A Drug Looking for a Second Life

A 57-year-old woman, 4 years postmenopausal, in a stable 25-year marriage, tells you her desire has been absent for years and it is wrecking her sense of self. She has tried couples therapy. She is not depressed. She asks if there is a pill.

That question and the frustration behind it are what PLUMERIA was designed to answer. Flibanserin, a 5-HT1A agonist and 5-HT2A antagonist originally developed as an antidepressant, had already earned FDA approval for premenopausal HSDD on the strength of three trials showing modest gains in desire scores and satisfying sexual events. SNOWDROP, the first postmenopausal RCT, showed statistical significance on all endpoints. PLUMERIA was supposed to be the confirmation.

It never got the chance to be one.

The Trial That Was Abandoned Mid-Flight

PLUMERIA randomized 748 naturally postmenopausal women to flibanserin 100 mg nightly or placebo across 95 North American sites. The mean age was 56, mean HSDD duration 5 years, and 85% were white. The design was textbook. Double-blind, placebo-controlled, 24-week treatment course, with co-primary endpoints of satisfying sexual events (SSEs) and FSFI desire domain score. Women had to be in stable heterosexual relationships for at least a year, willing to engage in sexual activity at least monthly, and scoring 15 or higher on the Female Sexual Distress Scale-Revised.

Then Boehringer Ingelheim decided to exit the flibanserin business entirely. The trial was terminated. At that point, only 45.3% of patients had reached week 16, and a mere 16.4% had completed the full 24 weeks. The prespecified power calculation demanded 420 patients per arm. Neither arm hit that mark. The investigators salvaged what they could, shifting the primary analysis to week 16 using last observation carried forward imputation.

That analytical rescue matters. LOCF assumes a patient's last recorded value persists unchanged. An assumption that systematically biases toward whatever trajectory was emerging at dropout. In a trial where half the population dropped out because the sponsor pulled the plug, not because of lack of efficacy or adverse events, LOCF becomes more noise than signal.

What Moved and What Didn't

The FSFI desire domain score improved significantly more with flibanserin than placebo. A mean change of 0.6 vs. 0.4 (P = .011). On a scale that runs from 1.2 to 6.0, that is a between-group difference of 0.2 points. The FSDS-R total distress score also favored flibanserin (-8.4 vs. -5.4, P = .0002), and FSFI total score showed a significant difference (3.6 vs. 2.3, P = .023).

But the co-primary endpoint, satisfying sexual events, did not reach statistical significance. Mean change was 1.0 SSEs per month with flibanserin vs. 0.7 with placebo. The Patient Global Impression of Improvement showed no difference. The FSDS-R item 13 (distress specifically about low desire) trended toward significance but did not cross (P = .060). The Patient Benefit Evaluation did favor flibanserin (37.1% vs. 29.2% answering yes, P = .018), meaning roughly 8 extra women out of 100 felt they got a meaningful benefit.

Compare that to SNOWDROP, which ran to completion with 949 patients over 24 weeks and hit significance on every endpoint. SSEs (1.0 vs. 0.6), FSFI desire (0.7 vs. 0.4), FSDS-R item 13, FSDS-R total, and FSFI total, all P < .01. Same drug, same dose, same population. The difference between these trials is power, not pharmacology.

A pooled analysis published in the Journal of Sex & Marital Therapy calculated an NNT of 8-9 for postmenopausal women to achieve a desire-domain treatment response. Then questioned whether the threshold defining response was itself clinically meaningful.

The Side Effect Equation

Flibanserin's adverse event profile in PLUMERIA was consistent with previous trials. Insomnia hit 7.7% vs. 3.3% on placebo, somnolence 6.9% vs. 2.2%, dizziness 6.4% vs. 3.5%. Discontinuation due to adverse events ran 10.4% vs. 7.3%. One serious AE of liver function test abnormality was considered drug-related. No syncopal episodes occurred. But the trial did not restrict alcohol, and approximately 37% of participants reported consuming one to three drinks daily at baseline. The post-approval boxed warning about alcohol-flibanserin interactions makes real-world prescribing more fraught than these trial conditions suggest. Drug interactions with CYP3A4 inhibitors, oral contraceptives, and triptans are extensive. The Addyi label lists over 40 clinically significant interactions.

The Industry Fingerprints

This trial was funded by Boehringer Ingelheim, which abandoned flibanserin before it was picked up by Sprout Pharmaceuticals (later acquired by Valeant). The published analysis was written and funded by Valeant employees. Three of five authors had financial ties to the drug's manufacturers. The lead author served as consultant to Sprout. A senior author testified at the FDA advisory committee meeting that approved flibanserin.

The authors argue the SSE endpoint itself is flawed. That counting sexual encounters is a crude proxy for desire, and that the FSFI desire domain is more valid. That argument has merit, and the FDA has since moved away from requiring SSE counts. But you cannot run a trial with a co-primary endpoint, miss it, and then dismiss it in the discussion section as the wrong question.

Clinical Impact

Flibanserin does not have FDA approval for postmenopausal women, and PLUMERIA does not build a case to change that. One underpowered, prematurely terminated trial that missed its co-primary endpoint cannot confirm efficacy. This stands even when read alongside SNOWDROP's more complete but still modest results.

The patient who benefits from flibanserin, if she exists in this population, is a naturally postmenopausal woman with longstanding acquired HSDD, no psychiatric comorbidity, no meaningful alcohol intake, a short medication list, and a stable sexual partner. That is a narrow lane. Even for her, the expected gain is roughly one additional satisfying sexual event per month over placebo, with a 1-in-10 chance the side effects prompt her to stop.

For the physician sitting across from that 57-year-old patient, the honest answer has not changed. We do not have a good drug for this. And this trial, despite what its authors suggest, does not get us closer to one.

Current Landscape

A 19-year-old sits in your clinic, complaining of “brain fog” and a vague sense that people in the hallway are talking about him. He smells of stale tobacco and that unmistakable skunky top-note of cannabis. You ask about his use, and he shrugs: “A few cigarettes, a few bowls a week.”

We’ve long viewed cannabis as the primary suspect in the “Triggering Psychosis” case. Tobacco was usually relegated to a secondary role, a marker of poor health or a crude attempt at self-medication. But a new analysis by Bello and colleagues in *Nature Mental Health* suggests we might be looking at a lethal synergy rather than two isolated habits. They dug into the North American Prodrome Longitudinal Study-2 (NAPLS2) to see if the co-use of these substances is more than just the sum of its parts.

What They Tested

The researchers pulled data from 1,012 participants across eight sites, focusing on 734 individuals already deemed to be at “Clinical High Risk” (CHR) for psychosis. These aren’t just random kids on the street. These are patients already showing the “attenuated symptoms” we lose sleep over.

The investigators didn’t just look for “addicts”—they excluded anyone with a diagnosed substance use disorder in the six months prior to enrollment. This is a study of recreational, sub-diagnostic use.

The first witness to take the stand was baseline symptom severity. As expected, more frequent use of either substance correlated with worse scores across almost every domain: positive symptoms, anxiety, and depression.

But when they grouped the participants into “cannabis only,” “tobacco only,” and “co-users,” the symptoms were surprisingly similar. The co-users weren’t necessarily more symptomatic at the start than the single-substance users.

So, if the baseline symptoms don’t differentiate the co-user, why does this matter? Because the real mystery isn’t how they feel on Monday morning. It’s where they end up two years later. The investigators tracked these patients for 24 months to see who converted to full-blown psychosis.

Analyzing the Data

When looking at cannabis frequency alone, the risk of conversion climbed steadily. The hazard ratio was 1.17 per unit of frequency. Not a massive signal, but it was there.

Tobacco use alone, however, failed to reach statistical significance. If this were a simple linear story, the case would be closed: cannabis is the culprit, tobacco is the bystander.

But the plot twisted when they looked at the interaction between the two. When you combine heavy cannabis use—three or more times per week—with even light tobacco use of fewer than 10 cigarettes a day, the risk doesn’t just add up. It multiplies.

Heavy cannabis and light tobacco co-use produced a hazard ratio of 2.93. For every hundred patients we see in this high-risk bucket who use neither substance, a certain baseline will convert. If we look at a similar hundred who co-use heavily, we are looking at nearly triple the number of patients experiencing a psychotic break.

Why does “light” tobacco use paired with heavy cannabis carry such a punch? One theory is biological priming. Nicotine may enhance the absorption of THC or sensitize the dopamine pathways that we know are already firing too hot in these patients.

Another possibility is that the “co-user” represents a specific phenotype of vulnerability. This might be someone whose brain is uniquely sensitive to the reinforcing effects of multiple substances.

The investigators also looked at the E-values—a measure of how strong an unmeasured confounder would have to be to negate their findings. For cannabis alone, the E-value was 1.61. For heavy co-use, it jumped to 5.31. This suggests the co-use signal is much harder to explain away as mere coincidence or observational noise.

Clinical Impact

The takeaway for our clinics isn’t just that “drugs are bad.” We knew that. The impact is in the nuance of the risk assessment.

We often see a patient who “just smokes a few cigarettes” and “uses some weed” and treat it as a low-priority lifestyle issue. This data suggests that the presence of tobacco in a regular cannabis user is a high-yield red flag.

We must move beyond checking boxes for individual substances. The “co-user” is our highest-risk patient profile. If we want to prevent the transition to chronic psychotic illness, our interventions need to be as multi-modal as the patient’s habits.

The evidence is clear: the synergistic effect of cannabis and tobacco is a potent catalyst for conversion. For the 19-year-old in your clinic, that “and” between his substances is the difference between a manageable prodrome and a life-altering diagnosis. This changes Monday morning.

What clinical strategies do you find most effective for addressing co-use in the adolescent prodrome?

You order a preoperative stress test. The patient has no chest pain, no exertional symptoms, and feels fine. The result comes back: reversible ischemia. The reflex kicks in — catheterization, stent, problem solved.

Except the problem was never solved. Because five consecutive major RCTs over twenty years have tested the ischemia hypothesis — find it, fix it, save the patient — and it has failed every time for hard endpoints.

I wrote this article not because I did this myself, but because I came across a patient in clinic who was about to have a LHC after a benign arrythmia led to an unnecessary stress test which resulted in the planned LHC. The patient and his partner were now preparing for a possible stent placement “so we dont have a heart attack in the future”.

The Trial That Was Supposed to End the Debate

The ISCHEMIA trial (NEJM 2020) randomized 5,179 patients with stable CAD and moderate-to-severe ischemia to an invasive strategy (catheterization plus revascularization) versus optimal medical therapy alone. Unlike COURAGE — where critics complained that anatomy was known before randomization, letting physicians cherry-pick low-risk patients — ISCHEMIA blinded the coronary anatomy with CCTA before randomization. No peeking. Moderate-to-severe ischemia required. 320 sites, 37 countries.

The answer: HR 0.93 (95% CI 0.80–1.08) for the primary composite. All-cause mortality: 145 deaths in the invasive arm, 144 in the conservative arm. One death apart. Nothing moved individually — not CV death, not MI, not cardiac arrest, not stroke.

And 35% of ISCHEMIA participants had no angina at enrollment. Among these asymptomatic patients, the invasive strategy produced a quality-of-life improvement of roughly 0.1 points on the Seattle Angina Questionnaire. Clinically meaningless. The QoL benefit that did exist was driven entirely by patients with daily or weekly angina.

But here’s where the definition games begin. Under the protocol’s stringent SCAI definition, periprocedural MI occurred in 2.7% of the invasive arm. Under the 4th Universal Definition of MI — the one most hospitals actually use — it was 8.2%. That’s a five-fold difference depending on which ruler you pick up. The early hazard from periprocedural MI in the invasive arm (5.3% vs. 3.4% at 6 months) was offset by a late reduction in spontaneous MI, resulting in a wash. But that late reduction may be confounded: half the invasive arm was on DAPT at 12 months versus only a quarter of the conservative arm. Was it the stent or the extra antiplatelet?

The 21% crossover rate in the conservative arm is the other asterisk. One in five “medical therapy only” patients ended up getting revascularized anyway. And the trial still showed no benefit for the invasive strategy. Either the crossover rescued the conservative arm, or revascularization genuinely doesn’t help. Both interpretations point the same direction.

The Precedent Check

ISCHEMIA didn’t arrive in a vacuum. COURAGE (NEJM 2007): no benefit. BARI 2D (NEJM 2009): no benefit. FAME-2 (NEJM 2012): benefit driven entirely by urgent revascularization, not death or MI. ORBITA (Lancet 2018): no benefit on exercise time versus sham procedure. A meta-analysis of 7 RCTs encompassing over 10,000 patients with stable CAD and moderate-to-severe ischemia confirmed the same conclusion: PCI plus OMT offers no benefit over OMT alone for mortality, cardiovascular death, or MI. The signal is unanimous across two decades.

So where does PCI actually win? ORBITA-2 (NEJM 2023) pivoted the question — instead of asking “does PCI prevent death?”, it asked “does PCI relieve angina beyond placebo?” Double-blind, sham-controlled, antianginal medications withdrawn. The answer was yes: angina symptom score 2.9 vs. 5.6 (OR 2.21; 95% CI 1.41–3.47; P0.001). Real symptom relief, beyond placebo effect.

This result will be misused. Someone will cite ORBITA-2 to justify PCI in an asymptomatic patient. But ORBITA-2 enrolled symptomatic patients taken off their medications, with confirmed ischemia (median FFR 0.63). The asymptomatic patient with an incidental finding has no symptoms to relieve. The trial that applies to them is ISCHEMIA, not ORBITA-2.

The Stress Test That Started the Cascade

The false positive rate of the test that triggered this entire sequence is not trivial:

• Exercise ECG: 23–39%. Nearly 1 in 3 positive results may be wrong.

• SPECT nuclear perfusion: 17–30%, driven by soft tissue attenuation and motion artifact.

• Stress echocardiography: 14–21%.

• PET perfusion: ~11% — the best performer.

• Stress CMR: 11–15%.
  

These numbers get worse in low-pretest-probability populations. Asymptomatic patients undergoing incidental or preoperative stress testing are, by definition, low-prevalence. A meta-analysis demonstrated that exercise ECG can only reliably rule in CAD when pretest probability exceeds 80%. For the asymptomatic patient getting a preoperative clearance, pretest probability is nowhere near that.

And here’s the twist that reframes the entire concept: a JACC study by Sinha et al. found that when coronary microvascular dysfunction was added to the reference standard alongside obstructive CAD, the apparent false positive rate of exercise ECG dropped from 31% to 0%. Every single “false positive” had demonstrable microvascular disease on invasive physiological testing. The ischemia was real. The epicardial arteries were clean. A stent would have fixed nothing.

