1. IntroductionAccording to the Centers for Disease Control, it is estimated that national opioid prescription in the US has tripled since 1999, and the economic cost of prescription opioid misuse exceeds 78 billion dollars annually [1]. In 2015, 30,000 people died of an opioid drug overdose in the US [2] while 67,400 died in 2018 [3]. Most opioid abuse originates from the misuse of prescription medication used to treat chronic pain [4,5]. Alarmingly, persistent postoperative pain occurs in as many as 10–50% of surgical patients [4], which puts these patients at high risk of developing opioid medication-related adverse events with additional healthcare costs [6].Pain from craniotomy surgeries is commonly managed intraoperatively with fentanyl, remifentanil, or sufentanil infusions, intravenous fentanyl, morphine hydromorphone, and acetaminophen, as well as with the use of scalp blocks and infiltration of a local anesthetic at the craniotomy incision site. Even though it has been historically reported that post-craniotomy pain may be less severe compared to that after non-craniotomy procedures [7], several studies have demonstrated that acute post-craniotomy headaches can be severe and occur in as many as one-third to half of postoperative patients [8,9,10,11]. Some reasons for persistent post-craniotomy pain may include dural irritation, pericranial muscle retraction, surgical trauma, decreased cerebrospinal fluid pressure, persistent tension headache, and neck muscle spasm related to the patient’s position during the surgical intervention, and aseptic meningitis [5,12]. The location of surgical intervention may also be related to increased postoperative pain, with infratentorial surgeries having higher pain scores than supratentorial surgeries [13]. Despite using non-opioid analgesic techniques such as local anesthetic skin infiltration and/or scalp blocks and non-opioid medications, opioids remain the mainstay therapy for post-craniotomy headaches. While prior research has focused on intraoperative anesthetic management and pain experiences during their hospital stay [14,15,16,17], there is no report of the rate of opioid prescription at the time of hospital discharge, which may be associated with various patient and procedure-related factors.

To bridge this knowledge gap, we undertook a study to examine the prevalence and factors associated with opioid prescriptions in patients at the time of hospital discharge. We attempted to examine the continued prescription of opioids at the neurosurgical clinic follow-up. We studied this in a cohort of patients undergoing craniotomy for tumor resection/vascular repair. The specific aims of this study were (1) to examine the prevalence of opioid prescription at the time of hospital discharge and neurosurgical clinical follow-up within 30 days of the neurosurgical procedure and after 30 days of the neurosurgical procedure and (2) to examine the association between perioperative factors associated with the rate of opioid prescriptions. Knowledge gained from this study may increase awareness of opiate prescriptions amongst healthcare providers routinely caring for patients undergoing craniotomy, which in turn may inform responsible and appropriate discharge opioid prescription practices.

2. Materials and Methods 2.1. Study Design

This is a single-center, retrospective observational study conducted at Harborview Medical Center in Seattle, Washington, a comprehensive stroke, and Level I trauma center.

2.2. Patient Population

We included patients 18 years of age and older who underwent elective craniotomy by a single neurosurgeon (Author LS) at Harborview Medical Center between January 2015 and December 2019. We included craniotomy for tumor resections and vascular repair (microsurgical aneurysm repair, resection of arteriovenous malformation, and intracranial/extracranial bypass). Exclusion criteria included non-intracranial surgeries, emergency surgeries, more than one neurosurgical procedure during the same surgical encounter, expired patients, and patients with pain not located in the head/neck region.

2.3. Institutional Practice

All post-craniotomy patients were admitted to an intensive care unit, regardless of extubation status. None of the patients received scalp nerve blocks perioperatively. The pain was assessed by bedside nursing staff with the Critical-care Pain Observation Tool (CPOT) in intubated patients and by a numerical scale (range 0–10) in communicative patients.

2.4. Data Collection

Patient information and specific data are collected from the electronic health record (EHR) and refined via a departmental database, Perioperative & Pain Initiatives in Quality Safety Outcome (PPIQSO).

We categorized the variables of interest into three groups:

(a) Preoperative: age, sex, race category (White/non-White), ethnicity (Hispanic or Latino/Non-Hispanic/Non-Latino), Language category (English-speaking/Non-English speaking), insurance (Commercial/Non-commercial), history of chronic pain, pre-admission use of opiates, and oral morphine equivalent (OME), history of any depression/anxiety disorder, and a history of substance abuse.

(b) Intraoperative: type of procedure (tumor/vascular), location (supratentorial/infratentorial), surgical duration, use of remifentanil infusion, fentanyl infusion, lidocaine infusion, methadone, hydromorphone, ketamine infusion, and non-opioid analgesic such as intravenous acetaminophen.

