Chronic post‐operative opioid use after open cardiac surgery: A Danish population‐based cohort study

Abstract Background Knowledge of chronic opioid use after cardiac surgery is sparse. We therefore aimed to describe the proportion of new chronic post‐operative opioid use after open cardiac surgery. Methods We used prospectively registered data from a national prescription registry and a clinical registry of 29 815 first‐time cardiac surgeries from three Danish university hospitals. Data collection spanned from 2003 to 2016. The main outcome was chronic post‐operative opioid use, defined as at least one opioid dispensing in the fourth post‐operative quarter. Data were assessed for patient‐level predictors of chronic post‐operative opioid use, including pre‐operative opioid use, opioid use at discharge, comorbidities, and procedural related variables. Results The overall proportion of post‐operative opioid use was 10.6% (95% CI: 10.2‐10.9). The proportion of new chronic post‐operative opioid use was 5.7% (95% CI: 5.5‐6.0) among pre‐operative opioid naïve patients. The corresponding proportions among patients, who pre‐operatively used low or high dose opioid (1‐500 mg or > 500 mg cumulative morphine equivalent opioid), were 68.3% (95% CI: 66.1‐70.4) and 76.3% (95% CI: 74.0‐78.5) respectively. Risk factors associated with new chronic post‐operative opioid use included: female gender, underweight and obesity, pre‐operative comorbidities, acute surgery, ICU‐time > 1 day, and post‐operative complications. Strongest predictor of chronic post‐operative opioid use was post‐discharge use of opioid within one month after surgery (odds ratio 3.3, 95% CI: 2.8‐4.0). Conclusion New chronic post‐operative opioid use after open cardiac surgery is common. Focus on post‐discharge opioid use may help clinicians to reduce rates of new chronic opioid users.


| INTRODUC TI ON
Opioid use is a serious healthcare concern globally as well as in Denmark. [1][2][3] Denmark has one of the highest rates of opioid consumption in the world with 3%-5% of all Danes using opioid regularly. 4 This is concerning as chronic opioid use is associated with increased risk of opioid-related morbidity and mortality. [5][6][7] As one common indication for opioid prescriptions, special attention should be devoted toward opioid prescribing following surgical procedures. 8 In Denmark, approximately 4000 cardiac surgeries involving sternotomy are performed annually. 9 The use of intraoperative opioid treatment following cardiac surgery is necessary to alleviate pain caused by artery harvesting, rib retraction, sternotomy, and drain tubes. 10 Effective intraoperative pain management may reduce post-operative complications such as myocardial infarction, cardiac arrhythmias, hypercoagulability, and pulmonary complications. 10,11 While opioids may be necessary to control pain in the intraoperative period, studies suggest that the harms of opioid use after hospital discharge may outweigh the benefits, as post-operative opioid use has been associated with chronic opioid use, misuse, opioid use disorder, overdose, and cognitive dysfunction. [12][13][14][15] In fact, several studies show that 3-10% of patients who did not take opioids before surgery develop a chronic opioid use after surgery. [16][17][18] Several risk factors for chronic post-operative opioid use have been identified including male gender, older age, medical and psychiatric comorbidities, concomitant benzodiazepine use, and type and extent of surgery, among others. 16,17,19,20 Although previous studies have examined the associations between various types of surgeries and long-term opioid use, the knowledge of chronic post-operative opioid use after cardiac surgery is sparse.
We aimed to examine patterns of opioid use within the first year after open cardiac surgery and estimate the risk of chronic post-operative opioid use according to pre-, intra-, and post-operative characteristics.

| Study design and setting
This population-based cohort study used prospectively collected data from medical and administrative registries in Western Denmark.
Western Denmark has a population of approximately 3.3 million people corresponding to 55% of the total Danish population. 21 Danish citizens are assigned a unique 10-digit Civil Personal Register number allowing linkage of Danish registries on an individual level. 22 The study was approved by the Danish Data Protection Agency

| Data sources
Patients were identified from the Western Denmark Heart Registry (WDHR)-a mandatory internet-based clinical registry covering all adult patients undergoing cardiac surgery in Western Denmark.
Detailed patient, surgery, anesthesia, and intensive care data are prospectively collected on consecutive patients. 21,23 Data on opioid dispensing were retrieved from the National Prescription Registry (NPR)-a registry containing detailed nationwide information about patient, dispensing, prescriber, and pharmacy information on redeemed prescriptions since 1995. 24 Data in both WDHR and NPR are considered both complete and valid. 21,23,25 Baseline characteristics were identified from the WDHR using EuroSCORE I 15

and 2013
EuroSCORE II. 16 The specific definitions of the Euroscore I and II can be seen elsewhere. 15;16 The Danish Health Service provides universal tax-supported healthcare, guaranteeing all Danish residents free access to general practitioners and hospitals. In Denmark, general practitioners issue most prescriptions. Patient co-payment is required for prescription drugs. Any reimbursable medicines are covered by a tax-financed drug reimbursement scheme. 26

| Study population
The study included all first-time cardiac surgical procedures conducted in the three university hospitals in Western Denmark from Prior to study-time restriction, patients with missing data on time of surgery (n = 7) and patients with an invalid Central Personal Registration number were excluded (n = 223). After study-time restriction, all non-first-time procedures, or if the surgical procedure did not involve sternotomy or was a heart transplantation were excluded ( Figure 1).
All patients completed the 1-year follow-up. Missing data on pre-, intra-, and post-operative characteristics ranged from 2.5% to 8.4%.

