Welcome back to the journal review. This is the December twenty twenty-five issue of Dermatologic Surgery, and we've got four pieces worth your time tonight — two touch on opioid stewardship from very different angles, one is a retrospective look at a wound closure adhesive on the lower leg, and one tackles an old technique question that's never had good numbers behind it: curettage alone for keratinocyte carcinoma. Let's get into it.

First up is a retrospective cohort analysis titled "Cannabis Use Is Associated With Increased Postoperative Pain and Opioid Use After Mohs Micrographic Surgery." The clinical gap here is pretty simple to state — we know a lot about postoperative pain after Mohs from a management standpoint, and we know from the general surgical literature that cannabis-related disorder is linked to worse acute postoperative pain and higher narcotic use after non-dermatologic surgery, but nobody had really looked at this specifically in the Mohs population, where pain is usually mild and non-opioid regimens dominate.

Methodologically, this is a big-database study using TriNetX, which aggregates de-identified records across more than a hundred thirty million patients. The authors pulled everyone with a Mohs CPT code, then split them into a cannabis-related-disorder cohort and a no-cannabis-related-disorder cohort using ICD-ten codes, and used propensity score matching to balance the two groups one-to-one across age, sex, obesity, nicotine dependence, diabetes, anxiety disorders, prior opioid use, and prior emergency department visits. This is the right tool for the job here — you obviously can't randomize patients to have a substance use disorder, so propensity matching is the standard workaround to approximate an apples-to-apples comparison from observational data, and it's worth noting the unmatched cohorts were wildly different at baseline, so the matching step was doing real work — nicotine dependence alone was nearly ten-fold higher in the cannabis cohort before matching.

After matching, each group had a little over sixteen hundred patients. Within a week after surgery, the cannabis cohort was about eighteen percent more likely to receive an opioid prescription, which was statistically significant. At thirty days, they were nearly three times more likely to have a coded diagnosis of acute postprocedural pain — that's both statistically significant and clinically meaningful, not a trivial effect. They were also modestly but significantly less likely to attend their postoperative follow-up visit. Emergency department visit rates were similar between groups, so this isn't translating into a crisis-level utilization signal, at least not at the ED level.

The authors frame this as consistent with the broader substance-use literature — higher pain intensity and pain catastrophizing are well described in patients with cannabis-related disorder, and they appropriately caution that this doesn't mean you should reflexively hand out more opioids to these patients; it means you should recognize the pattern and lean harder into multimodal, non-opioid strategies, citing the older Sniezek data showing that alternating acetaminophen and ibuprofen actually outperforms acetaminophen-codeine combinations after Mohs. The missed-follow-up finding is the other practically important thread — postoperative visits are how you catch wound complications and decide on further reconstruction, so a population that's both more symptomatic and less likely to show up is a real setup for suboptimal outcomes.

Limitations are the ones you'd expect from a claims-code study: it's retrospective, so no causality; ICD-ten and CPT coding can misclassify both exposure and outcome; and critically, the codes can't capture wound-level variables we know drive pain, like scalp or lower-extremity location, same-day multiple procedures, or healing by primary versus secondary intention. The "acute postprocedural pain" code is also a blunt instrument — you can't tell if that's mild or severe, just that someone coded it. So take the effect sizes as directionally real but not finely calibrated. Practically, this isn't going to change your surgical technique, but it is a reasonable nudge to specifically ask about cannabis use during your preoperative risk conversation, set expectations proactively about a non-opioid pain plan, and maybe build in an extra check-in for these patients given the lower follow-up attendance.

Next, a ten-year single-center retrospective cohort study out of UC San Diego looking at two-octyl cyanoacrylate — the surgical skin glue — and surgical site infections after lower-extremity dermatologic procedures. The background problem is well known to anyone doing leg surgery: SSI rates on the lower extremity run meaningfully higher than elsewhere, commonly cited in the range of four to seven percent, and the traditional response has been liberal prophylactic antibiotics, which nobody's thrilled about given resistance concerns.

The design here is clever and worth walking through. Rather than a straight two-arm comparison, the authors built three cohorts across a ten-year window: a pre-adhesive era where everything was closed with sutures, a post-adhesive-availability era where the surgeon still chose sutures, and a post-adhesive-availability era where the surgeon actually used the glue. The reason this three-way split matters methodologically is that it lets you separate two different possible effects — an effect of the adhesive itself, versus a secular or "awareness" effect where simply introducing a new closure option into the clinic coincides with broader improvements in infection-control practice and antibiotic stewardship, independent of what's actually being smeared on any given wound. That's a smart way to interrogate causality in a retrospective design without needing randomization.

