Article Text
Abstract
Introduction The reported use of cannabis within surgical population is increasing. Cannabis use is potentially associated with increased harms and varied effects on pain control. These have important implications to perioperative care.
Methods We conducted a retrospective cohort study comparing surgical patients reporting cannabis use preoperatively to control patients with no cannabis exposure, in a 1:2 ratio. To control for confounding, we used a propensity score-matched analysis to assess the adjusted association between cannabis use and study outcomes. Our primary outcome was a composite of (1) respiratory arrest or cardiac arrest, (2) intensive care admission, (3) stroke, (4) myocardial infarction and (5) mortality during this hospital stay. Secondarily, we assessed the effects on pain control, opioid usage, induction agent dose and nausea-vomiting.
Results Between January 2018 and March 2019, we captured 1818 patients consisting of cannabis users (606) and controls (1212). For propensity score-matched analyses, 524 cannabis patients were compared with 1152 control patients. No difference in the incidence of composite outcome was observed (OR 1.06, 95% CI 0.23 to 3.98). Although a higher incidence of arrhythmias (2.7% vs 1.6%) and decreased incidence of nausea-vomiting needing treatment (9.6% vs 12.6%) was observed with cannabis users vs controls, results were not statistically significant. No significant differences were observed with other secondary outcomes.
Conclusion Our results do not demonstrate a convincing association between self-reported cannabis use and major surgical outcomes or pain management. Perioperative decisions should be made based on considerations of dose, duration, and indication.
- outcomes
- analgesia
- postoperative complications
Data availability statement
Data are available on reasonable request from the corresponding author.
Statistics from Altmetric.com
Introduction
Rates of cannabis use have increased across several parts of the world and this may be attributable to increased use or reporting. Recent Canadian statistics indicate a prevalence of 18% of any cannabis use in individuals >15 years in the first quarter of 2019, compared with 14.8% in 2017.1 The 2019 United Nations World Drug Report identifies cannabis as the most commonly used recreational substance.2 We also observe that cannabis use continues to be on the rise for chronic pain3 and other indications,4 5 despite uncertain evidence.
Both plant-derived cannabinoids (phytocannabinoids) and synthetic cannabinoids have been considered for medicinal use. Crude cannabis contains >750 identified constituents,6 with the most well-known being tetrahydrocannabinol (THC) and cannabidiol (CBD). They act via the endocannabinoid system, notably via CB1 and CB2 receptors that are located on nerve terminals and immune cells.7 Animal studies have indicated the potential for peripheral, spinal and supraspinal mechanisms of pain modulation with cannabinoids.8 Apart from modulating neurotransmitter release, THC can interact with gamma aminobutyric acid, norepinephrine, and acetylcholine release and modulate actions on opioid and N-Methyl-D-aspartic acid receptors.9 Activation in the mesolimbic systemic can increase appetite and contribute to the antiemetic, psychoactive, and muscle relaxant properties.10
Both THC and CBD, being lipophilic, can easily cross the blood brain barrier.8 As a result, cannabinoids can interact with analgesics and anesthetics to affect pain and perioperative outcomes.7 Regular use of cannabis can lead to accumulation with estimated half-lives reportedly between 2–5 days or more.11 Chronic, heavy cannabis users can test positive for THC in the urine up to 30 days after use.12
Clinically, cannabis use has been shown to influence cardiovascular, respiratory, neurological, immune and other systems, apart from its interactions with medications.13 14 There have been concerns of potential harms in the form of arrhythmias,13 14 blood pressure changes,15 laryngospasm,16 and respiratory obstruction,17 central nervous system changes, increased perception of pain and severe nausea-vomiting.18 At the same time, cannabis could be beneficial for chronic pain,4 and considered in the context of opioid reduction,5 apart from its potential for improving nausea-vomiting.19 20 These findings have implications to perioperative care.21 22 There are reports observing increased sedation23 and induction dose requirements for surgical procedures.24 Considering the divergent effects (positive and negative), anesthesiologists and perioperative physicians could face a dilemma on continuing or stopping cannabis use around surgery.5 Additionally, stopping of cannabis can lead to withdrawal symptoms.16 Unfortunately, there is paucity of published literature to guide practice, with most being either small-sized studies25 or studies focusing only on heavy cannabis use.26 In this retrospective cohort study, we set out to assess the effect of cannabis on perioperative outcomes in a large sample of surgical patients. Since cannabis has the potential to affect major organ systems,15 22 26 and interact with anesthetic drugs, we considered the influence of cannabis on major perioperative adverse events as our primary objective. Effects of cannabis on perioperative pain27 and opioid needs, propofol induction dose,24arrhythmias,13 28 and postoperative nausea-vomiting (PONV),18 were considered as secondary objectives.
Methods
Design
This was a retrospective cohort study of elective surgical patients endorsing the regular use of cannabis, either for recreational or medicinal purposes, in their preoperative visit between January 2018 and March 2019, at St. Joseph’s Hospital (a tertiary care academic hospital) in Canada. The study was conducted after ethics board approval.
Study population
Eligibility criteria included patients aged 18 years or older, who underwent an elective surgical procedure under general or regional anesthesia. Patients were asked about current cannabis use during their preoperative anesthesia assessment, and based on their self-reporting, cannabis users were identified on their anesthesia record. Patients having their surgery under monitored anesthetic care were excluded because they do not undergo major physiological changes and experience minimal pain and risk of PONV, as compared with patients having full anesthesia. Inclusion of these patients could confound the study outcomes. Patients in the control group were identified with the same criteria during the study period and were selected randomly based on number of cannabis users for a particular month and matched for age (within ±5 years) and sex. To increase the efficiency of matched cohorts, we used a ratio of 1:2 (users:non-users).
Database
The study site uses Epic, a well-known digital health information and medical record system (https://www.epic.com/software%23PatientEngagement, for all its electronic medical records (EMR), including documentation of all perioperative and in-hospital records. This was initiated in December 2017, prior to the conduct of this study.29 We used this database to screen patients as well as to collect study data.
Study data collection
We collected baseline parameters of age, sex, body mass index (BMI), American Society of Anesthesiologists (ASA) risk classification, tobacco use (presence and pack-year history), alcohol use (standard drinks/day), recreational drug use, cannabis use, and presence of major systemic respiratory, cardiac or nervous system comorbidity. Surgical details included type of anesthesia, admission (ambulatory or in-patient), surgical category, duration, dose of induction agent for general anesthesia (GA), dose of intraoperative opioid, and airway obstruction. Postoperative data included pain scores in the postanesthesia recovery unit (PACU), use of antiemetics, dose of postoperative opioids administered in PACU, and incidence of adverse events.
Study outcomes
Our study outcomes were informed by a review of the literature conducted in November 2018 and updated in August 2020 within Medline. Our primary outcome was the incidence of a major perioperative adverse event during the hospital stay as a composite outcome that included: (1) respiratory arrest or cardiac arrest, (2) unplanned admission to intensive care, (3) new-onset stroke, (4) new-onset myocardial infarction (MI) and (5) in-hospital mortality. Secondary outcomes included (1) incidence of individual outcomes noted within our primary composite outcome, (2) new-onset arrhythmias during hospital stay, (3) incidence of PONV needing treatment in PACU, (4) incidence of respiratory obstruction in PACU, (5) dose of propofol as the anesthetic induction for GA, (6) dose of intraoperative opioids, (7) postoperative pain scores in PACU and (8) dose of postoperative opioids in PACU. Because these were exploratory analyses, we did not adjust for multiple comparisons. All outcomes were noted based on physician notes and documentation in the EMR. Complications were assessed as either present or absent for each patient and were not categorized by severity or the type of treatment warranted.
Statistical analysis
Patient characteristics, anesthesia and surgery-related information were summarized using descriptive summary measures, expressed as mean (SD) or median (IQR) for continuous variables, and number (%) for categorical variables. Baseline differences were estimated using two sample t-test, Mann-Whitney U test, χ2 test or Fisher’s exact tests. To control for confounding, we used a propensity score-matched analysis of the cohort to assess the adjusted association between cannabis use and study outcomes.30 This allows us to identify matched pairs in relation to their probability of exposure or non-exposure to cannabis. The propensity score was estimated using multiple logistic regression in which cannabis use status was regressed on 14 baseline variables based on clinical reasoning and observed baseline differences. These factors included age, sex, BMI, smoking status, alcohol, other recreational drug use, setting (inpatient vs ambulatory), type of surgery (six types) and GA. For our matching, we used 1:2 match using a caliper with 0.2 of the SD of the logit of the propensity score. The standardized mean difference (MD) was used to examine the balance of covariate distributions between cannabis use and control groups. Opioid use was converted to intravenous morphine equivalents.31 Outcomes were reported as OR or MD with 95% CI. We used Stata, V.15.1 (StataCorp) to conduct all descriptive and inferential analyses.32 A two-tailed p<0.05 was considered statistically significant.
Sample size
The hospital initiated the integration of perioperative records with the EMR in November–December 2017, and we intended to capture all patients since January 2018 up to March 2019. Based on our initial chart review we estimated approximately 30–40 cannabis users having surgery/month and within 15 months we would have approximately 450–600 patients with history of cannabis exposure. A formal sample size estimate was not conducted.