This means the catheterization that follows a “positive” stress test may be answering the wrong question. The patient has ischemia, but it isn’t coming from a lesion amenable to PCI. And is that ischemia affecting their quality of life, and even if you intervene on it, what benefit does it have for the patient? These were questions that the patient and his partner could not answer.

The Math in Plain Terms

At 5 years in ISCHEMIA, cumulative primary endpoint rates were 16.4% (invasive) vs. 18.2% (conservative). If you write a hundred prescriptions for optimal medical therapy and send none of those patients to the cath lab, roughly 18 will have a primary endpoint event. Catheterize all hundred and stent whoever is amenable — roughly 16 events. Two fewer composite events. Zero fewer deaths. And in the first 6 months, the invasive group had more events, driven by periprocedural MI. You have to wait years for the curves to cross, and when they do, the confidence interval still straddles zero.

Under the 4th UDMI definition, 1 in 12 patients in the invasive arm had a procedure-related MI. That’s the cost of admission — and the benefit on the other side of the ledger is a non-significant trend in a composite endpoint.

Clinical Impact

The patient who should not go to the cath lab: asymptomatic, incidental finding of reversible ischemia, no left main disease on CCTA (if performed), preserved LVEF, no recent ACS. This patient gets high-intensity statin, antiplatelet therapy, blood pressure optimization, and lifestyle modification. The 2023 AHA/ACC Chronic Coronary Disease Guideline recommends against routine periodic ischemia testing in asymptomatic patients — and when an incidental positive result does occur, the reflexive march to catheterization is not supported by evidence.

The narrow carve-outs where revascularization retains a role: left main disease ≥50% (CABG, not PCI — excluded from ISCHEMIA). LVEF ≤35% with ischemic cardiomyopathy (CABG per STICH — also excluded). Refractory angina despite maximally tolerated medical therapy (PCI for symptom relief, per ORBITA-2). These exceptions are real. They are also narrow.

The ischemia-centric model — find it, fix it, save the patient — is intuitive and emotionally satisfying. It is also, for stable disease, unsupported by the evidence. PCI relieves angina in symptomatic patients. That is its role. For the asymptomatic patient with an incidental stress test finding, the stent does not save them, and the test that triggered the cascade may not even be telling the truth.

The Bottom Line:

• Current Standard of Care: Acute Otitis Media (AOM) management focuses on analgesia and antibiotics for severe cases, with guidelines explicitly not endorsing steroids. Idiopathic Sudden Sensorineural Hearing Loss (ISSNHL) typically warrants systemic steroid “bursts” or intratympanic rescue.

• What This Study Adds: For AOM, experimental evidence suggests steroids can mitigate structural cochlear damage, but clinical data remains insufficient to change practice. For diabetics with ISSNHL, local steroid therapy (specifically intratympanic methylprednisolone) is significantly more effective and safer than systemic therapy, providing better hearing recovery with minimal glycemic disruption.

• The Fine Print: The AOM systematic review relies on a human sample size of just seven patients. The ISSNHL meta-analysis is dominated by studies from China, which may limit the generalizability of certain local administration techniques like post-aural injection.

Background

Hearing loss is often treated as a binary outcome—either you hear or you don’t—but the pathophysiology is a complex “crime scene” involving inflammatory cytokines and metabolic chaos. We are looking at two distinct scenarios: the common Acute Otitis Media (AOM) and the high-stakes Idiopathic Sudden Sensorineural Hearing Loss (ISSNHL). In both, the cochlea is the victim.

The Sampaio systematic review investigates whether we can prevent permanent sensorineural hearing loss (SNHL) during an AOM episode. Meanwhile, the Wang meta-analysis tackles a clinical dilemma every internist dreads: how to give high-dose steroids to a diabetic patient with sudden deafness without sending their blood sugar into the stratosphere. These papers ask the same fundamental question: Can we put out the fire in the inner ear without burning down the rest of the house?

What they looked at

Let’s look at the “alibi” for steroids in AOM. Sampaio et al. demonstrate that in animal models, bacterial toxins from the middle ear translocate through the semipermeable round window membrane. This triggers a local cytokine storm—specifically IL-1β, IL-6, and TNF-α—that peaks within 72 hours. The theory is solid: steroids should dampen this response.

But when we search for the human evidence, the case falls apart. Out of 15 included studies, only one was a clinical study involving exactly seven humans. While four of those seven improved, we can’t base a global standard of care on a sample size smaller than a standard jury. The “jurisdiction” here is strictly experimental; until we see a large RCT, we are just speculating on animal “testimony”.

The Wang meta-analysis, however, has much more “evidence” to work with: 23 randomized studies and 1,777 patients. The “investigation” focused on local vs. systemic steroids for diabetics with ISSNHL. There is a clear incentive for local therapy here: avoiding the “sugar spike” of systemic prednisone. The researchers found that the local group wasn’t just “not worse”—it was actually more effective.

But look at the “Endpoint Trick.” The meta-analysis uses a “total effective rate,” a composite that includes anything from “recovery” to “effective” (15-30 dB improvement). This is a softer target than total hearing restoration. Furthermore, a massive chunk of this data comes from China, where techniques like post-aural subperiosteal injection are common—a practice that hasn’t quite achieved “detective” status in the West.

The Numbers

In the Wang meta-analysis, the total effective rate for local steroids was significantly higher than for systemic steroids (RR = 1.17, 95% CI 1.11–1.22, P < 0.05).

Let’s break that down into clinic-level terms. If you treat 100 diabetic patients with sudden deafness using local steroids instead of systemic steroids, approximately 12 more patients will achieve a clinically significant hearing improvement who wouldn’t have with the systemic approach.

Even more striking was the PTA (Pure Tone Average) improvement. The mean difference was 6.60 dB (95% CI 3.07–10.14) in favor of the local group. Using a broad analogy, choosing systemic steroids over local ones for a diabetic is like trying to water a single houseplant by turning on the whole-home fire sprinkler system: you might get the plant wet, but you’re going to cause a lot of unnecessary damage to the furniture (in this case, the patient’s metabolic stability).

Wait, look at the adverse reactions. The systemic group had 22 cases of abnormal blood glucose compared to just 5 in the local group. That is a nearly five-fold increase in metabolic “crime”.

Clinical Impact

Acute Otitis Media: For AOM, steroids remain a “person of interest” but are not ready for an arrest. While we know from Sampaio that AOM causes permanent high-frequency hearing loss in children and adults, we lack the “murder weapon”—the proof that steroids stop it in humans. The Verdict: File this under ‘nice to know, changes nothing’ for now. Stick to current guidelines until a real-world trial proves we aren’t just over-treating a self-limiting infection.

Diabetic Sudden Deafness: This changes Monday morning. The Wang meta-analysis provides the “conviction” we need. If a diabetic patient presents with sudden deafness (ISSNHL), systemic steroids should no longer be your first reflex. The Prescription: Intratympanic methylprednisolone (not dexamethasone, as methylprednisolone had the highest effective rate) should be the go-to. It offers better hearing recovery with a fraction of the glycemic risk.

The Final Verdict: The local approach is the superior detective. It targets the culprit in the cochlea while leaving the rest of the patient’s physiology alone. This should never be done in primary care, leave it to your Ear specialist colleagues.

We Almost Missed It Again

We’ve all been there. Last month: 62-year-old woman, restless legs for years, stable on pramipexole 0.5 mg nightly that we inherited from another provider. We were running late, her chart looked clean, we almost refilled it without thinking. Then we checked the date—she’d been on it four years. We asked about daytime symptoms. She’d been having them since last summer. “But that’s just getting older, right?”

It wasn’t. It was augmentation. We caught ourselves before making the same mistake we all made for the first decade of our practice.

That stable patient on pramipexole for four years? That’s not our patient anymore. That’s the clinical trial version—filtered, optimized, and nothing like what walks into your clinic.

Here’s the thing about dopamine agonists and RLS: brain iron deficiency disrupts dopaminergic signaling in the striatum. Dopamine agonists force the system, temporarily relieving symptoms while accelerating the real problem underneath. It’s like pouring gasoline on a fire you’re trying to put out. We’ve been doing it for 15 years because the FDA approved these drugs on 12-week trials—long enough to show benefit, too short to catch the augmentation disaster that shows up at month six.

The AASM just demoted dopamine agonists from first-line to “use with caution” and elevated alpha-2-delta ligands and IV ferric carboxymaltose to strong recommendations. If you’re still starting with pramipexole, you’re practicing 2012 medicine.

What to Start Instead

Start with pregabalin or gabapentin. All three alpha-2-delta ligands now carry strong recommendations with decent evidence. No augmentation risk. We default to pregabalin 75–150 mg at bedtime. The somnolence is real—patients will call about morning grogginess—but it’s predictable and reversible. Not a permanent worsening of their underlying condition the way dopaminergic augmentation is.

One thing the guideline makes explicit that we’ve learned the hard way: prior dopamine agonist use may reduce response to gabapentin enacarbil. A 2019 RCT showed exactly this. The longer we wait to transition patients off dopaminergics, the harder that transition becomes.Text within this block will maintain its original spacing when published

The Iron Study Most of Us Skip

Check iron studies on everyone, using RLS-specific thresholds—not the standard anemia cutoffs. Ferritin ≤75 ng/mL or transferrin saturation <20% means supplement. Ferritin 75–100 ng/mL, go IV only. Morning draw, off iron supplements for 24 hours.

We’ve all assumed “normal iron” meant no role for supplementation. That’s wrong. Brain iron deficiency doesn’t mirror serum levels—MRI and CSF studies show reduced brain iron even when CBC looks clean. We’ve missed this before.

When IV Iron Makes Sense

IV ferric carboxymaltose gets a strong recommendation. Improves symptom severity, sleep quality, and how patients actually function day-to-day. The catch—insurance rarely covers this for RLS alone. You’ll need another diagnosis or a fight. Trying to get it approved for a ferritin of 70 is like explaining to your grandmother why you need WiFi on a plane—there’s a perfectly good reason, but the person making the decision doesn’t believe your problem exists. When it works, though, it works.

Two caveats worth knowing: the iron thresholds come from consensus guidelines that “have not been actually tested in trials”—the AASM’s own words. And while ferric carboxymaltose has solid RCT data, ferumoxytol and low molecular weight iron dextran only have weak recommendations based on single observational studies each.

The FDA Approval Problem Nobody Talks About

Gabapentin enacarbil is now the only FDA-approved medication with a strong recommendation. Pramipexole, ropinirole, and rotigotine—all FDA-approved for RLS—are now conditional against standard use. The approval pathway is completely mismatched to the actual disease. Pivotal trials ran 12 weeks. Augmentation takes years. Of course we approved drugs based on endpoints that missed the most important long-term harm.

What You Do Differently In Clinic

When a patient presents with clinically significant RLS—at least twice weekly with moderate distress—check iron studies first, review aggravating factors (alcohol, caffeine, antihistamines, SSRIs, untreated OSA), and start with pregabalin or gabapentin. Reserve dopaminergics for short-term rescue only: plane travel, cases that won’t budge, while titrating other agents.

If you inherit a patient on long-term pramipexole who’s augmenting, don’t stop abruptly. Add an alpha-2-delta ligand or opioid, taper slowly, and warn them about rebound. And for severe refractory cases—oxycodone ER has a conditional recommendation, and low-dose methadone has real-world registry support. It sounds like using a flamethrower to light a candle until you realize you’ve been using a flamethrower all along. It was just called “escalating doses of ropinirole.”

The Patient We Still Don’t Know What to Do With

Here’s the part that still trips us up: the patient who’s been stable on a dopamine agonist for years, no augmentation yet. Do we switch before trouble hits? The guideline doesn’t say. The data on whether early switching prevents augmentation doesn’t exist. We’ve been watching them like a bomb we can’t defuse.

We still refill pramipexole sometimes—for the patient who refuses change, who’s stable and terrified of rocking the boat. We document the conversation. We set a reminder to ask about daytime symptoms every visit. And we wait for the day they tell us their symptoms started at 2 PM, wondering if it’s just getting older.

What are you doing with those patients—proactively transitioning them, or waiting for augmentation to force your hand?

Leave a comment

References

1. American Academy of Sleep Medicine. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications. Version 3. Darien, IL: American Academy of Sleep Medicine; 2023.

2. Winkelman JW, Armstrong MJ, Allen RP, et al. Practice guideline summary: Treatment of restless legs syndrome in adults: Report of the Guideline Development, Dissemination, and Implementation Subcommittee of the American Academy of Neurology. Neurology. 2016;87(24):2585-2593. doi:10.1212/WNL.0000000000003388

3. Silber MH, Becker PM, Earley C, et al. Willis-Ekbom Disease Foundation revised consensus statement on the management of restless legs syndrome. Mayo Clin Proc. 2013;88(9):977-986. doi:10.1016/j.mayocp.2013.06.016

4. Allen RP, Picchietti DL, Garcia-Borreguero D, et al. Restless legs syndrome/Willis-Ekbom disease diagnostic criteria: updated International Restless Legs Syndrome Study Group (IRLSSG) consensus criteria—history, rationale, description, and significance. Sleep Med. 2014;15(8):860-873. doi:10.1016/j.sleep.2014.03.025

5. Trenkwalder C, Beneš H, Grote L, et al. Prolonged release oxycodone-naloxone for treatment of severe restless legs syndrome after failure of previous treatment: a double-blind, randomised, placebo-controlled trial with an open-label extension. Lancet Neurol. 2013;12(12):1141-1150. doi:10.1016/S1474-4422(13)70239-4

6. Allen RP, Chen C, Garcia-Borreguero D, et al. Comparison of pregabalin with pramipexole for restless legs syndrome. N Engl J Med. 2014;370(7):621-631. doi:10.1056/NEJMoa1300736

7. Earley CJ, Heckler D, Allen RP. Repeated IV doses of iron provides effective supplemental treatment of restless legs syndrome. Sleep Med. 2005;6(4):301-305. doi:10.1016/j.sleep.2004.12.006

8. Grote L, Leissner L, Hedner J, Ulfberg J. A randomized, double-blind, placebo controlled, multi-center study of intravenous iron sucrose and placebo in the treatment of restless legs syndrome. Mov Disord. 2009;24(10):1445-1452. doi:10.1002/mds.22550

9. Earley CJ, Horská A, Mohamed MA, et al. A randomized, double-blind, placebo-controlled trial of intravenous iron sucrose in restless legs syndrome. Sleep Med. 2009;10(2):206-211. doi:10.1016/j.sleep.2008.02.005

10. Trotti LM, Becker LA. Iron for the treatment of restless legs syndrome. Cochrane Database Syst Rev. 2019;2019(1):CD007834. doi:10.1002/14651858.CD007834.pub3

11. Hornyak M, Scholz H, Kohnen R, et al. What treatment works best for restless legs syndrome? Meta-analyses of dopaminergic and non-dopaminergic medications. Sleep Med Rev. 2014;18(2):153-164. doi:10.1016/j.smrv.2013.05.004

12. Garcia-Borreguero D, Silber MH, Winkelman JW, et al. Guidelines for the first-line treatment of restless legs syndrome/Willis-Ekbom disease, prevention and treatment of dopaminergic augmentation: a combined task force of the IRLSSG, EURLSSG, and the RLS-foundation. Sleep Med. 2016;21:1-11. doi:10.1016/j.sleep.2016.01.017

13. Winkelman JW, Bogan RK, Schmidt MH, et al. Randomized polysomnography study of gabapentin enacarbil in subjects with restless legs syndrome. Mov Disord. 2011;26(15):2765-2772. doi:10.1002/mds.23932

14. Kushida CA, Walters AS, Becker P, et al. A randomized, double-blind, placebo-controlled, crossover study of XP13512/GSK1838262 in the treatment of patients with primary restless legs syndrome. Sleep. 2009;32(2):159-168. doi:10.1093/sleep/32.2.159

15. Allen RP, Ondo WG, Ball E, et al. Restless legs syndrome (RLS) augmentation associated with dopamine agonist and levodopa usage in a community sample. Sleep Med. 2011;12(5):431-439. doi:10.1016/j.sleep.2010.09.013

The Bottom Line:

• Current Standard of Care: Clinicians often treat the MRI rather than the patient, assuming that any narrowing in the canal is the culprit for gait instability or weakness.

• The High-Yield Pivot: Look for “non-spinal” clues—bulbar symptoms for ALS, magnetic gaits for NPH, or vision loss for MS. If the symptoms evolve in days (not months), stop looking at the disc and start looking for inflammation.

• The Clinical Compass: Utilize the DFT to screen for “hidden” Parkinsonian balance failure and the Gold Coast Criteria for a high-sensitivity ALS catch.

The Cold Case of the Aging Spine

The mystery isn’t just finding a lesion; it’s determining if that lesion actually committed the crime. In Differentiating Neurologic Disorders from Spinal Conditions, the authors lay out a sobering reality: we are frequently operating on the wrong suspect. Because degenerative changes are ubiquitous in older adults, a “positive” MRI often provides a false alibi for neurodegenerative diseases like ALS or Parkinson’s.

The stakes for a misdiagnosis are life-altering. Performing a laminectomy on an undiagnosed ALS patient often leads to a rapid postoperative decline. We must be clinical detectives, looking for the “fingerprints” that a spine surgeon’s scalpel can’t touch—slurred speech, urinary urgency, or cognitive lag.

Interrogating the Mimics

Amyotrophic Lateral Sclerosis (ALS) is the master of disguise, often presenting with asymmetric limb weakness that looks like a simple radiculopathy. Wait, look at the reflexes. If you see brisk reflexes in a weak, wasted muscle, the spine alibi is dead.

Then there is the inflammatory “Hit-and-Run” squad. Guillain-Barré Syndrome (GBS) hits fast with symmetric, ascending weakness and areflexia, often after a viral “pre-game”. Transverse Myelitis (TM) is just as aggressive, causing sudden motor and autonomic failure. If the case “closes” in less than four weeks, it’s rarely a slow-grinding disc.

The Forensics (Epidemiology & Benchmarks)

The “crime rate” of misdiagnosis is statistically significant in the aging population.

• The 16-Month Lag: Patients first referred to orthopedics or ENT face a massive delay in formal ALS diagnosis.

• The 15-Second Marker: On the DFT, a sit-to-stand time exceeding 15 seconds is a red flag for neurologic failure.

• The Fiber Split: Small Fiber Neuropathy (SFN) targets Ad and C fibers (pain/temp/autonomics), while Large Fiber (LFN) hits proprioception and vibration. The detective’s tip: SFN won’t show up on a standard EMG; it requires a skin punch biopsy to find the perp.

Relying solely on a spine MRI in an 80-year-old is like finding a fingerprint in a busy subway station. Everyone has one, but it doesn’t prove they are the person you’re looking for.

Clinical Impact

This changes the “conviction” process for surgery on Monday morning. We must evaluate the whole “grid,” not just the junction.

The Exact Patient Profile:

• The High-Suspect Candidate: A patient with “stenosis” who also has a shuffling gait, slowed mental processing, or bulbar symptoms.

• The “Pivoting” Case: If the patient’s symptoms are rapidly progressive (GBS/TM) or involve multiple levels that don’t match the imaging (MS), call for a neurologic consultation immediately.

• The Surgical Trap: Parkinson’s patients carry an elevated risk for construct failure and reoperation; they need dopaminergic optimization and the DFT before the OR.

The Final Verdict

We have to stop being “image technicians” and start being “clinical detectives.” The presence of a disc bulge is not an indictment of the spine if the patient’s gait and reflexes are testifying for a neurologic mimic.

🟢 This changes Monday morning. Integrate the Dubousset Functional Test (sit-to-stand and dual-task walking) into your standard preoperative workup for any patient over 65. If they can’t walk and count backward at the same time, the “crime” likely didn’t happen in the spine.

Amyotrophic Lateral Sclerosis (ALS) is the master of disguise, often presenting with asymmetric limb weakness that looks like a simple radiculopathy. But look closer at the “bulbar” clues: dysarthria, dysphagia, or involuntary tongue biting. Wait, look at the reflexes. If you see brisk reflexes in a weak, wasted muscle, the spine alibi is dead. That is a signature of motor neuron disease, not a pinched nerve.

Then there is Normal Pressure Hydrocephalus (NPH), which mimics compressive myelopathy with its shuffling, “magnetic” gait. Unlike spinal disease, NPH often brings two partners to the crime: urinary incontinence and cognitive decline. If the gait doesn’t improve after you “fix” the neck, you’ve likely missed the real suspect hiding in the ventricles.

The Forensics (Epidemiology & Benchmarks)

The “crime rate” of misdiagnosis is statistically significant in the aging population. Parkinson’s affects over 2% of people over 80, and NPH hits up to 6% of the same group. If you aren’t screening for these, you are operating on a bystander.

• The 16-Month Lag: Patients first referred to specialists like orthopedics or neurosurgery face a diagnostic delay of 10 to 16 months for ALS.

• The 15-Second Marker: On the Dubousset Functional Test (DFT), a sit-to-stand time exceeding 15 seconds is a red flag for neurologic failure or high fall risk.

• The 90% Rule: Spinal cord lesions are present in 90% of established MS cases, making it the ultimate mimic of chronic spondylotic myelopathy.

Analogy: Relying solely on a spine MRI in an 80-year-old is like finding a fingerprint in a busy subway station. Everyone has one, but it doesn’t prove they are the person you’re looking for.

Clinical Impact

This changes the “conviction” process for surgery on Monday morning. We must look beyond the scan and evaluate the whole patient.

The Exact Patient Profile:

• The High-Risk Candidate: Any patient with “stenosis” who also has a shuffling gait, slowed processing, or bulbar symptoms.

• The “Pivoting” Case: If the patient’s symptoms are rapidly progressive or involve multiple levels that don’t match the imaging, call for a neurologic consultation immediately.

• The Surgical Trap: Parkinson’s patients carry an elevated risk for construct failure and reoperation; they need dopaminergic optimization before the OR.

The Final Verdict

We have to stop being “image technicians” and start being “clinical detectives.” The presence of a disc bulge is not an indictment of the spine if the patient’s gait and reflexes are testifying for a neurologic mimic.

🟢 This changes Monday morning. Integrate the Dubousset Functional Test (sit-to-stand, stair climb, and dual-task walking) into your standard preoperative workup for any patient over 65. If they can’t walk and count backward at the same time, the “crime” likely didn’t happen in the spine.

The Bottom Line:

• Current Standard of Care: Red light therapy is a mainstream wellness staple for collagen and recovery, but clinicians lack high-fluence safety data for diverse skin types.

• What This Study Adds: High-fluence LED-RL is safe up to 480 J/cm² in Caucasians, but the Maximum Tolerated Dose (MTD) drops significantly to 320 J/cm² for patients with skin of color.

• The Fine Print: Melanin absorbs red light energy; exceeding the MTD in darker skin triggers subepidermal blistering and a ~35% risk of transient hyperpigmentation.

BACKGROUND

When Jennifer Aniston spotlights a “Strength for 50+” ritual involving red light therapy, a niche dermatological tool becomes an overnight “longevity” essential. It feels aspirational, accessible, and—according to the lifestyle blogs—entirely risk-free. But for the practicing clinician, this popularity creates a mystery: Is there a point where “biohacking” becomes “biotoxic”?

The stakes are higher than a missed collagen boost. If your patients follow the “more is better” logic popularized by celebrity endorsements, they might unknowingly cross a line where their own skin physiology turns a therapeutic light into a source of necrotic damage.

THE STUDY

The STARS researchers didn’t just look for the “glow”; they looked for the breaking point. Using a “3+3” dose-escalation design—the same rigorous setup used for experimental oncology drugs—they tested a handheld Omnilux device against a temperature-matched mock control. This “alibi check” ensured that any blistering was due to the photons themselves, not just a hot device.

The interrogation of STARS 1 and 2 reveals a dramatic “Pivot” in risk between populations. While fair-skinned subjects in STARS 2 hummed along at 480 J/cm², the diverse cohort in STARS 1 hit a wall. An African American subject developed a 5mm blister at that same dose—histology confirmed subepidermal clefting and necrotic keratinocytes.

The “Incentive” for clinicians here is to recognize that melanin, usually the body’s ultimate UV shield, behaves differently in the visible light spectrum. It stops being a protector and starts being an absorber.

CAVEAT

Let’s look at the “Hyperpigmentation Tax.” In the MTD cohorts, the incidence of darkening was 33.3% to 36.7% for those receiving active light versus 0% for controls ($p \leq 0.003$).

If you write 100 “prescriptions” for high-fluence red light, approximately 35 of your patients will walk away with transient hyperpigmentation.

Think of it like a “metabolic speed limit.” For fair-skinned patients, the road is clear up to 480 mph. For patients with skin of color, the “radar” (melanin) catches the red light energy at 320 mph. If they push the pedal past that threshold, the result isn’t a better complexion—it’s a clinical crash of blisters and prolonged erythema.

CLINICAL IMPACT

This study provides a necessary reality check for the “biohacking” era. Clinicians should adopt these three rules for Monday morning:

1. Dose for Diversity: Do not recommend universal settings. Cap fluences at 320 J/cm² for patients with skin of color (Fitzpatrick IV-VI) to avoid phototoxicity.

2. Warn of the “Side Effect”: Ensure patients understand that the “collagen glow” carries a 1-in-3 chance of transient hyperpigmentation, particularly in darker skin types.

3. Monitor the Ritual: Advise patients that erythema lasting more than 24 hours is a dose-limiting event, not a sign of “efficacy”.

The Final Verdict: 🟡 Not yet — but I’m paying attention. Red light therapy is a valid clinical tool, but it is not inert. While celebrities focus on the ritual, we must focus on the dose. Darker skin is more photosensitive to this spectrum, and our safety protocols must reflect that 160 J/cm² gap in tolerance.

THE BOTTOM LINE:

• Current Standard of Care: Guideline-directed medical therapy (GDMT) for HFrEF mandates the rapid initiation of four classes: Beta-blockers, RAS inhibitors (ACEi/ARB/ARNI), MRAs, and SGLT2 inhibitors. Expert consensus now favors simultaneous or rapid-sequence initiation within weeks of diagnosis to maximize mortality benefit.

• What This Study Adds: In a real-world cohort of over 50,000 patients, only 21% ever achieved quadruple therapy over nearly three years. For those who did, it took a median of six months—a dangerous “therapeutic inertia” that likely costs lives during the highest-risk period post-diagnosis.

• The Fine Print: This is a VA population (97% male), so the findings may not perfectly map to the general public. Additionally, pharmacy fill data is a proxy for adherence; we don’t know if patients were prescribed the drugs but never filled them, or if clinical barriers like borderline hypotension prevented titration.

THE BACKGROUND

We have the “Four Pillars.” We’ve seen the DAPA-HF, EMPEROR-Reduced, and PARADIGM-HF trials prove, beyond a reasonable doubt, that these drugs save lives. We are no longer debating if they work. The mystery now is why we aren’t using them.

Jacobs et al. set out to track 52,850 patients with incident HFrEF in the Veterans Health Administration. They wanted to see how long it takes to reach the “finish line” of quadruple therapy. You’d think in an integrated system with standardized electronic records and low-cost drugs, we’d be hitting home runs.

But the data tells a different story. It’s a story of a system moving in slow motion while a progressive disease runs a sprint. If you were expecting a victory lap for modern cardiology, look elsewhere.

WHAT THEY DID

The researchers looked at the “time to quadruple therapy” (TTQ). This wasn’t just about getting a prescription; it was about the overlap—the moment all four drugs were actually in the patient’s medicine cabinet at the same time.

The incentives here are fascinating. In the private sector, we blame “prior auths” and fragmented care. But the VA is a closed system. So why did only 21.2% of patients reach quadruple therapy over a median follow-up of nearly three years?

Let’s pause on that denominator. We are talking about 52,850 human beings. Only about 11,000 of them ever got the full cocktail. That’s a 79% failure rate on the primary goal of modern HFrEF management.

The study interrogated the usual suspects: race, sex, and money. They found a plot twist. Black patients (HR 1.22) and Hispanic patients (HR 1.21) actually reached quadruple therapy faster than White patients.

Why? It might be targeted equity initiatives within the VA. It might be “healthy user” bias in specific urban clinics. Or perhaps, in a system where the baseline is this slow, any proactive outreach stands out.

But then we find the fingerprints of the real culprit: the copay. Even in the VA, where “expensive” drugs are relative, patients in “priority groups” who had to pay a small copay were 8% less likely to reach the goal.

SIX MONTHS

Let’s look at the “speed” of this care. The median TTQ was 197 days.

Six months.

In the first six months after a HFrEF diagnosis, the risk of sudden cardiac death and rehospitalization is at its peak. We are essentially giving the fire department a six-month “lead time” to show up after the 911 call.

