This Week in General Medicine — Aug 31, 2026

Welcome to This Week in General Medicine. This week we're covering 10 notable papers spanning cardiology's ongoing habit of questioning its own long-standing practices, prevention and adherence at the system level, and a review of where psychiatry is heading with schizophrenia. Let's dive in.

We'll start with a run of trials that all ask the same uncomfortable question: are we doing things because they work, or because we've always done them? The clearest example comes from the New England Journal of Medicine, where Bundgaard and colleagues report POET II, an international open-label trial in 508 adults with left-sided infective endocarditis caused by Staph aureus, Enterococcus faecalis, or streptococci [1]. Everyone received at least two to four weeks of antibiotics and had to be clinically stabilised before randomisation, at which point one group simply stopped and the other completed a standard four to six weeks. The tailored group gained about thirteen extra days alive and off antibiotics, and the safety composite of death, unplanned cardiac surgery, or symptomatic embolism met noninferiority, at roughly eight percent versus eleven percent. But relapse of bacteraemia or endocarditis was about three times more common with the shorter strategy, five percent versus under two percent, and that difference was statistically significant. So this is a genuine trade-off rather than a clean win: fewer antibiotic days, more relapses. For a stable patient who has responded well, a shortened course is now defensible, but it demands a plan for follow-up and a low threshold for re-culturing.

The de-prescribing theme continues in The Lancet, where Silvain and colleagues pooled individual patient data from the ABYSS and SMART-DECISION trials, just over six thousand stable patients more than six months out from a myocardial infarction, with ejection fraction at or above forty percent and no heart failure [3]. Median time since the infarct was three and a half years. Stopping the beta-blocker met noninferiority for the composite of death, infarction, stroke, or cardiovascular hospitalisation, and also for the harder secondary endpoint of death, infarction, or heart failure hospitalisation, which occurred in about six and a half percent of patients in each group. The authors are appropriately cautious, noting the primary composite was hospitalisation-heavy, but the practical message is that for a stable post-infarct patient with preserved function and no other indication, indefinite beta-blockade is not obligatory.

Also in The Lancet, and perhaps the most provocative result of the week, is PVI-SHAM-AF from Wachter and colleagues, a genuinely double-blind, sham-controlled trial of catheter ablation for symptomatic atrial fibrillation across nine sites in Germany and Poland [4]. Two hundred sixty-two patients were randomised two to one to ablation or a sham procedure. Quality of life improved substantially in both arms, by roughly twenty points with ablation and roughly sixteen with sham, and the between-group difference of under three points was not statistically significant. Ablation did not demonstrate superiority over sham for atrial fibrillation-related quality of life at six months. That is a striking result given how confidently we counsel patients about symptom benefit, and it suggests a large placebo component. It does not address rhythm outcomes or the higher-burden persistent population in detail, and twelve-month follow-up is pending, but it should temper the way we frame expectations in the clinic.

Two trials this week actually strengthen the case for intervention. In the New England Journal of Medicine, the SINGLE-AF trial from Kim and colleagues randomised just over eighteen hundred South Korean patients with atrial fibrillation at intermediate stroke risk, meaning a CHA2DS2-VASc score of one in men and two in women, to a direct oral anticoagulant or nothing [2]. Mean age was sixty years. At twenty-four months the composite of stroke, systemic embolism, major bleeding, or cardiovascular death occurred in four patients on anticoagulation versus thirteen without, an absolute difference of about one percentage point, driven mainly by stroke. Major bleeding looked similar between groups and there were no cardiovascular deaths at all. Event rates are very low, so the number needed to treat is large, but this is the first randomised evidence supporting what has been a class IIa recommendation, and it nudges that borderline conversation toward treating.

The second positive signal is ACACIA-HCM, also in the New England Journal of Medicine, testing the cardiac myosin inhibitor aficamten in over five hundred patients with symptomatic nonobstructive hypertrophic cardiomyopathy, a condition with no proven medical therapy [6]. At thirty-six weeks both dual primary endpoints were met, but the margins are modest: about a three-point advantage on the Kansas City questionnaire and a peak oxygen uptake difference of under one millilitre per kilogram per minute. Meanwhile about one in ten patients on aficamten had a reversible drop in ejection fraction below fifty percent, versus under one percent on placebo, and serious adverse events were more frequent. Statistically positive, clinically incremental, and it will require echo surveillance.