The results are genuinely nuanced, and I'd encourage you not to skim past this one. Confirmed, culture-positive infection rates dropped significantly from about five percent in the pre-adhesive era to under two percent once the adhesive era began — but here's the catch: that drop happened in the sutures-only post-adhesive group too, and there was no significant difference between the sutures-only post-adhesive group and the actual-glue-used group. In other words, the improvement tracks with the era, not with the adhesive itself. The same pattern held for clinically suspected infections — significant drop from about eight percent pre-adhesive to roughly three percent in both post-adhesive groups, sutures or glue, with no difference between them. Prophylactic antibiotic use fell from eighty-six percent to seventy-seven percent, and postoperative antibiotic prescribing fell from about eight percent to three percent, both significant. Closures were shorter and Mohs stage counts were slightly lower in the glue-used group, but the authors are appropriately cautious here — surgeons were preferentially choosing the adhesive for smaller, presumably lower-risk wounds, so that's very likely selection bias rather than the glue somehow causing fewer Mohs stages.

There's also a useful negative finding buried in the confounder analysis: neither prophylactic nor postoperative antibiotic use actually correlated with infection rates in this cohort at all. That's a meaningful stewardship data point on its own, separate from the adhesive question — it suggests a lot of that antibiotic use wasn't doing measurable protective work.

Limitations are squarely what you'd expect: single center, retrospective, non-randomized closure choice with clear selection bias toward smaller wounds for the adhesive, a large imbalance in immunosuppression prevalence between eras that's hard to fully adjust for, and ten years of secular trends in wound care that could confound any before-after comparison regardless of adhesive availability. So the honest takeaway is not "glue reduces infections" — the data don't really support that causal claim once you look at the sutures-only comparison arm. What it does support, and this is genuinely useful, is reassurance that two-octyl cyanoacrylate is a safe, non-inferior closure option for appropriately selected smaller lower-extremity wounds, and that the broader move away from routine prophylactic and postoperative antibiotics in this setting doesn't appear to be costing you in infection rates. That's a stewardship-supportive finding even if it isn't a glue-specific one.

Third article: a Medicare claims analysis of opioid prescribing among Mohs surgeons from twenty fourteen through twenty twenty-two. This is a straightforward descriptive and predictive epidemiology piece, not an interventional study, but it's a nice large-scale check on whether all the pain-management literature from the last decade has actually changed practice at the population level.

Methodologically, the authors identified Mohs surgeons through Medicare's provider and service files using the Mohs CPT codes, then cross-referenced Part D prescriber data to capture opioid prescriptions. Because the data aren't at the individual encounter level, they had to get creative — they used the Medicare reimbursement amount for each repair as a proxy for how painful that repair likely was, reasoning that reimbursement roughly tracks the extent of tissue manipulation, since more extensive flaps and grafts pay more than a simple linear closure. That's an inferred workaround rather than a validated pain measure, but it's a reasonable one given the data constraints, and they used it to bucket repairs into low, moderate, and high complexity.

The topline result: the percentage of Mohs surgeons prescribing any opioids to Medicare beneficiaries fell from forty-eight percent in twenty fourteen to thirty-three percent in twenty twenty-two — so by the end of the study, roughly two-thirds of Mohs surgeons weren't prescribing opioids to this population at all. Among those who did prescribe, the average days' supply dropped from three-point-seven days to three days, and the proportion of surgeons classified as high-volume prescribers — meaning four or more days' supply or opioids for more than half their cases — fell from thirty-one percent to twenty percent. The gap between the heaviest and lightest prescribers also narrowed somewhat over time, though it didn't close completely.

Predictors of prescribing at all were fairly stable across the study period: graduating medical school in nineteen ninety or later, practicing outside the Northeast, higher Mohs case volume, and doing more complex repairs. Things that didn't predict prescribing — physician sex, MD versus DO, urban versus rural location, patient age, and interestingly, the opioid overdose death rate in the surgeon's own state. When they zoomed in specifically on high-volume prescribing in the most recent two years, the picture shifted a bit: male sex and rural location predicted high-volume prescribing, while post-nineteen-ninety graduation and practicing in a high-overdose-mortality state predicted against it — and notably, repair complexity dropped out as a predictor of high-volume status specifically, even though it predicted prescribing at all. There's also a worthwhile drug-selection finding tucked in here: hydrocodone prescriptions fell by about a quarter over the study period, while tramadol prescriptions rose two and a half fold. That's worth flagging out loud, because tramadol is sometimes reached for as the "safer" opioid, but it carries its own dependency and serotonergic interaction risks, and this kind of substitution effect deserves a second look rather than being read as pure stewardship progress.

