Association of opioid use disorder with outcomes of hospitalizations for acute myocardial infarction in the United States

Highlights • Patients with opioid use disorder experience greater in-hospital mortality and readmissions following MI.• Paradoxically, these patients experience superior cardiovascular outcomes compared to their counterparts.• Opioid use disorder remains associated with high resource utilization and healthcare expenditures.


Introduction
The misuse of opioids in the United States is a growing crisis with nearly 47,600 attributable deaths in 2017. 1 The widespread prevalence of Opioid Use Disorder (OUD) has previously been linked to several structural determinants, including socioeconomic status, race, economic opportunity, and social cohesion. 2 A recent retrospective study found that cardiovascular disease is the second most common cause of mortality in patients with OUD, second only to direct drug poisoning. 3 Yet, associations between cardiovascular outcomes and OUD are underexamined.
The effects of chronic opiate exposure on the cardiovascular system remain controversial. Several in vivo and in vitro studies have demonstrated the cardioprotective and anesthetic preconditioning effects of opioids, mediated primarily by delta receptor activation and protein kinase C transduction pathways. 4−6 In fact, Marmor reported OUD to be associated with a lower risk of significant Coronary Artery Disease (CAD) 7 . Conversely, several recent clinical and population studies have challenged this association, showing a significantly increased risk of Acute Myocardial Infarction (AMI) and arrhythmia in patients with prolonged opioid use, seen in as few as 6 months. 6,[8][9][10] In a study of 2253 propensity-matched pairs, Ranka and coworkers noted OUD to be associated with increased complications but not mortality after an AMI. 11 However, this study suffers from several limitations, including a lack of information on associated hospitalization costs and readmissions.
In the present work, the authors examined the clinical and financial outcomes of OUD in a national cohort of AMI hospitalizations. The authors hypothesized OUD to be independently associated with a greater risk of complications but not in-hospital mortality. The authors further hypothesized increased readmissions within 30 days of index discharge and hospitalization costs among the OUD cohort.

Methods
All adult (≥ 18 years) non-elective admissions for AMI, classified as ST-Segment Elevation (STEMI) or non-ST-Segment Elevation Myocardial Infarction (NSTEMI) were abstracted from the 2016-2019 Nationwide Readmissions Database (NRD). The NRD is the largest all-payer readmissions database in the US and provides accurate estimates for ∼60% of all hospitalizations. 12 The NRD is maintained by the Agency for Healthcare Research and Quality's Healthcare Cost and Utilization Project. Unique patient identifiers allow for tracking of re-hospitalizations within the state over each calendar year. International Classification of Disease, Tenth Revision (ICD-10) codes (Supplemental Table 1) were used to identify relevant diagnoses while additional patient and hospital characteristics were defined according to the NRD data dictionary. This included age, sex, income quartile, primary insurance, hospital teaching status, and bed size. Records missing data for age, in-hospital mortality, and costs were excluded (0.46%). Those with a primary diagnosis of infective endocarditis or type 2 AMI were similarly excluded (14.3%).
The primary outcome of interest was in-hospital mortality and cardiac complications, which included cardiogenic shock, ventricular tachycardia/fibrillation, cardiac arrest, acute heart failure, and other cardiac complications. Other cardiac complications were considered as a composite of cardiac tamponade, pericarditis, hemopericardium, atrial and ventricular septal defect, chordae tendinea rupture, papillary muscle rupture, angina, intracardiac thrombus, and ventricular wall rupture. Secondary outcomes included respiratory, infectious, thromboembolic, acute kidney injury, and neurologic complications, and key markers of resource use. Specifically, the authors considered Length Of Stay (LOS), hospitalization costs, non-home discharge (defined as discharge to a skilled nursing facility, acute care facility, intermediate care facility, or against medical advice), and 30-day non-elective readmissions. Hospitalization costs were converted from charges using center-specific costto-charge ratios and adjusted for inflation to the 2019 Personal Health Index. Diagnosis-related groups were used to identify admitting diagnoses upon re-hospitalization.
Categorical variables are reported as proportions while normally distributed continuous variables are shown as means with standard deviation. Non-normally distributed continuous variables are reported as median and interquartile range. The significance of intergroup differences of continuous variables was assessed using the adjusted Wald and Mann Whitney U tests, as appropriate. Pearson's Chi-Square test was used to compare categorical variables. The authors used Cuzick's nonparametric test to assess the significance of temporal trends in the study (NPtrend). 13 Prior to regression, entropy balancing was used to obtain a weighted comparison group with comparable covariate distributions. Entropy balancing has several advantages over traditional propensity score matching and preserves the entire cohort for further analysis. 14 Model covariates were selected based on clinical relevance. Subsequent multivariable mixed regressions were developed to examine the association of OUD with described outcomes of interest. Models were adjusted for age, sex, and concomitant substance use disorder, among other factors (Supplemental Table 2). Models were Optimized Using Receiver-Operating Characteristics (AUROC) as well as Akaike and Bayesian Information Criteria to reduce overfitting. Regression outputs are reported as Adjusted Odds Ratios (AOR) or Beta (β) coefficients, both with 95% Confidence Intervals (95% CI). An α less than 0.05 was considered significant. Stata 16.1 (StataCorp, College Station, TX) was used for all statistical analyses. Given the deidentified nature of the data, this study was deemed exempt from full review by the Institutional Review Board at the University of California, Los Angeles. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline was followed for this study.

Results
Of an estimated 3,318,257 hospitalizations for AMI included for analysis, 36,057 (1.1%) had a concomitant diagnosis of OUD. The prevalence of OUD modestly decreased from 11.0 per 1,000 hospitalizations in 2016 to 10.5 per 1,000 hospitalizations in 2019 (NPtrend < 0.01).

