A 2024 propensity-matched study of older surgical patients found that those who used cannabis before surgery reported modestly higher pain scores and received slightly more opioids in the first 24 hours after surgery. The differences were statistically significant but small, and the observational design cannot establish that cannabis caused these outcomes.

Cannabis use among Americans over 65 has risen dramatically in the past decade, and many of these older adults will undergo major surgery. Yet until now, essentially no published research has examined what preoperative cannabis use means for postoperative pain management specifically in this age group. Older surgical patients are disproportionately vulnerable to both inadequately treated pain and opioid-related adverse events, making this intersection of cannabis use, aging, and perioperative care an urgent gap in the clinical evidence base that demands systematic investigation.

Cannabis use among adults over 65 has risen sharply alongside the expansion of medical and recreational legalization, yet its perioperative implications in this population have remained almost entirely unexamined. This study addressed that gap by using a validated natural language processing algorithm to identify cannabis users from clinical notes in the electronic health record at the University of Florida Health, overcoming a well-known limitation of prior research that relied on ICD codes alone. The NLP system achieved 93% precision and 95% recall, capturing approximately 26 times more cannabis users than ICD codes would have identified. The investigators then applied propensity score matching on a broad set of clinical covariates to create balanced comparison groups of older surgical patients.

In the final matched sample of 504 patients, cannabis users had significantly higher median postoperative pain scores (4.68 vs. 3.88 on a 0 to 10 scale; difference 0.80 points; 95% confidence limits 0.19 to 1.36; p=0.01) and received more opioids (42.5 vs. 30.0 mg oral morphine equivalents; difference 12.5 mg; 95% confidence limits 3.80 to 21.20; p=0.02) in the first 24 hours. However, the pain score difference falls below most accepted minimal clinically important difference thresholds, and the opioid difference corresponds to roughly one additional standard tablet of oxycodone. The authors appropriately acknowledge that cannabis dose, frequency, route, and reason for use were unmeasured, and they call for prospective studies with detailed exposure characterization.

Older Cannabis Users Report Modestly Higher Pain After Surgery: What a New EHR Study Does and Doesn’t Show

Nearly one in twenty Americans over 65 now uses cannabis, and many of them will undergo surgery. A new study asks what that means for postoperative pain, and the answer is more nuanced than the headlines suggest. The paper appears to claim that cannabis use is associated with increased postoperative pain and opioid requirements in older adults, and what it actually tested is whether, after balancing a set of measurable clinical characteristics, older cannabis users had higher pain scores and opioid consumption in the first 24 hours after major inpatient surgery. Before any criticism, the study deserves genuine credit: the NLP-based cannabis ascertainment is a real methodological step forward. The finding that ICD codes captured only 3.75% of the cannabis users identified by the algorithm is striking, and it is the kind of result that should make every researcher who has relied on billing codes alone for cannabis exposure feel uneasy. To put it plainly, using ICD codes to count cannabis users is like trying to count all coffee drinkers in a hospital by looking only at caffeine toxicity diagnoses. You would miss nearly everyone. This single finding may prove to be the study’s most lasting contribution to the field.

The central methodological challenge, however, is that propensity score matching can only balance the confounders you can see. The investigators carefully matched on age, sex, BMI, comorbidities, surgical type, and anesthesia approach. But they had no information on what kind of cannabis these patients were using, how much, how often, by what route, or why. A patient taking low-dose CBD oil once a week for arthritis and a patient smoking high-THC flower daily for chronic pain were placed in the same “cannabis user” category. This is like grouping daily espresso drinkers with weekly decaf consumers under “coffee users” and then drawing conclusions about the effect of coffee; the category is simply too broad to support precise clinical interpretation. Perhaps more importantly, the unmeasured factor most likely to confound these results is underlying chronic pain severity. People who use cannabis before surgery may well be those with more severe baseline pain, and if that underlying pain also predicts higher postoperative pain and opioid need, then cannabis is a marker for the confounder rather than the cause. Cannabis withdrawal in the perioperative period, an entirely unstudied phenomenon in this cohort, represents another plausible alternative mechanism.

To a patient, I would say this study suggests that older adults who use cannabis before surgery may experience slightly more pain and need a bit more pain medication afterward, but we do not yet know if cannabis itself is the reason, and the most important thing you can do is tell your surgical team about your use so they can plan accordingly. To a colleague, I would note that the effect sizes here are modest enough that I would not change my analgesic protocols based on this alone, but I would use it to reinforce the value of routinely documenting cannabis use in preoperative assessments. To a policymaker, I would argue that the appropriate response is not to mandate clinical protocols, but to fund prospective trials that actually characterize cannabis exposure in surgical populations. In EHR-based research on behavioral exposures like cannabis use, the quality of exposure ascertainment is at least as important as the sophistication of the statistical adjustment, and in most existing cannabis research, that ascertainment has been far too crude to support the conclusions drawn.

