Welcome back to the journal club. This is our walkthrough of the April twenty twenty-six issue of the Journal of the American Academy of Dermatology — four pieces this month, and it's an interesting mix: two Ethics Journal Club letters bookending two brief reports, one on health economics and one a systematic review and meta-analysis. Let's get into it.

First up is an Ethics Journal Club piece, structured as an advice column, on balancing anxiolytics and patient autonomy during Mohs surgery. The question comes from a hypothetical "Dr. Careful," who raises a real tension a lot of us live with day to day — you want your anxious patients comfortable, but Mohs is unusual among procedures in that you're having substantive, decision-relevant conversations with the patient in real time, mid-case, about margins, staging, and reconstructive options. So what happens to informed consent if that patient is on board?

The piece frames this as a collision between two ethical principles: autonomy and nonmaleficence. On one side, untreated periprocedural anxiety isn't just an experience problem — it has physiologic consequences, tachycardia and hypertension among them, and benzodiazepines have a good track record in dermatologic surgery for reducing anxiety and pain and improving satisfaction, with short-term use carrying minimal dependence risk. On the other side, the data are clear that even single low doses of these drugs measurably impair cognition — the authors cite midazolam affecting memory, language, and attention, and lorazepam prolonging reaction time. So if a patient takes their anxiolytic and then you're negotiating flap versus graft with them intraoperatively, you have to ask whether that's a conversation with someone who has full decisional capacity.

The recommendation isn't to avoid anxiolytics — it's sequencing. All substantive discussion of the procedure, anticipated intraoperative decisions, and the associated risks, benefits, and alternatives needs to happen before any benzodiazepine is on board, as part of a proper informed consent conversation. The medication is taken only after that consent process is complete, and only under direct instruction from the clinical team — meaning if a patient shows up having already dosed themselves at home, the practical answer is to reschedule them. They also recommend having a designated driver present, with preference for someone who's actually authorized as a surrogate decision-maker, identified before the drug is given, in case intraoperative decisions do need to be made after sedation. And they suggest using front-office staff — schedulers — as an early screening layer, using a standardized questionnaire to flag anxious patients for either a presurgical consultation or non-pharmacologic options like music or guided breathing for patients who'd rather stay fully engaged.

There's no data or outcomes here to weigh — it's a normative, practice-guidance piece. The actionable takeaway for your practice is really operational: build the consent conversation and anxiolytic timing into your workflow explicitly, rather than leaving it to case-by-case judgment on the day of surgery, and have a clear surrogate-decision-maker identification step baked into your pre-op process if you're routinely offering perioperative benzodiazepines.

Second article, a brief report, and this one is squarely about your reimbursement environment: an analysis of Medicare payment escalation for new skin-substitute Q-codes between twenty twenty-two and twenty twenty-five. Background here is the wave of anecdotal and lay press reporting — including a New York Times piece — suggesting a proliferation of new cellular and tissue-based products, or CTPs, priced well above existing products without any demonstrated added clinical value. The authors wanted to quantify whether that narrative actually holds up in the claims data.

Methodologically, this is a straightforward descriptive analysis of publicly available CMS files — the Part B average sales price pricing spreadsheets — pulled quarterly from twenty twenty-two through the second quarter of twenty twenty-five, filtered to the Q41 series of Healthcare Common Procedure Coding System codes, which are the temporary codes CMS assigns to new products to allow billing while longer-term coverage and efficacy determinations are pending. That's a sensible design choice given the question — you don't need a clinical trial to answer "did prices go up and did the number of new codes increase," you need the actual government pricing files, which are the authoritative source for what Medicare pays. They used Wilcoxon rank-sum testing to compare price distributions between new and established codes, and adjusted for inflation using the CPI so that the price trend over time isn't confounded by simple currency inflation.

