascular
Footnotes
Acknowledgements
Presented at the American College of Surgeons Annual Meeting, 4 October 2020, Chicago IL, USA.
Disclosure
Dr Brummett is a consultant for Heron Therapeutics and Alosa Health.
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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by National Institute on Drug Abuse (NIH RO1 DA042859) and the Michigan Department of Health and Human Services (MDHHS, E20180672-00 Michigan DHHS - MA-2018 Master Agreement Program) as well as the Substance Abuse and Mental Health Administration (SAMHSA: E20180568-00 MA-2018 Master Agreement Program). Craig S Brown is supported by the Ruth L Kirschstein Postdoctoral Research Fellowship Award administered from the National Institute on Drug Abuse F32-DA050416.
Supplementary material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