Discussion
In the context of the nearly five-fold increase in US medical hospitalizations for OUD over the past two decades, 15 the present study examined outcomes of 36,057 patients with OUD admitted with AMI. Although OUD patients were younger, they had a higher burden of Other Payer includes self-pay, no charge, and other payer, as defined by the NRD.  several chronic medical conditions, but no difference in mortality compared to others. However, patients with OUD experienced lower rates of cardiovascular complications and procedural intervention at index hospitalization. Nonetheless, OUD appears associated with several indicators of increased healthcare expenditures, including in-patient costs, LOS, non-home discharge, and 30-day readmissions. Several of these findings warrant further discussion. As expected, given their younger age at presentation, patients with OUD less frequently had age-related diagnoses of hypertension, diabetes, and peripheral vascular disease. Conversely, they had a higher burden of chronic lung and liver diseases. Such comorbidities may be sequelae of chronic polysubstance use, 16 which was present in nearly 75% of OUD patients analyzed. Although the authors were unable to examine the extent of coronary artery disease in the present study, OUD patients less frequently presented with STEMI and required fewer revascularization procedures compared to non-OUD patients, suggesting a less severe AMI course. The relative absence of other major age-related comorbidities may explain the reduced cardiovascular complications among OUD patients.
Animal studies have generated hypotheses that could explain why opioids may be paradoxically cardioprotective. Opiates have been implicated in Ischemic Preconditioning (IPC) in animal models. IPC is a process by which brief periods of coronary artery occlusion and reperfusion activate opioid receptors δ and μ and stimulate signal transduction via  protein kinase C. 17 This has been found to reduce the size of infarcted myocardial tissue and mitigate the resultant ischemic injury in isolated rat hearts. 18−20 Since these effects have been primarily described in animal models, the true effects of long-term opioid exposure on the conditioning of the cardiovascular system remain to be elucidated in OUD patients. Nonetheless, acute opioid administration (intravenous morphine sulfate) has been the recommended treatment for cardiac pain refractory to nitroglycerine. 21 Further studies are needed to clarify the mechanism behind the apparent cardioprotective effects of opioids. Those with OUD presented with AMI at a much younger age, potentially suggesting a greater burden of atherosclerosis. While the younger presentation of AMI in OUD patients has not been widely studied, several investigators have noted the extent of opioid use to correlate directly with the degree of coronary artery disease, hyperlipidemia, 22,23 and AMI rates. 8,24 OUD has been linked to poor nutritional status, caused by both metabolic and lifestyle derangements. 25 Specifically, the xerostomia and appetite suppression caused by opioids predisposes those with OUD to have diets high in fat and sugar. 25,26 Furthermore, Strike et. al. found that over 50% of intravenous drug users experience food insecurity, defined as the systematic and chronic lack of access to nutritional food. 27 Although not mechanistic, these findings may, in part, explain the accelerated presentation of AMI among opioid users.
As noted above, OUD patients had higher odds of respiratory, renal, and infectious complications. This may contribute to the longer and more expensive hospitalizations the authors found to be associated with OUD. In a study of 56,278 patients with OUD, Clark and colleagues found that the degree of comorbidities experienced by OUD patients raises the cost of their hospitalizations. 28 As the annual opioid-related healthcare expenditures in the US reach an estimated $89.1 billion, mitigating costs and resource use for OUD patients is increasingly relevant. 29 Furthermore, OUD patients were more likely to be readmitted within 30 days, which is consistent across various elective and emergent hospitalizations. 30−32 Similar to prior literature on AMI, 33 cardiovascular reasons comprised a majority of all readmissions across the study population. However, OUD patients were less likely than their counterparts to have a cardiac-related readmission diagnosis. The authors noted OUD patients to be more frequently readmitted for respiratory and psychiatric diagnoses. Known complications of prolonged opioid use include opioid-induced respiratory depression and comorbid psychiatric disorders, as previously reported. 16−34 Consequently, the present findings may indicate continued post-discharge opioid use, possibly resulting from suboptimal addiction counseling at the time of index hospitalization. A recent randomized controlled trial found that patient navigation services increased the OUD patients' likelihood of adhering to medical treatment and finding resources to meet their social needs. 35 Schoenfield and colleagues outlined barriers to accessing OUD treatment, which includes physician inexperience, stigma of substance use, and a lack of community resources to locally implement best practices. 36 Multifaceted approaches to increase social services and treat OUD at index hospitalization could minimize these preventable readmissions.
The present study has several limitations inherent to its retrospective nature. The accuracy of the data is subject to potential coding errors of the NRD. As an administrative database, the NRD lacks detailed data on race, geographic region, and other demographics that are important predictive factors of OUD and associated outcomes. Furthermore, the NRD does not track patients who are readmitted in a state different from that of the index hospitalization. Therefore, it is likely that the burden of OUD among AMI hospitalizations and associated readmissions have been under-reported. Although the authors were able to adjust for various patient and hospital characteristics using validated diagnostic coding, the authors were unable to adjust for important factors that may have influenced key outcomes, due to the limited nature of the NRD. In particular, this study lacks patient-level data on medications, compliance, and extent of opioid use.

Conclusion
In conclusion, OUD patients admitted for AMI presented with lower severity and had fewer procedural interventions with no change in mortality. However, they continue to have increased LOS, hospitalization costs, non-home discharge, and 30-day non-elective readmission. This may be the result of increased non-age-related comorbid conditions, persistent drug use, and socioeconomic disparities in spite of the potential protective effect of younger age. The present findings underscore the need for improved access to OUD treatment, as well as directed strategies to address financial and healthcare inequities.

Funding
The authors have no financial disclosures or conflicts of interest to report.

Conflicts of interest
The authors declare no conflicts of interest.