| Outcome
Opioid dispensings were identified from 6 months before surgery to 12 months after surgery. We defined the primary outcome "chronic post-operative opioid use" a priori before data extraction. We divided opioid use into four 3-month periods (quarters) after surgery and defined opioid use as at least one opioid dispensing in one of these quarters. We defined chronic post-operative opioid use as at least one opioid dispensing in the fourth post-operative quarter. We defined late initiators as individuals with a first opioid dispensing later than three months after surgery.

| Covariates
Included covariates were defined a priori before data extraction.
Pre-operative opioid use was included as a possible risk factor for chronic post-operative opioid use. Pre-operative opioid use was defined as at least one opioid dispensing within 6 months before cardiac surgery. We further divided pre-operative opioid use into low (≤500 mg) or high dose (>500 mg) based on cumulative morphine equivalent opioid use within 6 months before cardiac

| Statistical analyses
Proportions and numbers are used to describe the study population stratified by opioid naïve patients and opioid users before surgery. Categorical data are presented as numbers and percentages (%) with an estimated 95% confidence interval (CI). Comparisons of categorical data were made using a chi-squared test and results with a P-value < .05 were considered significant. Continuous data are presented as means with a 95% CI for normally distributed data and as medians with interquartile ranges for skewed data.
Patients had to be alive on the last day of each of the four postoperative quarters to be included in the calculation for that particular period. Logistic regression was applied to estimate odds ratios (OR) and 95% CI of chronic post-operative opioid use according to pre-, intra-, and post-operative patient characteristics.
ORs were stratified according to pre-operative opioid use in order to minimize any risk of effect modification. Crude estimates with 95% CIs were reported. All statistical analyses were conducted in SAS version 9.4.  (Table 1). Pre-operative opioid users were more likely to be female, to be obese, or to have a diagnosis of chronic obstructive pulmonary disease, peripheral artery disease, diabetes, or pulmonary hypertension (all P < .01). Less pre-operative opioid users were alive and the end of the fourth post-operative quarter (90% compared to 94% of pre-operative opioid naïve patients). Pre-operative opioid users were more likely to have acute surgery, to have an ICUtime > 1 day, or to be under extracorporeal circulation during the operation (all P < .01). Pre-operative opioid users had a higher frequency of post-operative myocardial infarction (P = .04), renal replacement therapy (0 < 0.01), and sternal wound infection (P < .01).

| Chronic post-operative opioid use
The overall proportion of chronic post-operative opioid use was 10.6%  (Table A3 in the appendix).

| Risk factors for chronic post-operative opioid use
Pre-operative opioid use was highly associated with chronic post- post-discharge OME use of more than 500 mg was highly associated

| Opioid use distribution
In

TA B L E 1 (Continued)
chronic post-operative opioid use based on prescription dispensing according to the NPR may represent a more accurate measure of medication intake, even in the presence of some misclassification. 30 The proportion of chronic post-operative opioid users did not change during the study period. However, a decrease in opioid dispensing in the first post-operative quarter was observed.   patients with intraoperative prescription size of more than 450 OME with a threshold effect between 300-450 OME and > 450 OME. 28 Our findings support their conclusion that post-discharge opioid prescription is a modifiable predictor of new chronic post-operative opioid use, and clinicians should certainly avoid prescribing more than 60 tablets of 5 mg Oxycodone for full sternotomy. However, more elaborated, patient-tailored and evidence-based prescribing guidelines are needed to respond to the risk of inducing new chronic post-operative opioid users among adult cardiac surgery patients.

| Strengths and weaknesses
A major strength of the present study is its multi-institutional study design covering an unselected patient population with free access to a universal healthcare system and with individual patient-data linkage. Furthermore, prospective data collection in the WDHR and complete patient follow-up minimized the risk of selection bias. The present study is based on prescription dispensings rather than on issued prescriptions, which provides an important advantage as a prescription dispensing is a better surrogate marker for actual drug intake than a written prescription. 24 Another important feature of the NPR is its inclusion of drugs used by nursing home residents, which limits differential misclassification of exposure status due to frailty among elderly individuals. 7 A limitation of this study is the lack of data on indication for opioid use. Thus, it is likely that some late initiators took opioids for pain conditions unrelated to the cardiac surgical procedure. Further limitations include unavailability of data on intended duration, and whether the drug was taken as recommended by the prescriber.
Another limitation is the unavailability of data on inpatient opioid use and opioids dispensed in hospitals.