The Analogy: Imagine you’re building a bridge. We know it needs four main support cables to stay up. We put the first one up on day one. We wait two months for the second. Three months for the third. By the time we get to the fourth cable six months later, the bridge has already sagged or collapsed.

If you manage 100 new HFrEF patients this year, only 21 of them will ever get the full evidence-based treatment they deserve. The other 79 will be left walking on a bridge with missing cables. Of those 21 who “make it,” half will wait longer than 197 days.

Think about the SGLT2i. It’s the newest pillar. In this study, it was the “bottleneck” drug. It’s like we’re still treating it as a luxury add-on rather than a foundational requirement.

CLINICAL IMPACT

This study is a mirror, and the reflection isn’t pretty. It proves that our current “sequential titration” model—start one drug, wait two weeks, check labs, start the next—is a recipe for failure.

The Nuance: The patients who did reach the goal were younger, had diabetes (likely because they were already on an SGLT2i for renal protection or glucose), and were diagnosed in an outpatient setting.

If your patient has CKD, you are likely hesitating. If they were diagnosed during an inpatient stay, surprisingly, they were less likely to hit the goal (14.2% vs 22.2%). That’s an indictment of our hospital-to-home transition. We stabilize them, discharge them on two drugs, and then they get lost in the “we’ll up-titrate later” void.

THE VERDICT

This study doesn’t change the what of our practice, but it should radically change the when.

We need to stop viewing quadruple therapy as a “destination” we reach after a year of careful visits. We need to view it as a “day one” or “week two” mandate. The barrier isn’t just the copay; it’s the clinical inertia of waiting for the “perfect” time to add the next drug.

In HFrEF, “later” often means “never.”

🟡 A mandatory change in urgency. If you aren’t aiming for quadruple therapy by the first follow-up, you’re already behind the clock

Molnupiravir improves recovery rates and reduces persistent symptoms at 3-6 months in high-risk outpatients with COVID-19, with moderate certainty evidence from over 25,000 patients—but we’ve moved from “antivirals save lives” to “antivirals help you feel better sooner” without adequately updating our risk-benefit framework or our patient conversations.

We’ve all written that Paxlovid prescription reflexively, mentally checking drug interactions while the patient waits, assuming we’re preventing hospitalization like we did in 2021. The PANORAMIC trial gives us better evidence for what we’re actually accomplishing now. Molnupiravir improved recovery at 14 days maintained through day 28: 31.8% versus 22.6% with usual care alone. That’s an absolute difference of 9.2%, or an NNT of approximately 11 to achieve one additional recovery. Time to recovery dropped from a median of 15 days to 9 days. The benefit persisted: at 3-6 months, persistent symptoms occurred in 8.5% versus 11.0% (NNT ~40). No mortality benefit emerged, but baseline mortality was 0.04% in the usual care group—too low to show improvement.

Of course the underpowered secondary analysis gets printed—it answers a “hot question” without actually answering it. The PLATCOV trial comparing nirmatrelvir-ritonavir to molnupiravir randomized just 144 participants with zero serious adverse events in either arm, meaning we still have no head-to-head safety data. But it gets cited because we desperately want to know which drug to choose.

Why don’t these patients look like mine?

These PANORAMIC patients are not our patients. They’re the cardiology version of Instagram—filtered, optimized, and nothing like what walks into your clinic. PANORAMIC enrolled 94% with three or more vaccine doses. The mean age was 56. These aren’t the unvaccinated, high-risk outpatients of 2021 where antivirals showed mortality benefit. They’re vaccinated, partially protected, and experiencing milder disease. The trial design adapted to this reality by powering for recovery rather than hospitalization or death—endpoints that would require impractically large sample sizes in this population.

What about the other drugs?

Simnotrelvir-ritonavir also shows clear benefit. The Chinese phase 2 trial showed a median 35.8-hour reduction in time to recovery with high certainty evidence. Adverse events increased: 28.9% versus 21.6% (NNH ~14). The drug isn’t yet widely available in the US, but the mechanism is similar to nirmatrelvir-ritonavir—both are SARS-CoV-2 main protease inhibitors combined with ritonavir boosting.

Nirmatrelvir-ritonavir’s evidence base looks thinner by comparison. The PLATCOV trial showed 70.7% versus 53.6% recovery at 20 days, but this is low certainty evidence from a single small study. A large UK cohort study found adverse events of 1.3% versus 1.0%—a modest absolute increase, but in a population where benefit is less certain. The head-to-head comparison with molnupiravir showed no difference in recovery (70.7% versus 68.8%), with zero serious adverse events in either arm.

Ensitrelvir shows the adverse event signal without clear benefit. At 125mg, adverse events hit 44.2% versus 24.8% with placebo. At 250mg, 53.6% versus 24.8%. Recovery outcomes showed no meaningful improvement. The SCORPIO trials establish that higher doses increase toxicity without proportional efficacy gains.

So what should I actually prescribe?

You don’t need to default to Paxlovid when molnupiravir offers comparable recovery benefit with cleaner pharmacology. Molnupiravir has the best evidence for the outcomes we can now measure—symptom recovery and reduction in persistent symptoms. It’s oral, five days, minimal drug interactions. Nirmatrelvir-ritonavir remains reasonable but carries the well-documented interaction burden with common medications without clear superiority data.

The bigger picture: we’ve shifted from treating a life-threatening disease to treating a symptomatic respiratory infection without adequately updating our clinical framing. Early pandemic antiviral use prevented hospitalization and death. Current antiviral use accelerates return to normal activities and reduces prolonged symptoms. Both are reasonable clinical goals, but they require different conversations with patients. “This could keep you out of the hospital” differs meaningfully from “this might help you feel better a few days sooner.”

For putting this into practice, you should target molnupiravir at high-risk outpatients within five days of symptom onset who want faster recovery—older patients, those with cardiovascular or metabolic comorbidities, the populations where PANORAMIC showed benefit. Set expectations accurately: we’re treating symptoms, not preventing catastrophe. The baseline risk has changed, and our prescribing rationale needs to change with it.

What don’t we know yet?

What’s still unknown? Which drug works better between all three agents with recovery as the primary endpoint. Real-world effectiveness of simnotrelvir-ritonavir outside trial conditions. Whether any antiviral retains mortality benefit in the unvaccinated or severely immunocompromised—populations systematically excluded from recent trials.

What I want to know: Are you still reflexively reaching for nirmatrelvir-ritonavir, and has your prescribing rationale shifted as the evidence base moved from mortality to symptom endpoints?

Leave a comment

References

1. Butler CC, Hobbs FDR, Gbinigie OA, et al; PANORAMIC Trial Collaborative Group. Molnupiravir plus usual care versus usual care alone as early treatment for adults with COVID-19 at increased risk of adverse outcomes (PANORAMIC): an open-label, platform-adaptive randomised controlled trial. Lancet. 2023;401(10318):281-293. doi:10.1016/S0140-6736(22)02597-1

2. Harris V, Holmes J, Gbinigie-Thompson O, et al; PANORAMIC Trial Collaborative Group. Health outcomes 3 months and 6 months after molnupiravir treatment for COVID-19 for people at higher risk in the community (PANORAMIC): a randomised controlled trial. Lancet Infect Dis. 2025;25(1):68-79. doi:10.1016/S1473-3099(24)00431-6

3. Schilling WHK, Jittamala P, Watson JA, et al; PLATCOV Collaborative Group. Antiviral efficacy of molnupiravir versus ritonavir-boosted nirmatrelvir in patients with early symptomatic COVID-19 (PLATCOV): an open-label, phase 2, randomised, controlled, adaptive trial. Lancet Infect Dis. 2024;24(1):36-45. doi:10.1016/S1473-3099(23)00493-0

4. Cao B, Wang Y, Lu H, et al. Oral simnotrelvir for adult patients with mild-to-moderate Covid-19. N Engl J Med. 2024;390(3):230-241. doi:10.1056/NEJMoa2301425

5. Bechman K, Green AC, Russell MD, et al; OpenSAFELY Collaborative. The safety of antivirals and neutralising monoclonal antibodies used in prehospital treatment of Covid-19. J Infect. 2024;89(3):106227. doi:10.1016/j.jinf.2024.106227

6. Yotsuyanagi H, Ohmagari N, Doi Y, et al. Efficacy and safety of 5-day oral ensitrelvir for patients with mild to moderate COVID-19: The SCORPIO-SR randomized clinical trial. JAMA Netw Open. 2024;7(1):e2354991. doi:10.1001/jamanetworkopen.2023.54991

7. Luetkemeyer AF, Chew KW, Lacey S, et al. Ensitrelvir for the treatment of nonhospitalized adults with COVID-19: results from the SCORPIO-HR, phase 3, randomized, double-blind, placebo-controlled trial. Clin Infect Dis. 2025;80(6):1235-1244.

We’ve all been there—staring at the CAP patient on hospital day 5 who’s been afebrile for 48 hours, wondering if we can stop the antibiotics or if cardiology will freak out when they see “only” 5 days on the discharge summary. A hospitalist-led stewardship initiative finally gives us cover to do what we know makes sense: cut median antibiotic duration from 6 to 5 days, drop macrolide use by 37%, and stop harming patients with unnecessary therapy.

The actual lesson isn’t the specific order set they built. It’s that three intervention components work regardless of your EMR vendor: default to short duration, automate oral switch to narrow agents, and have pharmacists reinforce daily. You can do this now without fancy tech.

What They Actually Did

University of Utah Health updated their CAP order set in late 2021 based on emerging evidence that 3-day courses and beta-lactam monotherapy were safe for ward patients. The new defaults: one dose IV ceftriaxone, automatic switch to oral amoxicillin, total 3-day duration, no empiric azithromycin, no routine blood cultures.

The results were immediate and stuck. Total antibiotic duration dropped 8% (adjusted rate ratio 0.92). Patients getting ≤5 days jumped from 14% to 33%. Azithromycin use fell from 62% to 39%. Amoxicillin use rose from 5% to 51%. These weren’t just order set selections—culture changed.

Clinical outcomes didn’t suffer. Thirty-day mortality was not significantly lower (4.9% vs 6.9%). ICU transfers dropped. The combined endpoint of readmission or mortality was unchanged (12.9% vs 13.8%). The readmission rate alone did increase (6.3% to 9.4%), but multivariable analysis found no association with antibiotic duration, amoxicillin use, or azithromycin avoidance—suggesting messy data from hospital expansion and COVID-era disruptions rather than harm from shorter therapy.

Why This Probably Won’t Work at Your Hospital (And Why That’s Okay)

This was a single-center pre-post study, not an RCT. The post-intervention period included 639 additional encounters from hospital growth and spanned ongoing pandemic disruptions. The order set was used in only 38% of cases—meaning most practice change came from pharmacist feedback, education, and workflow nudges rather than the tool itself.

That’s kind of encouraging. It suggests you don’t need perfect buy-in or fancy tech. You need defaults that make the right choice easy and pharmacists who review antibiotics daily. The 38% usage rate is probably the most honest number in the paper—because we all know how these things go. You build the perfect order set, you educate everyone, and then you watch as people free-text their way around it because “my patient is different.”

Of course this worked at an academic center with an antimicrobial stewardship pharmacist rounding daily. These POST-PCI patients are not our patients. They’re the cardiology version of Instagram—filtered, optimized, and nothing like what walks into your clinic at 2 AM when you’re the only doctor in the building.

What You Can Do Tomorrow

1. Prescribe 3-5 days total for non-ICU CAP. The order set recommended 3 days; median duration was 5. Even this modest reduction cut antibiotic exposure meaningfully. Don’t let perfect be the enemy of better—if you discharge on day 4 with 1 day of oral antibiotics to complete at home, you’ve beaten the old 7-10 day standard.

2. Switch to oral amoxicillin early. The study used one dose IV ceftriaxone then oral amoxicillin. This works because Streptococcus pneumoniae—the bug you’re actually treating in most CAP—remains highly penicillin-susceptible. Skip the fluoroquinolones and extended-spectrum agents unless you have specific concerns.

3. Drop routine atypical coverage. Azithromycin went from standard to selective without harm. Reserve macrolides for confirmed Legionella (they kept urine antigen testing) or severe disease requiring ICU. For the ward patient with CURB-65 1-2, beta-lactam monotherapy is enough.

4. Partner with pharmacy—if you have one. You don’t? Then you’re the stewardship team now. Good luck. The order set was the visible intervention, but the daily pharmacist feedback and stewardship presence drove behavior change. If you don’t have embedded stewardship, start with the defaults you can control in your own prescribing.

   Text within this block will maintain its original spacing when published

What Would Make Me Trust This More

For spreading this to other hospitals, you’d want to see this replicated in community settings without a pharmacy team that actually rounds. You’d want cost data—shorter courses save money, but what’s the price tag for implementation? You’d want to know whether patients actually felt better; the study tracked readmissions and mortality, not whether they felt better or were happy with shorter therapy.

The 30-day ED visit bump is classic—cut the antibiotics, watch the patients circle back because we didn’t give them a safety net. You know who shows up in the ED at day 10? The guy who didn’t understand he was supposed to follow up, not the guy who got 10 days of levofloxacin and felt invincible. If you’re cutting antibiotic duration, you may need to compensate with clearer return precautions or earlier follow-up.

The Verdict

This study validates that shorter, narrower antibiotic courses are safe for non-ICU CAP—something European guidelines already recognized and US guidelines are slowly accepting. The specific order set matters less than the principles: default to the minimum effective therapy, automate de-escalation, and have colleagues who reinforce these choices daily.

Your current practice is probably still 5-7 days with empiric azithromycin. You can change that tomorrow for your next ward CAP patient. Start with 3 days total, oral amoxicillin once they stabilize, and skip the macrolide unless you have specific indication. The evidence supports it. Your antibiotic stewardship team will thank you. And your patients will avoid the diarrhea, resistance selection, and drug interactions that come with unnecessary therapy.

What I want to know: Who’s actually doing 3-day CAP courses already, and who’s still writing for azithromycin “just to be safe”? And be honest—how many of those 7-day courses are because you don’t want the specialist to give you that look?