Results
Among 15 048 patients having surgery between January 2018 and March 2019, 606 patients were identified as cannabis users, and another 1212 patients were identified as matching controls based on their age and sex, for a total sample of 1818 patients (figure 1). Baseline data indicated that cannabis users had a lower BMI; higher use of current smoking, alcohol, and other recreational drug use; and a higher prevalence of coronary artery disease. For matching, 142 patients were excluded due to missing BMI, giving us 524 cannabis users and 1152 control patients. Baseline and surgical data for the 1676 patients after matching is shown in tables 1 and 2. Matching ensured a standardized difference of 10% or less for known covariates, indicating appropriate balance figure 2, tables 1 and 2, except for history of heart failure. The prevalence of reported cannabis use was 4% (606/15048) and specifically in women having elective C-sections was 1.6% (10/606). Cannabis users tended to have a higher ASA score and underwent more orthopedic and plastic surgical procedures (32.4%) than the control group (22.8%). Cannabis users also had a higher rate of combined GA with regional compared with only GA (table 2).
Study outcomes
We summarize our matched analysis of study outcomes in table 3. Outcomes of propofol induction dose, intraoperative opioid dose, postoperative opioid dose, and postoperative pain scores were not normally distributed.
Primary outcome
There were 11 events for our composite outcome in the non-user control (0.9%) and seven in the cannabis user group (1.2%). One patient in each group had a respiratory/cardiac arrest and MI. No patients had stroke during this study period. The incidences of unplanned postoperative intensive care unit (ICU) admissions were nine in the control and five in the cannabis group. Three patients in the control group died in hospital. Post matching analysis indicated no difference in the odds of having our primary outcome (OR 1.06, 95% CI 0.23 to 3.98). As the event rates were very low for individual outcomes within the composite, we did not perform specific matched analysis.
Secondary outcomes
Although there was a significant difference in the outcome of new-onset arrhythmia in the prematched analysis, matched analysis indicated no such differences (OR 1.67, 95% CI 0.76 to 3.58). In patients having GA, the induction doses of propofol (mg) suggested a slight increase in the cannabis group, but it was not statistically significant (MD 8.06, 95% CI −0.20 to 16.33, p=0.06). Cannabis exposure did not result in any differences in the opioid dose used during surgery or in PACU. Pain intensity at PACU discharge was not statistically different with 30.4% in the control group and 33.5% cannabis users reporting moderate to severe pain. In PACU, 12% (n=130) of control patients compared with 9.5% (n=50) of cannabis users needed treatment for PONV, but this was not statistically significant.
Discussion
Our large observational study of surgical patients observed a cannabis use prevalence of 4% based on self-reporting. Compared with matched controls, cannabis use was not associated with increased odds of major cardiovascular or neurological outcomes. Exposure to cannabis did not influence propofol induction doses, intraoperative pain or opioid use, postoperative pain, or opioid use. Based on incidences, there was a possibility of higher risk of new-onset arrhythmias during hospital stay and lesser risk of PONV needing treatment with cannabis use, but neither was statistically significant.
Cannabis use among surgical patients
The prevalence of cannabis use in surgical population is unclear, as most existing reports rely on databases that capture cannabis use disorder (CUD) as a diagnosis and not average cannabis use.26 33 McAfee reported a prevalence of 5.9% in their prospective study.5 Similar figures have been noted in the general population, 5.4% in theUSA and 6.1% in Canada.3 The prevalence in our study of 4% (606/15 048), is probably reflective of a relatively older population and potential under reporting. The increased rate of regional anesthesia in cannabis users is likely reflective of their surgical groups with higher cannabis use (orthopedic and plastic surgeries).34
Cannabis and major perioperative outcomes
Our study is perhaps the largest cohort study evaluating the effect of self-reported current cannabis use in surgical patients. Most of the existing randomized controlled trials (RCTs) are small with methodological limitations.22 Although Ladha et al identified 13 RCTs around the perioperative period, nearly all involved THC or similar compounds. They had significant heterogeneity in their study population and outcomes.22 Most of the existing observational studies involving large sample sizes have been based on healthcare databases recording CUD, which reflects heavy and inappropriate use of cannabis from a health and societal perspective. Goel et al reported a large retrospective cohort study on 27 206 prospective matched cohort patients with a mix of cardiac and non-cardiac procedures, identified using the Nationwide Inpatient Sample from 2006 to 2015.26 The exposure was identified as a diagnosis of CUD, which differs from our population of self-reported current cannabis use. Although there was no difference in the composite outcome, the adjusted odds of postoperative MI were higher with CUD (OR 1.88, 95% CI 1.31 to 2.69).26 McGuinness et al looked at a surgical cohort of patients having six common vascular procedures and within them identified a subset of patients with CUD in a dataset administered by the Agency for Healthcare Research and Quality.33 Among 4684 patients with propensity matching, CUD patients had a higher incidence of perioperative MI (OR 1.56, 95% CI 1.09 to 2.24) and perioperative stroke (OR 1.59, 95% CI 1.20 to 2.12) than patients without CUD. Interestingly, patients with CUD had a lower incidence of sepsis. We did not specifically look into infection or sepsis; however, the incidences of unexpected ICU admissions were similar.33 In contrast to both these studies, we did not observe any increased risk of MI in current cannabis users, perhaps because of the dose or other factors associated with CUD. Jung et al looked at studies assessing the use of cannabis in patients having obesity surgery.35 Having only six observational studies with 1167 patients of whom only 116 had history of cannabis intake makes it hard for making any reasonable conclusions.35 There was a suggestion that cannabis use can improve weight loss without influencing other outcomes. On a related observation, the BMI in our cannabis user group was significantly lower than the non-cannabis group at baseline, p<0.001. Whether the use of cannabis predisposes to weight loss is not clear in literature.