(c) Post-operative: first 24 h post-op OME, first 24 h post-op maximum pain scores, maximum pain score during the hospital stays, daily OME from postoperative day one to postoperative day 30, total hospital OME, hospital OME/per day of stay in the hospital, and hospital length of stay.

(d) Follow-up: timing of visits in days since the original neurosurgical procedure and opiate prescriptions.

2.5. Outcome Measures

Opiates at discharge, OME at discharge, and Opiates prescribed at follow-up (within 30 days and 30 days after the original neurosurgical procedure). Total discharge OME and discharge OME per day is inferred from the number and dose of tablets and the administration frequency in the prescription. The OME conversion was as follows: oral morphine: conversion factor 1, oral controlled release Morphine: 1, oral codeine: 0.15, oral hydromorphone: 4, oral hydrocodone: 1, oral oxycodone: 1.5, oral oxymorphone: 3, oral meperidine: 0.1, oral levorphanol: 15, oral tramadol: 0.25, oral tapentadol: 0.3, transdermal fentanyl: (mcg/h): 2.4, oral methadone (1–20 mg/day): 4, oral methadone (21–40 mg/day): 8, oral methadone (41–60 mg/day): 10, oral methadone (>61–80 mg/day): 12, sublingual buprenorphine: 30.

2.6. Data Analysis and Statistics

The descriptive analysis described the study sample characteristics. After normality testing using the Shapiro–Wilk test, continuous variables such as age, body mass index (BMI), 24 h OME, hospital OME, OME at discharge, and hospital length of stay were expressed as median [interquartile range, IQR]. Categorical variables were expressed as counts and percentages.

Univariable analysis to examine associations with opiate prescription at hospital discharge was performed using the following parameters: age, sex, ASA physical class, insurance type, type of case (tumor/vascular), location (supratentorial/infratentorial), race category (non-white/white), ethnicity category (declined/unknown/Hispanic or Latino/not Hispanic or Latino, language category (English/non-English), history of chronic pain, history of a depression/anxiety disorder, history of substance abuse, preoperative opioid use, preoperative OME, length of surgical procedure, intraoperative ketamine infusion, hospital OME, hospital LOS in days: 1–4 days/5–12 days/>12 days. Multivariable analysis was performed using the following factors: intraoperative ketamine infusion, preoperative opioid use, and hospital LOS (days): 1–4 days/5–12 days/>12 days. Hosmer-Lemeshow goodness-of-fit test was performed on the multivariable model. Comparisons between patients receiving opioids at the time of hospital discharge and those that did not were performed using the Wilcoxon test for continuous variables and the Chi-square test for categorical variables. Odds ratio (OR) and 95% confidence intervals (CI) were calculated to examine the odds of receiving opioids at the time of discharge. A linear regression model examined the correlation between the first-24 h OME, daily OME, hospital LOS and discharged OME use. We also examined the association between discharge OME age, sex, preoperative opioid use, history of psychiatric disorder, history of chronic pain, and the use of ketamine infusion intraoperatively. A Bonferroni-corrected p-value of 18]/RStudio version 1.554 [19] was used for statistical analysis. 4. Discussion

We examined the opioid burden and the prescription of opiates for head and neck pain at the time of hospital discharge after elective craniotomy for tumor resection/vascular procedures. The main findings of the study are described and discussed below.

Our study finds that seven out of ten patients are prescribed opioids at the time of their discharge from the hospital. Those with a history of pre-operative chronic pain were more likely to be prescribed opioids at discharge. In contrast, those with more extended hospital LOS were less likely to be prescribed opioids at discharge. Practitioners were more likely to prescribe patients with pre-existent chronic pain opioid pain medications on discharge since the presence and severity of pre-existing pain may correlate with the intensity of acute postoperative pain [20]. Given the lack of association between hospital OME and whether the patient is discharged with an opioid, there is a high likelihood that this trend may stem from medical practitioner perception rather than patient condition as a primary driver of opioid prescribing practices. For example, Brandal et al. looked at colorectal surgery. They found that 70% of opioid-naive patients with below-average opioid use during their hospital stay and low pain scores on discharge still received an opioid prescription upon discharge [21]. Since there is no institutional protocol with respect to the prescribing OME, the dose prescribed may very well be influenced by the subjective opinion of the treating physician, a finding observed in the neurocritical care setting [22]. Our study finds that opioid use is common in the postoperative period [23]. While in a small sample study, opioid-free craniotomy [24] was found to be non-inferior to traditional pain control concerning pain scores in the postoperative period, opioid-free post-craniotomy pain management remains elusive and warrants further examination. Only a minority of patients remained “opioid-free” during their hospital stay, given the “severe” pain cited by a substantial proportion of patients during their post-craniotomy period. Some patients may remain opioid-free, and others require a high OME rate. Expectations of pain and pain-control/pain-free state, variability in surgical techniques, location of the surgical site, and institutional pain regimens for an intraoperative and postoperative period may affect the frequency and severity of OME use.