There's no dedicated limitations section spelled out in what we have, but the inherent constraints are worth naming yourself: this is Medicare-only, so it says nothing about prescribing to younger, privately insured, or cash-pay patients; the "percent of cases prescribed an opioid" is an estimate built from claims ratios rather than true encounter-level linkage; and small-state numbers can get skewed by a single outlier prescriber, which the authors themselves note happened with Wyoming's statewide average being driven by one physician. Practically, this isn't a study that changes what you do in your own operating room — it's a population-level trend confirmation. The genuinely useful takeaway is the tramadol substitution pattern, which is worth a moment of individual reflection on your own prescribing habits, and the reminder that rural and male-prescriber-associated high-volume patterns represent a plausible target for further stewardship outreach if you're involved in any quality-improvement effort at that level.

Last article, and this one's a nice piece of practical evidence-building: a retrospective cohort and interview study from the University of Utah on five-year recurrence rates of keratinocyte carcinoma treated with curettage alone — meaning curettage without the electrodessication step. The background gap is straightforward — electrodessication and curettage has decades of recurrence data behind it, but curettage alone, despite being used by some dermatologists for just as long, has never had a solid recurrence dataset, let alone one benchmarked against electrodessication and curettage.

The design is a hybrid retrospective chart review plus telephone survey, and the rationale for that combination is worth spelling out because it's a genuinely elegant solution to a common single-center problem: patients who move on to other clinics or providers become invisible to a pure chart-review study. So the authors split their cohort into an Arm A, patients who'd had documented follow-up back at the same institution and could be assessed entirely through the electronic record and clinical photographs, and an Arm B, patients without that in-house follow-up, who were instead contacted by phone, consented verbally, and asked directly whether their treated lesion had recurred, with medical records requested from outside clinics to confirm when needed. They then ran both a per-protocol analysis, excluding anyone truly lost to follow-up, and an intention-to-treat analysis where lost patients were conservatively counted as recurrences — a sensible way to bracket the true recurrence rate between a best-case and worst-case estimate.

Working from an initial pool of over eighteen hundred lesions, after exclusions the per-protocol analysis included just over a thousand lesions. The overall five-year recurrence rate was about three and a half percent, breaking down to roughly two and a half percent for basal cell carcinoma and four and a half percent for squamous cell carcinoma. For context, that lines up closely with published electrodessication-and-curettage recurrence rates from prior studies, which have generally clustered in the three to six percent range — so on its face, this supports the idea that skipping the electrodessication step doesn't meaningfully compromise cure rates, at least in this cohort and in experienced hands.

But there's an important anatomic caveat, and it's a big one: lesions on the head, neck, and scalp recurred at just over seven percent, significantly higher than the under-two-percent rate seen below the neck. Drilling into the face specifically, nose lesions had the highest site-specific recurrence at nearly eighteen percent, and Bowen-type squamous cell carcinoma in situ had the highest recurrence by subtype at twenty percent. There's also a subtler biological signal worth flagging for a Mohs audience specifically — among the basal cell carcinomas that did recur, several that started out as low-risk superficial or nodular subtypes came back with more aggressive micronodular or infiltrative features. The absolute numbers are small, but that's exactly the kind of finding that should keep you cautious about assuming a benign initial biopsy guarantees a benign recurrence if one happens.

The limitations here are real and worth naming plainly, since the authors themselves lean into a fairly confident conclusion. This is a single center, three providers, one specific curettage technique — a single vigorous pass with aluminum chloride for hemostasis — so it may not generalize to other curettage approaches. The comparison to electrodessication-and-curettage recurrence rates is indirect, pulled from separate historical studies rather than a head-to-head randomized or even concurrent cohort comparison. The phone-survey arm relies on patient self-report and recall, which introduces its own bias, particularly for patients who might not recognize a subtle recurrence. And critically, these providers were already selecting which lesions were appropriate for curettage alone in the first place — so this recurrence rate reflects a curated, lower-risk lesion population by definition, not an unselected cross-section of keratinocyte carcinoma.

So here's the practical read: this is genuinely useful supportive evidence, not a practice-changing mandate. For low-risk basal cell and squamous cell carcinomas below the neck in experienced hands, these numbers give you real reassurance that curettage alone performs comparably to electrodessication and curettage, and that's a legitimate option to keep in your toolkit for the right patient and lesion. But the head-and-neck data, especially the nasal and Bowen-type numbers, argue against extending this technique casually to facial or scalp lesions, where recurrence climbed several-fold — those sites still deserve more definitive treatment, whether that's electrodessication and curettage, excision, or Mohs, depending on the clinical picture.

That's the December issue. A cannabis-and-pain cohort study reinforcing that substance use history belongs in your preoperative risk conversation, a nuanced adhesive study that turns out to be more about the era of practice than the glue itself, a Medicare-wide confirmation that opioid stewardship among Mohs surgeons keeps improving with an interesting tramadol wrinkle worth watching, and solid new numbers behind an old technique question on curettage alone. Thanks for listening, and I'll see you next month.