Contrast that with CARDIO-TTRansform, reported by Fontana and colleagues in the New England Journal of Medicine, in which over fourteen hundred patients with transthyretin amyloid cardiomyopathy received eplontersen or placebo every four weeks for one hundred forty weeks [7]. Despite the antisense oligonucleotide's clear mechanism, the primary composite of cardiovascular death and recurrent cardiovascular events was not reduced; events occurred in about twenty-nine percent of the eplontersen group and thirty-two percent on placebo, and the difference was not statistically significant. Cardiovascular deaths were numerically slightly higher with the drug. This is a negative trial, and it is a reminder that knocking down transthyretin production does not automatically translate into event reduction in this population.

Turning to how care is organised, PRESC1SE-MI in The Lancet is a pragmatic stepped-wedge cluster-randomised trial across nineteen hospitals in ten countries, covering nearly sixty-eight thousand emergency presentations with suspected myocardial infarction [5]. Switching from a zero-to-three-hour high-sensitivity troponin pathway to the guideline-recommended zero-to-one-hour pathway was safe, meeting noninferiority for thirty-day death or new type 1 infarction at just over one percent in each group. But median emergency department length of stay was identical at 309 minutes. Faster troponin does not equal faster discharge if the rest of the department's workflow is the bottleneck. If you are implementing this pathway, do not promise your administrators throughput gains.

On prevention, JAMA reports ADHERE-ASCVD from Bhatt and colleagues, a pragmatic trial embedded in Kaiser Permanente Northern California enrolling over twenty thousand adults with established atherosclerotic disease or high risk plus recent statin nonadherence [9]. Digital outreach by portal message, text, or email raised fourteen-day refill rates from about twelve to about fourteen percent, and a second outreach to nonresponders produced a further gain. Switching modality added nothing over simply repeating the same one. By twenty-eight days about a quarter of patients receiving outreach had refilled versus about a fifth with usual communication. Small absolute effects, but at population scale and near-zero marginal cost, that is a reasonable systems investment. A more intensive approach fared less impressively in JAMA Internal Medicine, where Leong and colleagues randomised nearly twenty-five hundred men with prostate cancer on androgen deprivation therapy to usual care or routine referral to an internist or cardiologist mandating a statin and a systolic target of 130 or below [8]. Over a median of nearly six years the hierarchical composite favoured referral, driven almost entirely by lower cholesterol; blood pressure differed by less than two millimetres of mercury and there was no difference in cardiovascular death, infarction, stroke, or heart failure. In other words, specialist referral moved a lab value, not an outcome.

Finally, the BMJ carries a clinical review from Sawa and colleagues on schizophrenia, summarising the neurodevelopmental and gene-environment picture, the complementary roles of pharmacological and psychosocial treatment, and the argument that progress in new drug development is blocked by the biological heterogeneity of a diagnosis defined purely by clinical criteria [10]. Their case is that accumulating objective biomarkers to stratify patients is the necessary first step toward precision psychiatry.

If you only have time for one paper this week, make it the sham-controlled ablation trial in The Lancet [4]. It challenges the symptom-relief rationale we routinely give patients when referring for atrial fibrillation ablation, and it will change how you frame that conversation starting tomorrow.

Here are the key takeaways from this week in General Medicine. First, shortened response-tailored antibiotics for stable left-sided endocarditis buy patients about two weeks off therapy and appear safe overall, but relapse is roughly three times more common, so plan close follow-up. Second, beta-blockers can reasonably be stopped in stable post-infarct patients with preserved ejection fraction and no heart failure. Third, catheter ablation did not beat a sham procedure for quality of life at six months, so counsel accordingly. Fourth, anticoagulation now has randomised support in atrial fibrillation at intermediate stroke risk, though absolute benefit is small. Fifth, eplontersen failed to reduce cardiovascular events in transthyretin amyloid cardiomyopathy, and aficamten's benefit in nonobstructive hypertrophic cardiomyopathy is real but modest and requires ejection fraction monitoring. And finally, faster troponin pathways are safe but will not, by themselves, shorten emergency department stays.

That's your roundup for This Week in General Medicine. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And if it saves you time, a quick rating in your podcast app helps other physicians find the show.

This is an automated summary generated by artificial intelligence, which can make mistakes. Always review the original source materials.