Leave a comment

References

1. Ciarkowski CE, Howard K, Imlay H, et al. Streamlining antibiotic use in community acquired pneumonia: a quality improvement initiative. J Hosp Med. 2026;1-8. doi:10.1002/jhm.70000

2. Dinh A, Ropers J, Duran C, et al. Discontinuing β-lactam treatment after 3 days for patients with community-acquired pneumonia in non-critical care wards (PTC): a double-blind, randomised, placebo-controlled, non-inferiority trial. Lancet. 2021;397(10280):1195-1203. doi:10.1016/S0140-6736(21)00313-5

3. Moussaoui R, de Borgie CA, van den Broek P, et al. Effectiveness of discontinuing antibiotic treatment after three days versus eight days in mild to moderate-severe community acquired pneumonia: randomised, double blind study. BMJ. 2006;332(7554):1355. doi:10.1136/bmj.332.7554.1355

4. Metlay JP, Waterer GW, Long AC, et al. Diagnosis and treatment of adults with community-acquired pneumonia. An official clinical practice guideline of the American Thoracic Society and Infectious Diseases Society of America. Am J Respir Crit Care Med. 2019;200(7):e45-e67. doi:10.1164/rccm.201908-1581ST

5. Wiersinga WJ, Bonten MJ, Boersma WG, et al. Management of community-acquired pneumonia in adults: 2016 guideline update from the Dutch Working Party on antibiotic policy (SWAB) and Dutch Association of Chest Physicians (NVALT). Neth J Med. 2018;76(1):4-13.

6. Vaughn VM, Flanders SA, Snyder A, et al. Excess antibiotic treatment duration and adverse events in patients hospitalized with pneumonia: a multihospital cohort study. Ann Intern Med. 2019;171(3):153-163. doi:10.7326/M18-3640

7. Garin N, Genne D, Carballo S, et al. β-Lactam monotherapy vs β-lactam-macrolide combination treatment in moderately severe community-acquired pneumonia: a randomized noninferiority trial. JAMA Intern Med. 2014;174(12):1894-1901. doi:10.1001/jamainternmed.2014.4887

8. Postma DF, van Werkhoven CH, van Elden LJ, et al. Antibiotic treatment strategies for community-acquired pneumonia in adults. N Engl J Med. 2015;372(14):1312-1323. doi:10.1056/NEJMoa1406330

Look, there’s a visit type that’s becoming standard in my clinic, and I bet you’ve had it at least once this month: the patient who arrives with lab results they ordered themselves—often from companies I’ve never worked with—asking me to interpret findings and recommend next steps.​

This isn’t some future problem we can kick down the road. In a recent survey of about 200 physicians on Doximity, two-thirds of us reported that patients have brought results from direct-to-consumer health tests into our visits. Less than 10% said their institution or professional organization had clear guidelines for how to handle these results or whether to fold them into routine care.​

And the companies behind this? They’re not small players. Function Health offers more than 100 tests for $365 annually and has hundreds of thousands of customers. Quest Diagnostics runs its own consumer business advertising “no doctor visit required”. Hims launched its own testing service in November 2024. Full-body scan company Prenuvo has performed over 100,000 scans, and millions have used 23andMe for genetic assessments.​

Professional medical organizations have expressed concern—and for good reason. Groups like the American College of Radiology and College of American Pathologists have cautioned that direct-to-consumer screenings may not be clinically validated and could lead to unnecessary health anxiety, expensive or harmful follow-up testing, and false positives.​​

So what do we actually do when these results show up in our exam rooms?

Why I’m telling you this matters right now

The Association for Diagnostics & Laboratory Medicine released a position statement in 2024 supporting consumer-initiated testing—but only from CLIA-certified laboratories committed to providing clear guidance for result interpretation. They explicitly call on us as healthcare providers to help patients understand appropriate use of these tests.​

The problem: most of us are making this up as we go. I certainly was.

Dr. Alexa Mieses Malchuk, a family medicine physician in Pennsylvania, told STAT that when patients bring in these tests, she spends “most of the time explaining the background of each test, what it means, and whether it’s a meaningful result in their specific situations in the first place”. That’s exactly what I found myself doing—spending 20 minutes on something that shouldn’t have been ordered at all.​

Dr. Amy Molten, who chairs the American Academy of Pediatrics’ section on advances in therapeutics and technology, put it perfectly: these tests “are often not designed to be used within a clinical care system,” which forces us to act as “interpreters and safety nets”.​

Without a structured approach, these visits eat my time, generate confusion for my patients, and—honestly—can kick off diagnostic cascades that end up harming people.

A four-step framework I actually use now

The National Academy of Medicine and multiple professional societies have laid out principles for judicious laboratory test ordering. I figured out these same principles work when I’m evaluating externally ordered tests:​

1. Do a thorough clinical assessment

2. Think through the probability and consequences of a positive result

3. Talk with patients openly about risks, benefits, and how results actually change what we do

4. Stick to clinical guidelines when they exist​

When a patient brings me external lab results, I’m basically doing retrospective pretest probability assessment—having the conversation that should’ve happened before ordering. It’s like being handed the punchline without hearing the joke.

Step 1: Reconstruct what actually happened

I start by figuring out why the test was ordered and what clinical question it was supposed to answer.​

Three questions I always ask:

• “What made you decide to get these labs?”

• “Were you having specific symptoms, or was this more about prevention?”

• “What did the company tell you these tests would help answer?”

Research on appropriate test ordering emphasizes that we should only order tests when results will actually change clinical management. If there’s no clear clinical question driving the test, there’s usually not a clear evidence-based action on the other side either. Simple as that.​

Step 2: Figure out if the test is even legit

Not all tests deserve equal weight in my decision-making—you know this already.

The Clinical Laboratory Improvement Amendments (CLIA) regulate laboratory operations, and the FDA oversees certain diagnostic tests—but many direct-to-consumer platforms operate in regulatory gray areas. The Association for Diagnostics & Laboratory Medicine says consumers should only use CLIA-certified laboratories.​

I triage external results into three buckets in my head:

1. Clinically validated, guideline-supported tests (lipids, A1c, TSH, CBC, BMP from reputable labs): I fold these into care using established clinical guidelines from USPSTF, professional societies, or UpToDate.​

2. Tests with murky clinical utility (proprietary biomarker panels, “better range” micronutrients, high-sensitivity inflammatory markers ordered without clear indication): These spit out data but we don’t have strong evidence linking them to better patient outcomes.​​

3. Experimental or non-validated tests (biological age scores, proprietary risk algorithms): No established clinical guidelines exist for these, and I could cause harm by acting on them.​

   Text within this block will maintain its original spacing when published

Step 3: Interpret like you normally would (pretest probability and all)

Laboratory medicine experts emphasize that test interpretation depends on pretest probability, whether the assay is actually measuring what it claims, and clinical context.​

For any abnormal finding, I ask myself:

• Does this match my patient’s symptoms, risk factors, and prior results?

• Could preanalytic factors—timing, specimen handling, different reference ranges—explain the result? Direct-to-consumer labs may use completely different reference ranges than my usual lab.​

• Is there actual evidence that acting on this result improves outcomes?

Research on laboratory test use shows that “unnecessary tests with insignificant results do little to reduce patient anxieties”. In fact, they often crank up medicalization and worry. I’ve seen this happen more times than I can count.​

Step 4: Land the plane with one of three decisions

I force myself to end every external result in one of three places:

1. Document but don’t chase it: When tests lack validation, don’t match guidelines, or won’t actually change what I do.​

2. Watch it over time: For borderline findings where following the trend or repeating in my standardized lab might tell me something useful.​

3. Act according to guidelines: When results show something clinically important (new diabetes-range A1c, scary-high LDL, real anemia) that I’d care about regardless of who ordered the test.​

   Text within this block will maintain its original spacing when published

How I talk about this without tanking the visit

Patients order these tests for reasons I actually get: frustration with long wait times, feeling dismissed by previous doctors, slick marketing promising health insights, and genuine desire to take charge of their health.​​

My communication has to acknowledge these motivations while still setting boundaries—it’s a tightrope walk, I know.

I validate first, then reframe: “It makes sense you’d want answers. Let me walk you through which of these results actually guide treatment decisions based on research evidence.”

I name the business model out loud: “Companies selling these tests make money when more testing happens. My job is different—I focus on what’s proven to help you stay healthy.”

I set explicit boundaries: “I’m happy to review results that come from CLIA-certified labs and line up with clinical guidelines. I won’t manage treatment plans based on tests that lack validation or could cause harm.”

Research on patient-physician communication shows that transparent discussion of risks, benefits, and limitations builds trust way better than pretending I have all the answers. People actually respect honesty.​

What I’m doing differently this week (and you can too)

Based on current evidence and professional guidance, here’s what I’m actually changing:

1. I created a standardized intake process: When patients bring external labs, I document the source, whether it’s CLIA-certified, why they ordered it, and my clinical read.​

2. I have templated language ready:

   • “This test isn’t validated for clinical use, so I can note the result but I won’t base treatment decisions on it.”

   • “If this specific test was clinically indicated based on my opinion, let’s repeat this in our lab with standardized methods so we can interpret it, and then track it reliably over time.”

   • “This finding does concern me; here’s what guidelines say we should do next.”

3. I stick to established guidelines: I pull up USPSTF, specialty society recommendations, or Dynamedex when deciding how to act on any lab result—whether I ordered it or not.​

4. I’m doing some patient education upfront: I put together a brief portal message that says: “Thinking about ordering your own labs? Send me a note first” that explains CLIA certification, why clinical context matters, and when self-ordered testing might not help.​

5. I set visit expectations early: If a patient books specifically for external lab review, I tell them upfront: “We’ll focus on clinically relevant findings and I may not be able to discuss every result”.​

The evidence behind why I’m taking this seriously

A 2023 qualitative study of direct-to-consumer laboratory testing companies found wild variability in practices and policies, raising real concerns about quality and appropriate use.​

A 2025 review concluded that “direct-to-consumer laboratory testing cannot function as a standalone diagnostic strategy” and that “effective use requires thoughtful integration into existing healthcare frameworks”. In other words, these tests can’t replace us—they just add complexity.​

The FTC has been urged to go after providers using misleading or fraudulent marketing, and the Agency for Healthcare Research and Quality has been asked to fund research comparing outcomes from consumer-initiated testing versus conventional approaches.​

Professional societies are crystal clear: we have to make sure clinically relevant tests get used appropriately, explain what tests can and can’t do, and protect patients from low-quality or misleading products.​

When I draw a hard line

Some patients really want me to manage treatment plans based on tests I don’t consider valid. When that happens, professional standards back me up on setting boundaries:

“I don’t prescribe based on this test because current evidence doesn’t support its use. If you’d like to pursue this approach, you may need to work with a clinician who practices differently—but I want you to understand the risks.”

This isn’t about my ego. It’s about practicing evidence-based medicine and protecting my patients from potential harm. Full stop.​

What we’re still figuring out

The AHRQ has been asked to fund research comparing outcomes from consumer-initiated testing versus conventional approaches. We don’t yet have solid data on whether these tests improve or worsen long-term health outcomes, healthcare costs, or patient satisfaction.​

Regulatory frameworks keep changing, so I have to pay attention to CLIA requirements and FDA guidance as they come out.​

What I do know: my patients are already bringing these results to appointments, and I need a systematic approach instead of winging it every time. Period.

Bottom line for your practice

Direct-to-consumer lab testing is already here. Two-thirds of us are encountering it. Professional organizations have issued guidance, but most clinics don’t have protocols yet.​​

You can start today by:

• Creating one templated response for external lab results

• Pulling up USPSTF and specialty guidelines for common tests patients order

• Deciding clear boundaries about what you will and won’t act on based on test validity and evidence

This isn’t about controlling patients or dismissing their concerns. It’s about applying the same evidence-based thinking to external tests that I apply to my own ordering: thorough clinical assessment, consideration of pretest probability, honest communication with patients, and sticking to guidelines.​

My goal isn’t to make every number perfect. It’s to keep my patients safe, avoid unnecessary testing cascades, and spend our limited time together on things that actually improve outcomes.

References

1. Ravindranath M. As more patients order lab tests online, doctors don’t know what to do with the results. STAT. Published January 9, 2026. Accessed January 10, 2026. https://www.statnews.com/2026/01/09/patients-ordering-lab-tests-screenings-online-frustrate-doctors/

2. ADLM supports use of direct-to-consumer tests from reputable labs, clinically meaningful results. Association for Diagnostics & Laboratory Medicine. Published September 30, 2024. Accessed January 10, 2026. https://myadlm.org/media/press-release-archive/2024/10-october/adlm-supports-use-of-direct-to-consumer-tests-from-reputable-labs

3. A judicious approach to ordering lab tests. The Hospitalist. Published July 25, 2022. Accessed January 10, 2026. https://community.the-hospitalist.org/content/judicious-approach-ordering-lab-tests

4. Pretorius RW, Dillon J. Overview on ordering and evaluation of laboratory tests. In: StatPearls. StatPearls Publishing; 2023. Updated August 16, 2023. Accessed January 10, 2026. https://www.ncbi.nlm.nih.gov/books/NBK570615/

5. Direct-to-consumer lab testing: help or hindrance for internists? Global RPh. Published December 8, 2025. Accessed January 10, 2026. https://globalrph.com/2025/12/direct-to-consumer-lab-testing-help-or-hindrance-for-internists/

6. Krleza JL, Stavljenic-Rukavina A,Ijos-Marusic V. Accuracy and precision. Acute Care Testing. Published December 31, 2021. Accessed January 10, 2026. https://acutecaretesting.org/en/articles/interpretation-of-laboratory-results

7. Christman MP, Gutiérrez AM, Cota K, Freimuth RR, Monte AA. Policies of US companies offering direct-to-consumer laboratory testing: a qualitative content analysis. BMC Health Serv Res. 2023;23(1):1035. doi:10.1186/s12913-023-10023-4

Target: Diagnosis, risk grading, and treatment of MASLD in adults
What this is: A current update condensing recent society guidelines (AASLD, EASL) and new FDA drug approvals (resmetirom, semaglutide).

Why This Matters Now

The name change is actually sticking. It’s MASLD (metabolic dysfunction-associated steatotic liver disease) now—not NAFLD—and we need to get used to it [Tilg et al, JAMA, 2026]. But honestly, the numbers are what keep me awake. We’re looking at a global prevalence of 30% to 40% [Tilg et al, JAMA, 2026]. If you have a patient with type 2 diabetes in your exam room, there’s a 60% to 70% chance they have this. If they have obesity, it’s 70% to 80% [Tilg et al, JAMA, 2026]. This is officially the most common chronic liver disease worldwide.

And here is the part that keeps me on my toes: we have a specific window to act. Progression is slow, sure, but it’s relentless. It takes about 14 years to advance one fibrosis stage in isolated steatosis, but once they hit metabolic dysfunction-associated steatohepatitis (MASH), that shrinks to about 7 years [Tilg et al, JAMA, 2026]. For the first time, we have drugs conditionally approved by the FDA explicitly to treat MASH with moderate to advanced fibrosis [Tilg et al, JAMA, 2026]. We cannot be the generation of doctors that lets the “fatty liver” wave turn into a “cirrhosis” wave.