With regard to the amount of opioid pain medications with which patients were discharged among those who were prescribed any after their craniotomy procedure, we found a statistically significant linear relationship between discharge OME and the first 24 h OME, daily OME, and total hospital OME. This may explain one of the reasons providers may base the amount of discharge OME to be prescribed. Though rich in risk factors for postoperative pain and the development of chronic pain, literature is scarce on factors associated with the amount of opioid medications patients are discharged with.

One observation that was not expected is the association between hospital length of stay and the rate of opioid prescription. The highest risk for opiate prescription in patients discharged within the first four days of their craniotomy, and its decrease with longer hospital length of stay suggest an association between pain scores and hospital OME, both of which decrease over time as the patient is observed after craniotomy. This would also be congruent with current studies, which showed that severe pain is commonly experienced within the first 48 h after surgery [13], presumably leading to an increased need for opioid pain medication, whether inpatient or upon discharge. In addition, when patients are discharged early, providers may fear that they will continue to have pain after leaving the hospital and are more likely to prescribe opioids on discharge. The literature has no comparable evidence regarding the association between LOS and opioid prescription in craniotomy cases. A study on abdominal transplant surgery found that LOS was protective against opioid over-prescription, likely because the initial postoperative opioid-requiring phase and/or acute complications have subsided [25]. On the other hand, other studies found that the longer the patients stay in the hospital after spine surgery [26], or wrist and ankle surgery [27], the more likely they are to be prescribed opioids at discharge.Findings from our study suggest that hospital discharge opioid prescription was not associated with age, sex, prior opioid use, surgical location and approach, and history of depression and/or anxiety, which were thought to predispose patients to postoperative pain and opioid use [5,13,20,28]. Specifically, pre-existing opioid use, female gender, and young age were shown to be associated with generally more postoperative opioid use [5,20]. However, a study with a small cohort found that the male (rather than female) gender can be associated with increased postoperative pain [28]. In craniotomy surgeries, infratentorial approaches are associated with more postoperative pain than supratentorial approaches [13]. The parameter of ‘pre-existing opioid use’ in some of those studies referred to chronic pain, which would make our finding consistent with existing literature. Conversely, suppose a patient was on a short preoperative course of opioid pain medication for pain related to their intracranial pathology. In that case, the surgery should theoretically relieve the patient of the cause of their pain and therefore make prolonged postoperative pain unlikely.At this time, opioid-free craniotomy care remains elusive [21,29,30], yet aspirational. To achieve an opioid-free state, there needs to be a shared mental model of understanding between the patient, the anesthesiologist, the neurosurgeon, and all postoperative clinicians. While alarmed by the opioid epidemic, we would be remiss to look at this as a potential pathway to craniotomy care. Recent studies published regarding the use of non-opiates in the routine management of craniotomy care indicate that while more data is awaited, perioperative clinicians have been interested in this.This study puts into perspective the use of opioids after elective craniotomy and highlights possible reasons for prescribing opioids at the time of discharge. The findings of hospital LOS being an important factor raises concerns since there is growing interest in early discharge from the hospital after craniotomy [31,32]. Suppose patients are discharged earlier than historical controls. In that case, this may lead to an increased rate of opioid prescription due to the higher severity of pain and OME use in the first few days after craniotomy. Practitioners who are routinely involved in the care of these patients must balance the risks and benefits of prescribing opioids and the total amount of opiates prescribed for the post-discharge period. The findings of this study may be helpful in educating practitioners as well as patients. LimitationsLimitations are typical of a single-center retrospective study in that the data used for this study was not collected for research purposes, the sample size and power analysis were not performed a priori, and the follow-up data is missing more than half of the original sample size. If a patient received opioid-based patient-controlled analgesia during their postoperative period, there were institutional limitations regarding the accuracy of the recorded dosage. In addition, we could not assess if patients self-administered the opioids they were prescribed at discharge. We did not study the impact of patient positioning or tumor grading [33] on the rate of opioid prescription, nor can we say if prescription patterns for opioids may be same in a US-based vs. non-US-based sample where the prescription standards may vary. The study’s strengths are that we created a homogenous group of patients operated on by only one neurosurgeon, thus trying to reduce variability in surgical exposure and its potential effect on postoperative pain. We also restricted our study to head and neck pain, which, while excluding 20% from our original cohort, the final study sample was reflective of pain related to the surgical procedure. We also excluded those patients who had a non-neurosurgical procedure not to confound the potential reasons for using opiates for non-head and neck pain.

Being a retrospective study, findings are hypothesis-generating, and conclusions cannot be generalized to other institutions; however, the results from our research warrant a multicenter investigation in a larger cohort of craniotomy patients.