Cannabis use and perioperative pain control
There is no consensus within existing studies on the effects of cannabis on perioperative pain control.36 In our study, we did not find any differences in pain scores, and approximately 30% patients had moderate to severe pain at the time of discharge from recovery in both groups. Abdallah et al reported a systematic review and meta-analysis of studies looking at the analgesic efficacy of perioperative cannabinoid compounds for acute pain.27 Meta-analysis of eight trials indicated slightly higher resting pain scores within the control group at 12 hours (data provided by 3/8 trials) (MD 0.83, 95% CI 0.04 to 1.63).27 However, all included trials had predominant effects of THC, including three discontinued compounds: GW842166, AZD1940, and levonantradol.27 So, these clinical implications may not be presently relevant. In our study, opioid doses during the surgery and in recovery were not different between the two groups. In contrast, Bauer et al reported higher opioid requirements in bariatric surgery patients using cannabis (36/434) compared with non-users despite similar pain scores.37 McAfee et al observed that cannabis users present with worse pain and more centralized symptoms, apart from increased use of recreational substances, particularly opioid and benzodiazepine use.5 These factors and the increasing use of cannabis for chronic pain,3 may predispose patients to higher postoperative pain and opioid use. Unfortunately, they did not report on postoperative pain control or opioid use. At 3 and 6 months, the differences in presurgery characteristics continued but there was no change in relation to surgical outcomes, including persistent pain at 6 months, recovery or treatment efficacy.5
Effect of cannabis on PONV and anesthetic medications
The dose of propofol in our study was higher in the cannabis group by 8 mg; however, it was not statistically significant (p=0.06). Although we controlled for alcohol and history of substance use, we were unable to capture the use of preoperative opioids and benzodiazepines. Hence, there is a risk of confounding and needs further exploration.24 We observed a decreased potential for PONV with cannabis use (9.9%) compared with controls (13%). Suhre et al not only observed an increased overall risk, but daily users had a much higher absolute risk increase of 3.3% (95% CI 0.4% to 6.4%), compared with current users, 1.2% (95% CI −0.7% to 3.1%).18
Strengths and limitations
Compared with existing studies, our study had a larger sample size and evaluates major outcomes in the context of self-reported current cannabis use. Although we did not find significant differences in any of our outcomes, we observed important differences in the dose of propofol, new onset arrhythmia, and the potential for decreased PONV, all of which need to be evaluated in a more rigorous study. Our study had some important limitations. Being a retrospective study, we were unable to capture the reason for cannabis use or quantify the duration, amount, and type of cannabis used. This is a challenge commonly observed in the cannabis literature. Despite the intent to categorize patients’ on the reason for use, a substantial percentage report using it for both medicinal and recreational purposes.5 We were also unable to report on the type and amount of other recreational drug use in our study patients, as it was not included in the anesthetic record. Self-reporting its use before surgery may have led to underreporting. Lastly, a substantial number of our study patient had day surgeries (63%) and outcomes were only collected until the time of discharge. Thus, there may be delayed complications that were not captured in this study. We are planning to work on a prospective study that could allow us to overcome these limitations and better inform care.
Conclusion
Our results do not demonstrate a convincing association between self-reported cannabis use and major surgical outcomes or pain management. The risk of MI observed in other studies is likely reflective of heavy cannabis use in susceptible population. As suggested by the recent consensus guidelines, perioperative decisions should be made based on appropriate considerations of dose, duration, and indication of use.25
Data availability statement
Data are available on reasonable request from the corresponding author.
Ethics statements
References
Footnotes
Twitter @harshamd5
Contributors BHZ: acquisition of data, drafting and critical revision, final approval; HS, NS, MC, DB, and LR: acquisition of data; final approval; LW: analysis and interpretation of data, final approval; HS: conception and design; analysis and interpretation of data; final approval.
Funding This project received funding from the McMaster Medical Student Research Excellence Scholarship.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.