What Changed

1. The “positive” diagnosis. We aren’t diagnosing this by exclusion anymore. You need hepatic steatosis—usually by ultrasound—plus at least one cardiometabolic risk factor: obesity, type 2 diabetes, hypertension, elevated triglycerides, or low HDL cholesterol [Tilg et al, JAMA, 2026]. It’s a cleaner diagnostic framework, and it forces us to think metabolically from the start.

2. The MetALD category has arrived. This is the new gray zone for patients who drink more than the strict cutoff but aren’t “alcoholics”—specifically women drinking 140 to 350 g/week and men drinking 210 to 420 g/week [Tilg et al, JAMA, 2026]. This isn’t just “moderate drinking.” Accurate assessment is vital because the disease is driven by both metabolic syndrome and alcohol, and classification relies heavily on patient history, which may be affected by underreporting [Tilg et al, JAMA, 2026].

3. FIB-4 is now mandatory screening. The review explicitly pushes a two-tier screening approach for high-risk groups—diabetics, obesity, metabolic factors [Tilg et al, JAMA, 2026]. Primary care owns Tier 1: calculating the Fibrosis-4 (FIB-4) index using age, AST, ALT, and platelet count [Tilg et al, JAMA, 2026]. If you’re managing diabetics and you’re not calculating FIB-4, you’re missing advanced fibrosis cases that are sitting quietly on your panel.

   1. One caution here: the FIB-4 equation explicitly includes age as a variable, which means it can skew higher in older adults and potentially reduce specificity [Tilg et al, JAMA, 2026]. Don’t let a 72-year-old’s elevated FIB-4 panic you without context.

4. Liver biopsy has been downgraded for routine diagnosis. It remains the criterion standard, naturally, but it’s effectively reserved for when noninvasive tests are inconclusive or to rule out other diseases [Tilg et al, JAMA, 2026]. This is a practical win for primary care—we have better tools now to risk-stratify without subjecting every patient to a needle.

5. “Cryptogenic” is no longer a throwaway label. If a patient has steatosis but zero metabolic traits, don’t just shrug and move on. This is “cryptogenic steatotic liver disease,” and we need to hunt for the zebras: drug-induced liver injury, Wilson disease, or lysosomal acid lipase deficiency [Tilg et al, JAMA, 2026].

   Text within this block will maintain its original spacing when published

The Staging Workflow (And Where Primary Care Fits)

The review simplifies the workflow into a logical sequence. Think about every patient with type 2 diabetes, obesity, or two or more metabolic risk factors [Tilg et al, JAMA, 2026]. Order an abdominal ultrasound—it’s the first-line imaging [Tilg et al, JAMA, 2026]. If steatosis is present, calculate FIB-4 immediately.

Here’s where the numbers matter. A FIB-4 below 1.3 signals low risk of advanced fibrosis, with a negative predictive value of 85% to 90% [Tilg et al, JAMA, 2026]. You can manage metabolic factors and repeat in one to three years [Tilg et al, JAMA, 2026]. A FIB-4 between 1.3 and 2.67 is indeterminate risk [Tilg et al, JAMA, 2026]. Anything above 2.67 is high risk [Tilg et al, JAMA, 2026]. Both of those need the next step: vibration-controlled transient elastography (VCTE), also known as FibroScan, or an ELF test [Tilg et al, JAMA, 2026].

If the VCTE comes back below 8.0 kPa, the probability of advanced fibrosis is low [Tilg et al, JAMA, 2026]. Between 8.0 and 12.0 kPa is indeterminate—continue monitoring closely. At or above 12.0 kPa, refer to hepatology for confirmation and potential treatment. If the VCTE suggests cirrhosis (F4), the job changes instantly. Liver-related mortality and hepatocellular carcinoma risk become the predominant threat [Tilg et al, JAMA, 2026]. This is the “cirrhosis cliff.”​

The bottleneck in the real world is often access to elastography. If FIB-4 is elevated, guidelines demand measuring liver stiffness, but in primary care, getting this approved or finding a slot can take weeks or months. This is where the system breaks down—we identify risk but can’t act on it fast enough.

Management: What Actually Works

The foundation remains lifestyle modification, and the data here is actually quite compelling. Weight loss is dose-dependent. A 5% loss decreases steatosis. A 7% to 10% loss is needed to resolve MASH and improve fibrosis [Tilg et al, JAMA, 2026]. The Mediterranean diet wins here—high fruit and vegetables, whole grains, olive oil [Tilg et al, JAMA, 2026]. Pair that with at least 150 minutes per week of moderate aerobic exercise [Tilg et al, JAMA, 2026].

Alcohol avoidance is a first-line treatment recommendation [Tilg et al, JAMA, 2026]. No bargaining. And don’t ignore the heart—cardiovascular disease is the leading cause of death in these patients [Tilg et al, JAMA, 2026]. Treat lipids and blood pressure aggressively.

The New Drugs: Scalpels vs. Sledgehammers

We have two conditionally approved drugs for noncirrhotic MASH with moderate to advanced fibrosis (F2-F3) [Tilg et al, JAMA, 2026]. They’re not interchangeable coin flips—think of them by phenotype.

Semaglutide 2.4 mg weekly is the sledgehammer. It’s best for the “classic” phenotype: obesity plus type 2 diabetes plus MASH [Tilg et al, JAMA, 2026]. It treats the whole syndrome. In the phase 3 ESSENCE trial, 62.9% of patients achieved MASH resolution compared to 34.3% with placebo [Sanyal et al, N Engl J Med, 2025]. It achieved about 10.5% weight loss, debulking the metabolic risk that drives the liver disease [Sanyal et al, N Engl J Med, 2025].

Resmetirom is the scalpel. It’s a thyroid hormone receptor beta-selective agonist [Tilg et al, JAMA, 2026]. It’s best for the “lean” phenotype or GLP-1 non-responders. It targets liver toxicity without causing significant systemic weight loss—just 1.8% weight change in trials [Harrison et al, N Engl J Med, 2024]. In the MAESTRO-NASH trial, it achieved 25.9% to 29.9% MASH resolution and improved fibrosis in 24.2% to 25.9% of patients [Harrison et al, N Engl J Med, 2024].

Tirzepatide and other multi-agonists are currently in phase 3 trials [Tilg et al, JAMA, 2026]. The pipeline is filling up.

The Real-World Friction

The data is solid—phase 3 trials with clear endpoints [Tilg et al, JAMA, 2026][Harrison et al, N Engl J Med, 2024][Sanyal et al, N Engl J Med, 2025]. But the real-world friction is staging. We cannot prescribe these drugs safely without knowing if fibrosis is present at F2-F3, and we cannot know that without elastography or biopsy [Tilg et al, JAMA, 2026]. The danger—and this is what worries me—is that we treat the “easy” patients with mild fatty liver and expensive drugs they don’t qualify for, while the “quiet” F3/F4 patients sit on our panels unidentified until they crash.

• Screen: Every patient with T2D, obesity, or ≥2 metabolic factors [Tilg et al, JAMA, 2026]

• Calculate: FIB-4 using age, AST, ALT, platelet count [Tilg et al, JAMA, 2026]

• Escalate: FIB-4 >1.3 → order VCTE [Tilg et al, JAMA, 2026]

• Refer: VCTE ≥12.0 kPa → hepatology​

• Treat: Obese/diabetic MASH → consider semaglutide; lean phenotype → consider resmetirom [Tilg et al, JAMA, 2026]

References

1. Tilg H, Petta S, Stefan N, Targher G. Metabolic dysfunction-associated steatotic liver disease in adults: a review. JAMA. 2026;335(2):163-174. doi:10.1001/jama.2025.19615

2. Rinella ME, Lazarus JV, Ratziu V, et al; NAFLD Nomenclature Consensus Group. A multisociety Delphi consensus statement on new fatty liver disease nomenclature. J Hepatol. 2023;79(6):1542-1556. doi:10.1016/j.jhep.2023.06.003

3. Harrison SA, Bedossa P, Guy CD, et al; MAESTRO-NASH Investigators. A phase 3, randomized, controlled trial of resmetirom in NASH with liver fibrosis. N Engl J Med. 2024;390(6):497-509. doi:10.1056/NEJMoa2309000

4. Sanyal AJ, Newsome PN, Kliers I, et al; ESSENCE Study Group. Phase 3 trial of semaglutide in metabolic dysfunction-associated steatohepatitis. N Engl J Med. 2025;392(21):2089-2099. doi:10.1056/NEJMoa2413258

Clinical Takeaway:
Slow tapering (longer than 4 weeks) plus psychological support prevents relapse as well as staying on the medication—everything faster fails at nearly the same rate as stopping cold turkey.

We all have that patient. Stable on sertraline for three years, functioning well, but the moment you suggest tapering off, their eyes widen: “I tried that once. I crashed in two weeks. Never again.” A new network meta-analysis by Zaccoletti and colleagues validates that fear. Analyzing 76 trials with 17,379 patients, they compared every exit strategy: abrupt discontinuation, fast tapers (4 weeks or less), slow tapers (longer than 4 weeks), dose reduction, and each with or without psychological support. The results will change how I write taper plans starting tomorrow.​

What Actually Works

Slow Taper Plus Therapy Is the Gold Standard: Tapering over more than 4 weeks combined with structured psychological support had a relative risk of relapse of 0.52 compared to abrupt stopping (NNT 5.4). That’s statistically identical to just continuing the antidepressant and clinically meaningful—treat five patients this way to prevent one relapse. The therapy studied most was 8 weeks of mindfulness-based cognitive therapy (MBCT), though 7-10 sessions of CBT focusing on relapse prevention also worked.​

Slow Tapering Alone Doesn’t Work: Here’s the part that surprised me. Extending the taper beyond 4 weeks without adding psychological support offered no statistical benefit over abrupt discontinuation (RR 0.81). It’s not just about giving the brain more time—the support is what makes the difference.​

The Fast-Taper Trap: Tapering over 4 weeks or less offered zero benefit over abrupt discontinuation (RR 1.02). We might think we’re being careful with a 2-week taper, but to the brain—and the relapse risk—it’s functionally the same as stopping today.​

The Safety Net Strategy: If full discontinuation feels too risky, continuing at a sub-therapeutic dose (≤50% of minimum effective dose—think 25 mg sertraline or 5 mg escitalopram) outperformed both abrupt stopping (RR 0.62, NNT 6.8) and fast tapering. It provides a psychological and pharmacological cushion.​

Psychological Support Matters—But Only During Tapering: Adding therapy to slow or fast tapers improved outcomes significantly. But combining therapy with abrupt discontinuation or standard continuation showed no statistical benefit, likely because those strategies already have defined trajectories—continuing works, stopping doesn’t.​

Short Follow-Up Limits Long-Term Confidence

The mean follow-up was only 46 weeks. Depression doesn’t quit after a year, so we don’t know if these patients stayed stable at year three or five. Additionally, 79% of the data came from depression trials; anxiety disorder evidence was thinner and didn’t assess all strategies. I’d apply these findings to panic disorder or GAD cautiously until more targeted data emerges. The studies varied moderately in their populations and methods, but sensitivity analyses removing lower-quality trials showed similar results.​

Clinical Bottom Line

• Redefine “Tapering” as Months, Not Weeks: If a patient requests a 2-week taper, I’m explaining the relapse risk honestly—it’s effectively the same as stopping abruptly, and we can do better.

• Prescribe the Exit, Not Just the Pills: I’m referring patients for 8 weeks of MBCT or 7-10 sessions of CBT during tapers. The data shows structured therapy—not just “going slow”—is what prevents relapse.​

• Offer a Micro-Dose Option: For highly anxious patients, continuing indefinitely at 25-50% of their current dose is safer than rushing a full exit and keeps the pharmacological safety net in place.

If we want patients off antidepressants safely, slow and steady with structured support is the only path that works—everything else is just a faster way to the same crash.

EBM PEARL

Network Meta-Analysis (NMA): This wasn’t a standard head-to-head review. NMA lets us compare Strategy A (slow taper) against Strategy C (fast taper) even if they’ve never been in the same trial, by using a common comparator (like abrupt discontinuation) as the mathematical bridge. It creates a ranking system across all strategies—which is how we know fast tapering sits near the bottom with abrupt discontinuation.​

Reading time estimate: 4-minute read

Clinical Takeaway:
Adding T3 to levothyroxine doesn’t relieve persistent hypothyroid symptoms better than levothyroxine alone—but when patients insist they feel better on combination therapy, that gap between their experience and our data matters clinically.

You know that patient—the one who comes back every three months saying they still feel terrible despite a TSH of 2.1? I just finished reading a systematic review of 20 trials that tested whether adding T3 actually helps. Short answer: it doesn’t. Mood, cognition, fatigue, QOL—all came back neutral. But here’s what caught me: lots of patients still preferred combination therapy even when the objective measures showed absolutely nothing. Enough that a third of ATA members admit they’d prescribe it anyway. That disconnect between what we’re measuring and what patients are living through? That’s what we’re actually dealing with here.​

What the Data Shows

• Twenty studies, zero wins for combo therapy: LT4/LT3 versus LT4 alone showed no difference for mood, depression, fatigue, anxiety, QOL, or cognitive function—didn’t matter what T3 dose they used, how long they ran the trial (5 to 52 weeks), or which symptom scale they picked.​

• But patients tell a different story: Despite flat outcomes, many chose the combination when they had the option—like they’re picking up on something our instruments just can’t capture.​

• The studies tested the wrong crowd: Most trials had fewer than 60 people, and only 15% of hypothyroid patients actually complain of persistent symptoms despite normal TSH. So most participants didn’t even have the problem the study was trying to fix.​

• Dosing was all over the map: Ratios ranged from 50:12.5 mcg to weight-based formulas; nobody agreed on how to replace the 5–6 mcg of T3 your thyroid normally puts out each day.​

  Text within this block will maintain its original spacing when published

We’re Chasing the Wrong Diagnosis Half the Time

The bigger issue? We keep calling these persistent symptoms “hypothyroidism” when half the time it’s something else wearing a really good disguise. Fibromyalgia looks like undertreated thyroid disease. So does vitamin D deficiency, sleep apnea, depression—they all show up with the same fatigue, brain fog, weight gain. These trials didn’t systematically rule out the mimics, so who knows if we were even treating thyroid problems to begin with.​

The measurement tools made things worse. Most studies grabbed generic QOL questionnaires off the shelf—nothing thyroid-specific, nothing that accounted for how bad symptoms were or how long they’d been dragging on. When one trial defines “persistent symptoms” at a cutoff of 5 and another uses 8, you’re not comparing the same thing. And nobody factored in cognitive biases—optimism bias, authority bias—that explain why people report feeling better without anything actually changing on paper.​

What This Means Tomorrow

• Start with the checklist, not the prescription pad: Before you blame the thyroid, check vitamin D, B12, ferritin; screen for sleep apnea and depression; make sure they’re actually taking their levothyroxine; ask about fibromyalgia. Write down what you’ve ruled out—otherwise you’re just guessing it’s thyroid-related.