This study sits at an early point in the research arc for cannabis and perioperative outcomes in geriatric populations. Prior retrospective cohort studies in younger surgical patients have reported similar associations between cannabis use and higher postoperative pain scores or opioid consumption, and this work extends those findings to older adults with improved exposure ascertainment. However, none of these studies, including this one, has crossed the threshold from association to causation, and none has provided the dose-response or mechanistic data needed to guide perioperative pharmacology. The evidence base remains firmly observational and hypothesis-generating.

From a pharmacological standpoint, potential mechanisms for the observed associations include opioid cross-tolerance mediated through overlapping endocannabinoid and opioid receptor signaling, cannabis withdrawal-related hyperalgesia in the perioperative period, or simple confounding by preexisting chronic pain conditions. Each mechanism would call for a different clinical response, and the current data do not distinguish among them. Safety considerations are heightened in older adults, who metabolize opioids more slowly and are more susceptible to respiratory depression, falls, and delirium. The single most concrete recommendation supported by this evidence is to routinely and systematically document cannabis use, including product type and frequency when possible, as part of every preoperative assessment in older adults, and to incorporate that information into individualized postoperative pain management plans.

This is a propensity score-matched retrospective cohort study using electronic health records from a single academic medical center. In the evidence hierarchy, it ranks above unmatched observational studies but well below randomized controlled trials. Its single most important inference constraint is that retrospective observational designs cannot establish causation, and propensity score matching can only address measured confounders, leaving unmeasured factors such as cannabis dose, frequency, and underlying pain severity as potential explanations for the observed associations.

The findings are consistent with the direction of prior retrospective cohort studies in non-elderly surgical populations that have reported associations between cannabis use and modestly higher postoperative pain scores or opioid requirements. This study extends those findings specifically to adults over 65, a population previously unrepresented in this literature, and does so with substantially improved exposure ascertainment through NLP.

What the study adds beyond confirmation is the methodological demonstration that ICD-code-only approaches vastly undercount cannabis use, a finding that should prompt reappraisal of prior EHR-based cannabis research that relied on coded diagnoses alone. The ASRA-PM perioperative cannabis guidelines already recommend routine cannabis screening; this study provides additional, though still observational, support for that recommendation in the geriatric surgical population.

The most consequential analytic choice was the decision to treat all cannabis users as a single homogeneous group. If the investigators had been able to stratify by cannabis dose, frequency, or route of administration, the overall association might have resolved into a strong signal confined to heavy users and no signal among occasional or low-dose users, or alternatively, a more uniform dose-response gradient. Either result would have profoundly changed the clinical interpretation.

Additionally, including a measure of preoperative chronic pain severity or preoperative opioid use as a covariate in the propensity model could have attenuated or eliminated the observed associations if underlying pain burden accounts for the signal. The convention of scoring pain as zero during sleep may also have introduced differential measurement error if cannabis users and non-users had different sleep patterns in the postoperative period.

The most likely overinterpretation of this study is the claim that cannabis causes more postoperative pain in older adults. The retrospective observational design can only demonstrate association, not causation, and several plausible alternative explanations remain unaddressed. Patients who use cannabis before surgery may have higher baseline pain severity, and cannabis use may be a marker for that underlying condition rather than a direct contributor to postoperative pain.

A second common misreading would be to advise older patients to stop cannabis abruptly before surgery based on this evidence. The study did not test cessation, and perioperative cannabis withdrawal, which can include pain sensitization, anxiety, and sleep disruption, was not examined and could itself worsen postoperative outcomes. The 0.80-point pain difference should also not be characterized as a large or clinically definitive effect; it falls below most accepted thresholds for minimal clinically important difference on a 10-point scale.

This study contributes the first focused evidence on cannabis use and postoperative pain outcomes specifically in older adults, using improved NLP-based exposure ascertainment. It finds modest, statistically significant associations between preoperative cannabis use and slightly higher pain scores and opioid consumption. It does not establish that cannabis causes worse postoperative outcomes, and the effect sizes are small. What it supports most clearly is the practice of routinely documenting cannabis use in preoperative assessments and the need for prospective research with detailed exposure characterization.

Does this study prove that cannabis makes surgery more painful?

No. This study found an association between preoperative cannabis use and slightly higher postoperative pain scores, but it cannot prove that cannabis caused the difference. People who use cannabis may have more underlying pain to begin with, and the study could not separate these possibilities.

Should I stop using cannabis before my surgery?

This study did not test whether stopping cannabis improves outcomes. Abruptly stopping cannabis before surgery could potentially cause withdrawal symptoms that affect your recovery. The most important step is to tell your surgical and anesthesia team about your cannabis use so they can tailor your pain management plan.

How much more pain are we talking about?

The difference was less than one point on a 10-point pain scale (0.80 points), and cannabis users received the equivalent of roughly one extra standard pain tablet over 24 hours. These are modest differences that may not be noticeable to most individual patients, though they could matter in populations especially sensitive to opioid side effects.

Does this apply to all cannabis users, or just heavy users?

The study could not distinguish between different patterns of cannabis use. Someone taking low-dose CBD oil occasionally was grouped with someone smoking high-THC cannabis daily. Future research will need to examine whether the type, dose, and frequency of cannabis use matter for surgical outcomes.

Have thoughts on this? Share it: