This Week in Family Medicine — Sep 27, 2026

This AudioScholar briefing is generated by artificial intelligence for healthcare professionals and trainees. It is not medical advice.

Welcome to This Week in Family Medicine. This week we're covering 7 notable papers spanning cardiovascular safety of two of the most commonly prescribed drug classes in general practice, a run of studies where the intervention didn't deliver what we hoped, and two forward-looking pieces on how technology may reshape diagnosis. Let's dive in.

We start with prescribing safety, and two large analyses in the BMJ that refine rather than overturn what we tell patients. Berggreen and colleagues report a nationwide nested case-control study drawn from Danish registries, covering women aged fifty to sixty-nine between 2003 and 2021, with nearly ten thousand cases of venous thromboembolism, over eighteen thousand ischaemic strokes, and close to twelve thousand myocardial infarctions, each matched to controls of the same birth year [1]. Among women who had never used menopausal hormone therapy, the background rate of venous thromboembolism was about sixteen events per ten thousand person-years. Current use of oral oestrogen, alone or with a progestin, was associated with roughly a sixty percent higher rate of venous thromboembolism, about a thirty percent higher rate of ischaemic stroke, and about a twenty percent higher rate of myocardial infarction. In absolute terms those relative figures are modest: the authors calculate that around one thousand women would need to take oral therapy for a year to produce one extra clot, and closer to four thousand for one extra myocardial infarction. The dose and duration signal is the part worth holding onto. Oral oestradiol above one milligram a day, used for more than five years, was associated with roughly a doubling of both stroke and myocardial infarction rates. Transdermal therapy showed no increase in thrombotic events, with one exception the authors themselves flag as based on small numbers, an apparent doubling of myocardial infarction with transdermal combined cyclic regimens. This is observational data, so confounding by indication cannot be excluded, but it is consistent with the existing physiological case for the transdermal route in women at higher baseline vascular risk, and it puts absolute numbers into a conversation that patients often experience as pure alarm.

Alongside it, also in the BMJ, Ostinelli and colleagues tackle a question that surfaces every time we prescribe an antidepressant to an older or polypharmacy patient: does it prolong the QT interval, and does that translate into events? They pooled individual participant data from thirty-five double-blind randomised trials, nearly nine thousand patients, covering placebo and ten antidepressants, and ran a network meta-regression accounting for baseline QT, age, sex, body mass index, potassium and renal function [2]. Escitalopram and amitriptyline were the two agents associated with meaningful QT prolongation versus placebo, escitalopram by roughly nine milliseconds and amitriptyline by around five, while venlafaxine and vortioxetine sat at the low end. Crucially, a separate aggregate analysis of one hundred and thirty-nine trials and more than fifty-two thousand participants found no association between antidepressant use and early major cardiovascular events or non-suicidal sudden death over the first eight weeks. The authors are careful: there was substantial variability between individuals, particularly for escitalopram, fluoxetine and mirtazapine, and they frame their conclusion around shared decision making that weighs an individual's modifiable and non-modifiable risk factors rather than around a blanket avoidance rule.

Our second theme is the week's crop of null and near-null findings, and the most striking sits in The Lancet. The REFIT2 trial, led by Johnsen, randomised four hundred and fifty adults with moderate-to-severe irritable bowel syndrome at five Norwegian hospitals to a single rectal enema of donor faecal microbiota, or their own faeces as placebo, two to one, fully masked [4]. At ninety days, about forty percent of those receiving donor transplant met the threshold of a seventy-five point improvement on the severity score, versus about thirty-eight percent on placebo. The difference, roughly two percentage points, was not statistically significant, and adverse events were similar in both groups. This is a clearly negative phase three result in a well-powered trial, and the authors conclude that microbiota modulation alone may simply be insufficient to shift symptoms in irritable bowel syndrome. Given how often patients arrive having read about faecal transplant, that is a useful piece of evidence to have.

Two smaller studies land in the same territory. In the Scandinavian Journal of Primary Health Care, Thy Christensen and colleagues examined what happened when the microbiology department in the North Denmark Region restricted culture and susceptibility testing of wound, skin, eye, ear and throat infections to recurrent or treatment-failure cases [3]. Submitted samples fell by about two thirds, so uptake was real, but antibiotic prescribing dropped only about two percent, and a neighbouring region without the guideline changed similarly. This was a retrospective pre-post design without clinical outcomes, so it cannot tell us whether treatment quality suffered, but it does suggest that reducing testing does not by itself move prescribing. And in BMC Primary Care, Zhou and colleagues surveyed eight hundred and thirty-seven Berlin residents aged sixty-five and over and found that longer continuity with the same general practitioner was associated with better self-rated health, but the effect was very small, far smaller than the contributions of lifetime morbidity, depressive symptoms or subjective social status [5]. The authors read this as a measurement problem rather than a verdict on continuity, arguing that single-item self-rated health may be too blunt to capture what continuity actually does. It's cross-sectional, so causal direction is unresolved either way.

Finally, two essays on where diagnosis is heading. In the Scandinavian Journal of Primary Health Care, Melbye, Rudebeck and Pasterkamp weigh the arrival of artificial intelligence-assisted digital stethoscopes, which can already classify heart and lung sounds at or above the accuracy of the average clinician [6]. Their concern is deskilling, and their proposed antidote is using the machine output as a reflective check against one's own clinical perception rather than as a replacement. They expect the traditional stethoscope to remain the main instrument for the foreseeable future, with the smart version most useful in teaching and in research on what chest sounds actually signify. And in the New England Journal of Medicine, Hunter and Inouye review why polygenic risk scores, twenty-five years after the human genome draft, have had so little effect on the diagnosis and treatment of common adult-onset disease [7]. Thousands of small-effect variants do predict risk, but the authors set out the practical barriers to clinical diffusion and where polygenic scores and pharmacogenomics might plausibly earn a place as sequencing becomes cheap.

If you only have time for one paper this week, make it the Danish hormone therapy study in the BMJ [1]. It is the largest contemporary dataset separating route, dose and duration, and it reframes the menopausal hormone therapy conversation from a single yes-or-no risk question into one about formulation.

Here is what this week's evidence adds up to in Family Medicine. First, from large observational Danish data, the thrombotic risk of menopausal hormone therapy appears concentrated in oral formulations, with higher doses over longer durations carrying the arterial signal, and transdermal therapy largely unremarkable; absolute risks remain small. Second, randomised evidence suggests antidepressant effects on the QT interval differ meaningfully by drug, with escitalopram and amitriptyline at the higher end, but no signal of early cardiovascular events emerged across more than fifty thousand trial participants. Third, faecal microbiota transplantation for irritable bowel syndrome has now failed a well-conducted phase three trial, which is about as clear an answer as we get. Fourth, restricting microbiology testing cut testing substantially but barely touched antibiotic prescribing, and the clinical consequences remain unstudied. And fifth, continuity of care's benefit was small when measured by self-rated health, which may say more about our outcome measures than about continuity itself.

That's your roundup for This Week in Family 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.