• Recognize the measurement gap when you see it: Patient comes back saying combo therapy changed their life but their labs look identical? You’re not seeing placebo effect—you’re seeing the limits of our tools. Our symptom scales might be too crude to pick up real (but subtle) shifts in cognition and energy that genuinely matter to them.

• Reserve combo trials for the truly miserable after you’ve done your homework: Think about a 3–6 month trial of LT4/LT3 (16:1 ratio, dosed twice daily) only after you’ve excluded everything else and had an honest conversation about shaky evidence. Watch for palpitations, bone loss, anxiety—T3’s short half-life means you get peaks that can throw someone into AFib.​

Twenty trials say it doesn’t work, but patients keep choosing it. Maybe the question isn’t whether it works—maybe it’s whether we’re measuring what actually matters to them.

EBM Pearl
Spot this design flaw the next time you read a negative trial: if you’re studying “symptom relief” but 85% of your participants don’t have symptoms, you’ve built in a null result from day one. It’s dilution of effect in action—any benefit in the symptomatic 15% gets swallowed by noise from the asymptomatic majority. Always ask: did they study the people who actually have the problem?​

Leave a comment

4 minute read

Clinical Takeaway:
For our 60+ patients, the bivalent RSVpreF shot cuts RSV hospitalizations by 83% with a clean safety profile—this is one I’m comfortable routinely recommending.​

EBM Pearl:

Registry-based pragmatic trials like this one show you can keep the rigor of randomization while answering messy real-world questions at scale. Minimal on-site procedures, national administrative data pulling the weight on outcomes—no need to build a massive traditional trial infrastructure.

We finally have a real-world randomized trial where RSV vaccine doesn’t just chase soft endpoints. It actually keeps older adults out of the hospital. The DAN-RSV trial randomized more than 131,000 Danes aged 60 or older to either get the bivalent RSV prefusion F vaccine or nothing during the 2024–2025 winter. Over roughly six months, RSV-related respiratory hospitalizations dropped by 83%: 3 cases in the vaccine group versus 18 in controls.​

Sure, it’s still rare. But when it hits, RSV is the kind of admission that turns a functional 75-year-old into someone who needs rehab.

The Numbers That Matter

• RSV hospitalizations: Vaccinate 1,818 people to prevent one RSV admission over six months (vaccine effectiveness 83%, 95% CI 43% to 97%)​

• Lower respiratory tract disease: Just 1 hospitalization in the vaccine group versus 12 in controls—92% effectiveness (95% CI 44% to 99.8%)​

• All respiratory admissions: 15% fewer hospitalizations for any respiratory illness (284 vs 335 events), which tells me a lot of RSV never gets diagnosed​

Safety profile: Serious adverse events were nearly identical (2.1% vs 2.4%); five were judged vaccine-related, including one Bell’s palsy and one pericarditis, but zero Guillain–Barré in the six-week window​

Why I Trust This More Than Usual

This trial looks like real life. It was open-label, yes—but the outcomes were hard endpoints (hospitalizations, deaths) pulled straight from national registries, so there’s no massaging the data. They invited basically the entire Danish 60+ population through digital mail and got 8.6% to sign up, which feels closer to our clinic panels than your typical phase 3 cohort.​

Now, there’s definitely healthy volunteer bias here. And RSV testing was dismal—only 23% of respiratory admissions got tested. So the true RSV burden? Probably way higher than what shows up in the primary endpoint. The fact that all-cause respiratory hospitalizations dropped by 15% suggests plenty of undiagnosed RSV hiding in the data. But honestly, for what I care about—keeping people off oxygen, out of the ICU, and away from that downward spiral—the signal is strong enough.​

What I’m Doing Differently

• I’m offering RSV vaccination to every reasonably healthy patient over 60, and I push harder in those with COPD, CHF, or a history of bad winters​

• When I counsel patients, I frame it like this: one more layer alongside flu and COVID vaccines, similar safety, but you get at least two years of protection—not another annual shot​

• Since most people have never heard of RSV in adults, I mention there’s no outpatient antiviral and some early data suggesting it may trigger CV events (similar to flu), which tends to land with my high-risk cardiac folks​

This doesn’t feel like splitting hairs over NNTs. It feels like closing one more trapdoor that opens up every winter under our older patients.

4-minute read

Clinical Takeaway:
Most insulin-treated nursing home residents bounce between lows and highs in the same week—deintensification fails because we’re fighting the regimen itself, not just the dose.​

Researchers tracked 12,031 insulin-treated VA nursing home residents during their first 7 days of admission and found something that explains why diabetes management feels impossible in this setting. Sixty percent had hyperglycemia ≥250 mg/dL, 19% had hypoglycemia <70 mg/dL, and—here’s what matters—two-thirds of those with hypoglycemia also hit hyperglycemia in the same week. These aren’t patients who are “too tight” or “too loose.” They’re ping-ponging between both extremes under regimens we thought would prevent these swings.​

Basal-Bolus: Built for a World That Doesn’t Exist

The residents with the wildest swings? They were on combination basal and bolus insulin—60% of the cohort. Seventeen percent hit both hypoglycemia and hyperglycemia within one week; 54% had hyperglycemia alone; only 21% stayed in range. Severe excursions were everywhere: 10% spiked above 400 mg/dL, 8% dropped below 54 mg/dL—rates far worse than residents on simpler regimens.​

You know what basal-bolus demands? Precise meal timing. Consistent carbs. Coordinated injections. It’s choreography—and it collapses when residents skip meals, when staff turnover is high, when appetites shift unpredictably. I mean, I suspect what’s happening is this: we see a high, we add bolus insulin, the patient doesn’t eat, and boom—dangerous low. The regimen itself generates the chaos we’re trying to fix. It’s like trying to balance on a tightrope during an earthquake.​

We Can’t Tell Which Comes First

This is a cross-sectional look, so we can’t say whether aggressive correction of hyperglycemia triggers the lows, or whether fear of hypoglycemia leads us to undertreat highs that then need rescue boluses. The first week is noisy—acute illness, fluctuating intake, staff learning the patient. Still, two-thirds of residents with hypoglycemia also hit hyperglycemia in that same stretch; that’s not occasional mismanagement, it’s baked into the system. We need protocols built for nursing homes, not imported from outpatient endocrinology clinics where patients eat on schedule.​

Clinical Bottom Line

• Simplify before you deintensify. If your nursing home patient on basal-bolus swings between extremes, switch to basal-only insulin (or add non-insulin agents) and aim for symptom control instead of glucose perfection.​

• Prevent lows first; tolerate some highs. A few days at 250–300 mg/dL causes polyuria. One hypoglycemic episode? Falls, fractures, and staff too scared to ever deintensify again.​

• Match the regimen to the reality. Irregular meals, cognitive impairment, staffing shortages—these settings can’t handle complex insulin choreography. Simpler wins.​

The goal isn’t tighter glucose control. It’s preventing the catastrophic lows while tolerating manageable highs. In nursing homes, the real threat isn’t an elevated A1c—it’s the fall that follows a 45 mg/dL glucose at 3 AM.

EBM Pearl
Cross-sectional studies show associations, not causality. This study proves 63% of residents with hypoglycemia also had hyperglycemia, but fingerstick timing can’t tell us which came first. Time-sequenced glucose data (think CGM with timestamps) would clarify whether aggressive bolus dosing for highs triggers the subsequent lows, or whether reactive feeding after lows causes rebound spikes—and that’s what we need to design safer protocols.​

Leave a comment

5-minute read

When a single mega-trial contributes half your meta-analysis patients, you’re basically reading that trial’s results with extra steps—subgroups tell you where it really matters.

Trial sequential analysis says the blood loss reduction is legit, but we haven’t hit the sample size we’d need to call transfusion reduction a done deal.

So imagine you are in the OR for a Whipple on a 62-year-old—jaundice, weight loss, the whole picture—you are likely not thinking “does TXA work in surgery?” I mean, that’s too broad. You’re probably thinking: does it help *this* case, *this* bleeding pattern? General surgery isn’t one thing; it’s thyroids and liver resections and hernias, all bleeding differently. That’s why this paper caught my attention—someone actually tried to look at TXA across the whole spectrum instead of just borrowing data from trauma bays or cardiac ORs.

Quick refresher: TXA blocks plasminogen from turning into plasmin, which stabilizes clots. Sounds good on a whiteboard, but what most surgeons care about is whether it quiets the field, keeps them from calling for more units, and—critically—doesn’t trade bleeding risk for clotting risk. This meta-analysis pooled 26 randomized trials to answer exactly that.

What they actually did

Twenty-six trials, nearly 7,000 patients, covering everything from liver transplants and Whipples to sleeve gastrectomies, hernia repairs, thyroidectomies, and sinus cases. All compared IV TXA to placebo, but—and this matters—dosing, timing, and whether they gave a bolus or infusion varied all over the place. Right there, you know any blanket recommendation is going to be messy.

Primary outcomes: intraop blood loss, transfusion needs, major bleeding. Secondary: VTE, mortality, length of stay. Oh, and nearly half the patients came from POISE-3, so that trial’s voice basically drowns out a lot of the smaller studies.

The pooled numbers (before reality bites)

Across all trials, TXA reduced blood loss by about 36 mL—confidence interval puts it somewhere between 15 and 57 mL. Statistically significant, sure. But honestly? For a big hepatectomy, 36 mL is noise. For a thyroid with persistent mucosal ooze, maybe it matters.

Transfusion risk dropped 25% (RR 0.75)—roughly one fewer transfusion per 17 patients you treat. Major bleeding went down too (RR 0.72), preventing one event per 22 patients. And here’s the relief: no increase in VTE, no mortality difference, length of stay basically the same.

If I stopped reading there, I’d be tempted to just throw TXA into every case. But then you get to the subgroups, and that’s where things fall apart.

Where it works—and where it doesn’t

When they split by surgery type, the story changed fast. Head and neck cases showed a 42 mL reduction in blood loss, and that held up statistically. Makes sense, right? Tonsils, sinus work—it’s all diffuse mucosal oozing where stabilizing microclots can actually quiet things down.

But abdominal surgery? The signal disappeared. Blood loss reduction wasn’t statistically significant, transfusion trends never crossed the significance threshold. Even when they zoomed in on hepatobiliary cases, TXA cut major bleeding but didn’t budge transfusion rates or mortality—and when they looked *just* at liver resections and transplants, even that benefit vanished.

So for a 62-year-old with the Whipple, this doesn’t give most surgeons much. There might be something in the highest-risk hepatobiliary operations, but it’s not a home run. Not even close.

Safety—and where I still worry

Pooled VTE rates looked fine—no increase with TXA. But again, that’s leaning hard on POISE-3 and HeLiX, which gave us most of the clotting events. For your standard elective patient with reasonable comorbidities, I’m fairly comfortable TXA won’t tip them into a PE.

Where should a surgeon pump the brakes? Cirrhosis. Cirrhosis isn’t just “they bleed easily”—it’s a tightrope walk between clotting and bleeding, a whole rebalanced coagulation system that could tip either way. This meta-analysis is mostly noncirrhotic folks, so a surgeon shouldn’t walk away confident about using TXA in someone with decompensated cirrhosis heading for transplant or an emergent ex-lap. The uncertainty there is real, and in this scenario, they/you should make those calls case by case than follow a protocol built on averages.

What I’m recommending doing tomorrow

I’m not a surgeon but consider:

• **Head and neck cases**—thyroid, sinus, oncologic resections: yeah, use TXA. The bleeding reduction is consistent, and the safety profile looks solid.

• **Routine abdominal work**—colectomies, standard hernias, straightforward sleeves: nah, don’t see enough benefit to bother. May want to stick with good technique and standard blood conservation.

• **Complex HPB surgery**: maybe. High-risk resections, You may consider it selectively, but don’t expect miracles—and keep in mind that the big outcomes (transfusion, survival, LOS) aren’t moving much.

You know what this paper really tells me? General surgery is way too broad to slap one drug on every case and call it evidence-based. The more I match TXA to the actual bleeding biology—mucosal ooze versus vessel injury, cirrhotic versus normal coags—the more useful this gets. Otherwise it’s just another positive meta-analysis I cite in a note and never think about again.

⏱ 4-minute read

Reference: [Yan et al., Ann Intern Med, 2026]

📌 Clinical Takeaway:
Statins reduce mortality and cardiovascular events in type 2 diabetes across all baseline risk levels—including patients with 10-year cardiovascular risk below 10%.

We’ve known for years that statins prevent cardiovascular disease and death in patients with type 2 diabetes, but there’s always been this nagging question: what about the people whose 10-year risk calculators spit out reassuringly low numbers? Yan and colleagues tackled exactly that, pulling data from over 400,000 UK primary care patients and sorting them by QRISK3 scores. The answer? Statins worked even in those with baseline risk under 10%—and the benefit held across every risk category.

What the Numbers Actually Mean

Let me break this down in ways that matter when you’re sitting across from a patient.

In the low-risk group (QRISK3 <10%), starting a statin dropped 10-year absolute risk by 0.53% for all-cause mortality and 0.83% for major CVD. Translation: you’re treating roughly 189 people to prevent one death and 120 to stop one major cardiovascular event. Small numbers, sure—but they add up when you’re talking about younger patients with decades ahead of them.

Move up to intermediate risk (QRISK3 10-19%) and the absolute risk reductions jump to 1.88% for mortality (NNT ≈53) and 2.14% for major CVD (NNT ≈47). In the high-risk range (20-29%), you see reductions of 2.74% for mortality (NNT ≈36) and 2.59% for CVD (NNT ≈39). And for those in the very high-risk category (≥30%), the gains are biggest—4.30% reduction in mortality (NNT ≈23) and 4.57% for CVD (NNT ≈22).

