When pills are stopped: thyroid, statins after 65 and two more stories

In two phase 3 trials of 757 adults, an RNA drug cut triglycerides and pancreatitis episodes

On 5 November 2026 the New England Journal of Medicine published SHASTA-3 and SHASTA-4, two identically designed, double-blind, placebo-controlled phase 3 trials of plozasiran in 757 adults with severe hypertriglyceridaemia, a very high level of blood triglycerides that raises the risk of acute pancreatitis. Plozasiran is a small interfering RNA that switches off production of apolipoprotein C3, a protein that slows the clearance of triglyceride-rich particles from the blood. Participants were randomly assigned 2:1 to the drug or placebo and followed for 12 months.

At 12 months, median triglycerides were 79% and 81% below baseline on plozasiran in the two trials. Compared with placebo, those reductions were 45 and 47 percentage points greater (95% CI, 38 to 51 and 39 to 54 percentage points). A level below 500 mg/dL (5.6 mmol/L) was reached by 91% and 93% of people on the drug versus 51% and 50% on placebo. In the pooled analysis, acute pancreatitis occurred at a rate of 1 per 100 patient-years with plozasiran and 5 per 100 patient-years with placebo (rate ratio 0.22; P = 0.008). Adverse events were reported by 73% of participants in both groups.

The pancreatitis result comes from a pooled analysis of both trials over one year, not from the primary endpoint, and the trials were funded by the developer, Arrowhead Pharmaceuticals. For severe hypertriglyceridaemia the drug is not yet approved. On 19 October 2026 the UK regulator authorised it only for familial chylomicronaemia syndrome, a rare inherited form of very high triglycerides; these two trials test it in the much broader group of people with severe hypertriglyceridaemia.

In 23 studies, about a third of people stayed off thyroid hormone after stopping it

On 2 November 2026 Thyroid published an updated systematic review and meta-analysis by Toloza, Brito and colleagues on supervised discontinuation of thyroid hormone. Levothyroxine is widely prescribed, often for subclinical hypothyroidism, when TSH is raised but thyroid hormone levels are normal, and evidence of clinical benefit in people without overt hypothyroidism is limited. The review included 23 studies, 22 observational and 1 randomised trial; 20 of them looked at stopping levothyroxine alone.

Across 21 studies with 2020 participants, 37% (95% CI 26% to 48%) still had normal thyroid function after stopping. The share was 41% (21% to 65%) among people first treated for subclinical hypothyroidism and 11% (2% to 38%) among those treated for overt hypothyroidism. In 11 studies with 1104 participants, 60% restarted therapy, rising to 87% in overt hypothyroidism. Positive thyroid antibodies, an uneven thyroid ultrasound pattern and a higher TSH before treatment went with restarting. Symptoms were generally unchanged after stopping, and adverse events were uncommon.

Almost all evidence is observational and results varied greatly across studies (I² 94.5% for the main result). The review supports considering a structured discontinuation attempt in selected patients, particularly those without a clear initial indication, with close biochemical monitoring and predefined criteria for restarting therapy. Discuss any change with your doctor.

Among 197,000 statin users aged 65 and over in England, stopping was linked to more hospital admissions, except in the frailest

On 2 November 2026 The Lancet Healthy Longevity published a cohort study by a University of Oxford team built on routine English primary care records from 1998 to 2021. Of 383,342 people aged 65 and over who had been taking a statin steadily, 65,727 stopped; each was compared with two people from the same practice who carried on, 197,180 people in all, mean age 75.9, followed for a mean of 4.1 years. The main outcome was any hospital admission.

In the first year, stopping was associated with a higher risk of admission (subdistribution hazard ratio 1.22; 95% CI 1.20 to 1.25), about 38 more admissions per 1000 people. The link weakened as frailty grew: 1.30 in fit people, 1.22 in mild and 1.16 in moderate frailty, and no association in severe frailty (1.03; 0.94 to 1.12) or dementia (1.05; 0.99 to 1.12). Among care home residents, admissions were lower after stopping (0.76; 0.63 to 0.90). Over five years the pattern held: 1.11 in the whole cohort and no increase in severe frailty.

This is an observational study: people who stop a statin may differ from those who continue in ways the records do not capture, and the authors themselves warn about residual confounding. The outcome was hospital admission for any reason, not heart attacks and strokes separately. The study was funded by the National Institute for Health and Care Research and the Wellcome Trust.

In 9837 Dutch adults, an AI reading routine eye photos picked out 7 in 10 eyes with glaucoma

On 2 November 2026 Ophthalmology published an external test of the deep-learning algorithm that won the international AIROGS glaucoma-screening challenge. The algorithm was trained on 101,442 colour photographs of the back of the eye; the authors applied it to 19,088 eyes of 9837 participants aged 50 to 80 in the population-based Rotterdam Study. Of these eyes, 201 (1.1%) had glaucoma with confirmed visual field loss.

At a setting that cleared 95% of healthy eyes, the algorithm found 70.6% of eyes with glaucoma (95% CI 63.8% to 76.8%); the area under the ROC curve was 0.911. That meant 5.1 referrals for every detected case. Among correctly flagged people who had earlier photographs, 33 of 61 (54%) had been flagged before the diagnosis, typically 4 to 7 years ahead.

About 3 in 10 glaucoma eyes were still missed, and these had milder visual field damage than the ones found. Incomplete optic-nerve-head visibility was an important source of false positives. This external validation in a prospective population-based cohort supports the algorithm as a scalable triage tool and highlights image-acquisition standards; it measured accuracy, not whether such screening protects sight.

This article is for information only and does not replace a consultation with a doctor. Nothing in it is a recommendation to start, stop or change treatment.

Sources

• Watts GF, Ballantyne CM, Blanco Echevarria A, et al. Plozasiran for Severe Hypertriglyceridemia and Risk of Pancreatitis. N Engl J Med. 2026;395:1805-1817. PMID 42981569, doi:10.1056/NEJMoa2610014, NCT06347003, NCT06347016

• Toloza FJK, Ho K, Burgos N, et al. Clinical Outcomes after Deprescribing Thyroid Hormone Therapy: An Updated Systematic Review and Meta-Analysis. Thyroid. 2026. PMID 42969098, doi:10.1177/10507256261502928

• Seeley AE, van der Veen R, Wang A, et al. Efficacy and safety of statin discontinuation in patients 65 years and older according to frailty status: a population-wide retrospective cohort study in England using Clinical Practice Research Datalink data. Lancet Healthy Longev. 2026. PMID 42971235, doi:10.1016/j.lanhl.2026.100920

• van Meyel SWF, van Garderen KA, Vermeer KA, et al. External validation of a deep-learning algorithm for perimetric glaucoma detection on color fundus photographs in the Rotterdam Study. Ophthalmology. 2026. PMID 42971439, doi:10.1016/j.ophtha.2026.10.030