The results are fairly striking. A hundred twenty new Q-codes were added over the three-and-a-half-year window. Median price per square centimeter was flat, sitting around one hundred dollars through twenty twenty-two, then started climbing gradually in twenty twenty-three. Then there's a clear inflection point in the third quarter of twenty twenty-four, where the median roughly doubled to around two hundred forty-five dollars, and it kept climbing to about four hundred dollars by mid twenty twenty-five. Among the seventy-three codes launched after mid twenty twenty-three, prices ranged enormously — from roughly two hundred dollars up to over fifteen hundred dollars per square centimeter. And once you inflation-adjust and pool everything, new entrant products were about three-fold costlier than established ones — roughly five hundred dollars versus one hundred fifty dollars per square centimeter — a difference that was statistically significant and, importantly, also clinically and economically meaningful given the volumes at which these products are used. Products priced above a thousand dollars per square centimeter made up about two-thirds of new codes launched in twenty twenty-four to twenty twenty-five, versus less than one in ten of the older, established codes.

The authors tie the inflection point to a twenty twenty-three CMS policy change that simplified the Q-code application process, effectively lowering the barrier to market entry for new products — that's their proposed explanation, and it lines up temporally quite well with the data. Limitations are honestly stated: this is pricing data only, not clinical effectiveness data, so they can't and don't claim these newer, pricier products are equivalent or inferior to older ones — only that price rose sharply without any accompanying comparative efficacy evidence. Average sales price also excludes rebates and site-of-service adjustments, and it's reported per square centimeter rather than per case, so real-world total cost depends on wound size and how many applications are used.

For practical purposes, this isn't a paper that changes how you select a skin substitute tomorrow morning. But it's highly relevant background if you use CTPs after tumor extirpation or for chronic wound management adjacent to your surgical practice — it's a signal that payer scrutiny and potential bundled "episode-based" payment reforms for these products are coming, and it's worth knowing the pricing landscape you're operating in, particularly if you're fielding questions from administrators about product selection or facing prior-authorization pushback.

Third article, also a brief report, is a systematic review with meta-analysis estimating the burden of keratinocyte carcinoma attributable to ambient ultraviolet exposure — specifically through the lens of sunburn history. The gap they're addressing: we know UV drives basal cell and squamous cell carcinoma, but the field has lacked a clean, quantitative synthesis of how the timing of sunburn exposure — childhood versus adulthood versus lifetime — differentially affects risk of each tumor type.

Methods: they searched PubMed and Embase from nineteen sixty-four through mid twenty twenty-four, following PRISMA reporting guidelines, with three independent reviewers screening for eligibility. Studies had to clearly distinguish basal cell from squamous cell carcinoma cases and report actual case-control data on sunburn history and burden — quite a strict inclusion bar, which is why out of three hundred twenty-five identified studies, only thirty-five — about one in ten — made the final cut. That level of exclusion makes sense given their requirement for granular, tumor-type-specific, exposure-timed data, but it's also worth flagging as a limiting factor for statistical power. They ran random-effects models to pool odds ratios for dichotomous "ever sunburned" exposure, and a separate dose-response logistic regression using cumulative sunburn frequency, which lets them speak to causality rather than just association.

The results: for basal cell carcinoma, ever having been sunburned raised the odds by about half — a statistically significant and clinically meaningful effect. Risk was highest for people reporting sunburn throughout their whole life, a bit lower for childhood-adolescent sunburn, and lowest — though still significant — for adulthood-only sunburn. The dose-response data reinforced this: more cumulative sunburns per year tracked with higher odds of basal cell carcinoma in a clean dose-dependent fashion, and high lifetime sunburn frequency carried roughly two-thirds higher odds. Notably, cumulative adult sunburns alone showed a numerically doubled risk that did not reach statistical significance — likely an underpowered subgroup.

For squamous cell carcinoma, the pattern was similar but generally a touch weaker in magnitude: ever having been sunburned raised the odds by about forty percent, again statistically significant. Lifetime sunburn history carried the highest risk, followed by childhood-adolescent, then adult-only. The dose-response findings again showed lifetime cumulative sunburn conferring the highest risk among the three exposure windows.