Subscribe now

On the safety side, myopathy stayed rare across all groups, liver dysfunction showed no signal, and muscle pain only emerged in sensitivity analyses. One thing worth noting: women with T2DM faced higher absolute cardiovascular risk than men, which meant they got more absolute benefit from statins. Given the well-documented sex disparities in statin prescribing, this gives us one more reason to close that gap.Text within this block will maintain its original spacing when published

The Observational Caveat

This was a target trial emulation study, not an RCT. The investigators used propensity-score matching and inverse probability weighting to control for confounding, but residual bias is still possible—think unmeasured factors like diet, exercise, or patients quietly stopping their statin without telling anyone. Hospitalized events like MI or stroke might’ve been undercounted, though that probably affected both groups equally. The cohort was also predominantly White and UK-based, so how well QRISK3 performs in other populations remains uncertain.

That said, the findings lined up well with data from the CARDS trial, and negative control outcome analyses didn’t show red flags for unmeasured confounding. The consistency here gives me confidence these associations are real.

What This Means for Your Practice

I’m taking three things from this study:

• Don’t wait for the risk calculator to hit some magic threshold. Start statins in T2DM patients with dyslipidemia regardless of their short-term predicted CVD risk. The absolute benefit scales with baseline risk, but even low-risk patients gain meaningful protection over 10 years.

• Set realistic expectations about timing. In the lowest-risk group, mortality curves didn’t separate until year four. That’s a long runway—patients need to stick with therapy, and we need to help them understand why it’s worth it.

• Have a backup plan for muscle symptoms. Myopathy is rare, but when patients develop symptoms—real or perceived—therapy falls apart fast. Keep ezetimibe and bempedoic acid ready as alternatives.

EBM Pearl
Target trial emulation uses observational data to mimic an RCT by cloning patients into treatment arms at decision points and censoring those who deviate from their assigned strategy. This study compared “initiation” vs. “non-initiation” strategies—not just “users” vs. “non-users”—which sidesteps immortal time bias and mirrors how we actually make treatment decisions in practice.

Leave a comment

**3-minute read**

The UPF–adenoma signal persisted after adjusting for a detailed diet quality score (AHEI-2010), which tells me it’s not just *what* my patients eat but *how heavily it’s processed*.

Seeing a clear association with conventional adenomas but not serrated lesions reminds me to separate these pathways when thinking about risk and mechanisms—not lump “polyps” together.

The Pattern I Kept Missing

When a 42-year-old comes in for her pre-colonoscopy visit, my mental checklist is still pretty standard. Family history? No. Lynch syndrome? No. BMI 24, runs four times a week, doesn’t smoke, takes a multivitamin. Low risk, right?

But then we talk about food. Flavored yogurt cup for breakfast. Protein bar mid-morning. Frozen bowl—one of those “healthy” ones—for lunch. Diet soda all afternoon. Microwaveable stir-fry kit for dinner. She’s really trying. But nearly everything comes in a wrapper.

That’s where Wang and colleagues made me stop. Early-onset colorectal cancer (EOCRC)—before age 50—has been climbing since the ‘90s, especially in people born after 1960. And you know what else climbed in parallel? Ultraprocessed foods. They now make up nearly 60% of what Americans eat, though this cohort of nurses was closer to 35%. Hard to ignore that timing.

Why This One Got My Attention

We dropped the screening age to 45, which feels proactive. But that’s still reacting to lesions that have been forming for years. What caught me here is they didn’t just look at cancer—they looked at the precursors: adenomas and serrated lesions diagnosed before 50. That’s the window where I can actually change something.

They followed over 29,000 women from the Nurses’ Health Study II who filled out diet questionnaires every four years and had at least one colonoscopy before hitting 50. Polyp type and location were confirmed from actual medical records and path reports—not just patient recall. UPFs were classified using the NOVA system: industrially produced, ready-to-eat or heat products, usually low in fiber and bioactives, high in additives and emulsifiers.

What They Actually Found

As I went through the results, what stuck with me was how the UPF–adenoma link held up, even when they threw everything at it.

UPFs made up about a third of daily calories—median 5.7 servings a day. Mostly refined breads, breakfast foods, sauces, condiments, and sweetened drinks (both sugar and artificially sweetened). Women in the highest quintile of UPF intake had 45% higher odds of early adenomas compared to the lowest (adjusted OR 1.45, 95% CI 1.19–1.77). The dose-response wasn’t linear, though—it plateaued around 7 to 8 servings per day.

The signal showed up in proximal and distal colon, stronger distally (OR 1.44 for distal colon-rectum), which mirrors where we’re seeing the EOCRC surge clinically. And it wasn’t just the big, scary polyps. Small adenomas (<1 cm), low-risk lesions—these also tracked with UPF intake. That tells me we’re catching earlier steps in the pathway.

But serrated lesions? Nothing. OR 1.04, confidence interval crossed 1.0.

They adjusted for age, endoscopy timing, BMI, T2D, physical activity, smoking, alcohol, NSAIDs, fiber, folate, calcium, vitamin D, total calories, and the AHEI-2010 diet quality score. The UPF signal didn’t budge. Practically speaking, if I follow 20 similar patients over two decades—high UPF intake versus low—I might see one extra adenoma case, assuming this is causal.

The Biology Actually Makes Sense Here

Conventional adenomas arise through the classic adenoma-carcinoma sequence—APC hits, KRAS mutations, chromosomal instability piling up over time. Serrated lesions go a different route: epigenetic changes, CpG island methylation, probably needing longer exposures to show an effect. The fact that UPFs linked only to adenomas fits with them accelerating that classic pathway specifically.

And here’s where it gets interesting. Recent genomic work shows colibactin-associated mutational signatures—think genotoxic *E. coli* strains—are 2.5 to 4 times more common in EOCRC, especially distally. UPFs are loaded with emulsifiers and artificial sweeteners that mess with gut barrier function and reshape the microbiome. Could they be enriching colibactin-producing bacteria early in life, leaving mutational “scars” that push the adenoma pathway forward faster? Maybe. It’s speculative, but mechanistically it tracks.

Then there’s the calorie piece. Controlled feeding studies show UPF-rich diets drive about 500 extra kcal/day and rapid weight gain. Obesity nearly doubles EOCRC risk. You’d think BMI would explain a lot of this. But it didn’t modify the UPF–adenoma association in subgroup analyses, and adjusting for BMI and T2D didn’t kill the signal. So weight matters, sure—but even normal-weight people eating lots of UPFs seem to be at risk.

No single food group explained everything. Artificially sweetened beverages showed a modest bump (OR 1.21, highest vs lowest quintile). But overall, it looks more like a cocktail effect—multiple additives acting together rather than one villain ingredient.

The Caveats (Because There Always Are)

Look, this is observational. Food-frequency questionnaires don’t tell me whether someone’s “bread” is homemade sourdough or a shelf-stable sandwich loaf with 15 ingredients. Misclassification is real and probably blunts any true effect. The cohort is mostly White, health-conscious nurses whose UPF intake (35% of calories) is well below the national average. And they couldn’t look at actual EOCRC incidence—not enough cases yet.

Still. The adenoma signal was consistent across sensitivity analyses, including when they restricted to lesions diagnosed before age 45 to preserve a polyp-to-cancer window. That consistency makes it hard for me to just shrug this off.

What I’m Doing Differently Now

When I see someone under 50, especially hovering near that screening threshold, I’ve started asking different questions.

Not “Do you eat fruits and vegetables?” Everyone says yes to that. Instead: “Roughly what percentage of what you eat comes out of a package?” It’s a quick read on UPF exposure, and honestly, people know the answer immediately.

When I’m talking about EOCRC risk, I frame UPFs as their own thing—separate from obesity, separate from fiber deficiency—because the data suggest processing itself matters. And for patients with heavy UPF patterns but otherwise average risk? I push the colonoscopy conversation harder, knowing there’s real adenoma data behind it.

The shift for me is simple. The colon isn’t just counting calories; it’s reacting to what we ask it to process day after day. I still care about fiber and red meat—but now I’m just as likely to end the visit asking: “Does your pantry look like you cook, or mostly just heat things up?”

6 minute read

When I look at the workforce data, counselors are more likely than psychologists or social workers to be physically present in high-need neighborhoods, but reimbursement rules and insurance participation still determine whether that presence translates into actual access.

The Referral That Goes Nowhere

I’ve lost count of how many times I’ve handed a patient a list of therapists, only to have them return weeks later with the same problem and zero appointments scheduled. Ten calls, no callbacks. Six-month waitlists. Practices that stopped accepting their insurance. It’s a pattern I see constantly.

Federal data estimate that about 47% of the U.S. population—roughly half the country—lives in a designated mental health workforce shortage area. That’s not just a statistic; it means my discharge plans often land in communities where waitlists stretch for months, agencies are chronically understaffed, and school counselors or primary care teams absorb way more risk than they should.[2][3]

What’s Actually Driving This

Look, this shortage isn’t some overnight phenomenon triggered by the pandemic. It’s the predictable result of a system that’s treated mental health as optional for decades. When I look at the workforce data alongside what I see clinically, a few things jump out.[3][2]

1. **Chronic underpayment.** Mental health gets reimbursed like it’s an optional add-on to “real” medicine. Medicaid rates for behavioral health consistently trail Medicare and commercial plans—and psychiatric services often see particularly low rates. So when Medicaid expansion brought more insured patients, psychiatrists’ acceptance of Medicaid didn’t meaningfully budge. Honestly, if I were a new graduate deciding where to practice, those fee schedules would factor heavily.[2][3]

2. **Stagnant pay and poor retention.** Colleagues in counseling education tell me that in some community agencies, starting wages for master’s-level therapists were $18/hour in 2009—and still $18/hour in 2017. Same hourly rate, higher living costs, more student debt, less guaranteed supervision. I mean, I get why new clinicians choose private practice, telehealth-only panels, or just leave the field altogether.[3]

3. **Unsustainable caseloads.** In parts of rural western Colorado, one school-based clinician is responsible for roughly 470 people. When that’s your denominator, every new referral feels like a crisis you may not have capacity to manage. Similar patterns show up across rural and frontier areas—long distances, limited backup, and the constant worry that you’re one resignation away from losing the last therapist in town.[1][3]

4. **Geographic maldistribution.** A national analysis using the Area Deprivation Index found roughly 351 behavioral health clinicians per 100k people in the least disadvantaged communities, but only about 267 per 100k in the most disadvantaged areas. Psychologists and social workers were especially scarce in high-need neighborhoods; counselors were somewhat more common there—but until very recently, many counselors couldn’t bill Medicare at all. So even when a clinician is physically present in a high-need zip code, payment rules may prevent my Medicare or Medicaid patients from seeing them.[1]

Who Gets Left Behind

In clinic, the shortage doesn’t show up as “lack of FTEs.” It shows up as patients who get stuck. Older adults with depression and chronic disease living an hour from the nearest psychologist. Adolescents with escalating self-harm whose families have insurance but can’t find a child therapist taking new patients within a reasonable drive. They cycle through the ED, get admitted, and are discharged with recommendations for outpatient therapy that simply doesn’t exist locally.[3]

The data back this up: youth and adults in Mental Health Professional Shortage Areas have higher suicide rates, and that gap widened between 2010 and 2018. Rural communities and socially disadvantaged neighborhoods are more likely to have zero behavioral health clinicians nearby—and when clinicians are present, they’re stretched thin.[2][1]

Layer in race and culture, and the picture gets sharper. Black clinicians make up a small slice of the mental health workforce nationally, and Black patients who want a racially concordant therapist often run into even longer waitlists or travel distances. Given what we know about concordance and engagement, that’s not a small detail—it’s another way the shortage amplifies existing inequities.[2][3]

Coverage Doesn’t Equal Access

Here’s something that caught me off guard: expanding insurance coverage alone hasn’t been enough to “fix” access. A California analysis of adults with serious psychological distress (SPD) found that uninsured rates dropped under the ACA in both shortage and non-shortage areas, but changes in actual mental health service use were modest and didn’t differ by shortage status. Specialty mental health visits did rise in non-shortage areas, but not enough to close gaps.[2]

That’s consistent with what I see. Insurance cards are necessary but not sufficient. If a patient has a plan that nominally covers therapy, but there are no local clinicians taking that plan—or the nearest one has a three-month wait—the effect is the same as being uninsured.

The distinction here is between coverage and capacity. Coverage changes who could be seen. Capacity determines who actually gets seen, when, and by whom. Without deliberate workforce and payment reform, coverage expansions alone won’t meaningfully shorten the line.[1][2]

What I Do Differently Now

I can’t solve the workforce shortage from my clinic room, but I’ve adjusted how I practice in light of it—and I think that’s where these data become clinically useful.

When I’m about to say, “Just get a therapist,” I pause and ask about geography, transportation, language, and prior attempts to find care. For someone in a rural, high-disadvantage area, I assume access will be constrained and build more structured follow-up into my own plan.[3][1]

I lean harder on integrated models—embedding brief behavioral interventions in primary care, using telehealth options that are genuinely accessible, and collaborating with school-based or community counselors even when they’re outside my usual referral network.[1][3]

And I support policy changes that reflect where clinicians actually are. Expanding reimbursement to licensed counselors and marriage/family therapists—especially in Medicare and Medicaid—taps into a workforce that’s already overrepresented in high-need communities. Loan repayment, rural stipends, telehealth-friendly regulations—these aren’t abstractions anymore; they’re tools that may change whether my next patient finds someone to call back.[1]

Clinical Bottom Line

1. **Treat referral access as a clinical variable:** When patients can’t access therapy, outcomes worsen. Factor waitlist times and geographic barriers into your management plan.[3][2]

2. **Push for payment parity and billing reform:** Support policies that expand reimbursement for all licensed clinicians and raise Medicaid rates to match the complexity of behavioral health care.[2][1]

3. **Lean on telehealth and integrated care:** These approaches extend reach and reduce burnout, especially in underserved areas. Push for policy that sustains them beyond the pandemic.[3][1]

References

1. Kim LH, Hodgkin D, Larson MJ, Doonan M. Changes in mental health services use under the ACA in one state: role of mental health provider shortages. Psychiatr Serv. 2025;76(2):149-156. doi:10.1176/appi.ps.20230628

2. Lombardi B, de Saxe Zerden L, Jensen T, Galloway E, Gaiser M. Behavioral health workforce distribution in socially disadvantaged communities. J Behav Health Serv Res. 2024;51:168-179. doi:10.1007/s11414-024-09897-0

3. A closer look at the mental health provider shortage. Counseling Today. May 2023. Accessed January 1, 2026. https://ct.counseling.org/2023/05/a-closer-look-at-the-mental-health-provider-shortage