The authors' interpretation is that childhood and lifetime sunburn exposure matter more than adult-only exposure for both tumor types, and that the somewhat stronger association with basal cell than squamous cell carcinoma probably reflects differing biology — squamous cell carcinoma has other major contributors like immunosuppression and HPV that dilute the UV-specific signal in a pooled analysis. Limitations are the ones you'd expect from a sunburn-history meta-analysis: it's all retrospective self-report, so recall bias is a real concern, and they flag a subtler issue — several included studies examined the same participants across multiple age windows, meaning there's likely double-counting of individuals across their combined-exposure categories, which could be inflating the weight given to those pooled estimates.

Clinically, this doesn't change your Mohs practice directly, but it does reinforce a counseling point you're well-positioned to make, especially with patients and their families in clinic: early-life sun protection carries outsized weight in lifetime keratinocyte carcinoma risk, which is a useful framing for family-based prevention conversations, and for supporting broader public health messaging around childhood photoprotection rather than treating adult sunscreen habits as the primary lever.

Fourth and final piece, back to the Ethics Journal Club format — this one on the ethics of using artificial intelligence to polish letters of recommendation for dermatology residency applicants. The question posed: if you're writing an LoR and you run it through an AI tool purely to smooth out phrasing and improve flow, are you doing anything ethically fraught?

The authors work through three concerns. First, authenticity — letters of recommendation are supposed to carry the individualized voice and judgment of the writer, and AI editing tends to amplify positive language beyond what the writer originally intended, which can distort the letter's actual signal. This is compounded by the fact that free-text sections on standardized LoR forms are already brief, so AI smoothing risks homogenizing language across letters and washing out the applicant-specific detail that programs are actually trying to extract. Their guidance is that AI-polished text must be carefully proofread against the writer's original intent, and that responsibility for authenticity sits with the human writer, full stop.

Second, bias and fairness. They cite prior work showing large language models tend to describe women using communal language — "warm," for instance — and men using agentic language like "leader," which means unreviewed AI editing can quietly reintroduce or amplify gender bias into letters. There's also an access-and-skill dimension here that's easy to overlook — faculty who are more adept at prompting AI tools may end up with more polished, more embellished letters than colleagues who aren't, which introduces an inequity that has nothing to do with the applicant. On the flip side, if residency programs are using AI for applicant screening, the authors argue those screening models need to be calibrated to be style-insensitive, or they'll systematically favor AI-polished letters over fully human-written ones — a fairness problem on both ends of the pipeline.

Third, privacy and trust. Personalized letters often contain identifiable applicant information, so uploading a draft to a third-party AI tool creates a real data-privacy exposure that most faculty probably haven't thought through in detail, even though vendors technically disclose data-use terms. Their practical guidance is concrete: get consent from the applicant before using AI on their letter, use institution-approved tools only, strip identifiers where feasible, limit AI use to only the text that actually needs polishing, and disclose AI assistance to the program. They note that failure to disclose carries its own risk, since AI-generated text leaves detectable vocabulary fingerprints, and being caught having used AI without disclosure damages the letter writer's credibility more than the AI use itself would have.

They close with a broader observation worth sitting with: if AI-assisted editing becomes widespread across personal statements, letters, and screening algorithms simultaneously, the narrative components of residency applications may simply stop carrying reliable signal — pushing programs to lean more heavily on directly verifiable measures like standardized assessments and structured interviews. There's no data or outcomes to weigh here either — it's entirely a normative framework. The practical takeaway if you write resident or fellow letters is straightforward: light AI polishing for grammar and flow is probably fine if you proofread rigorously against your own original meaning and disclose it, but treat any AI-driven change to substantive descriptive language — especially anything touching an applicant's personality or competencies — as something requiring your explicit, deliberate sign-off, not passive acceptance.

That covers all four pieces for this month. A quick throughline across the issue: two of these are pure ethics-and-workflow pieces asking you to be more deliberate about sequencing and disclosure in your own practice, one is a pricing-transparency wake-up call about the skin-substitute market you may be interacting with more than you realize, and one is a solid epidemiologic reminder to keep pushing early-life photoprotection messaging. Thanks for listening, and we'll see you next issue.