By six months, 3 in 10 patients had quietly stopped seeing their dietitian, and among 470 people at a Japanese clinic, living with family went with leaving sooner

Guidelines for diabetes, high blood pressure and similar long-term conditions recommend ongoing nutrition therapy with a dietitian, and the benefit depends on people actually coming back. Researchers from Hokkaido University and a family medicine clinic in Sapporo went through the clinic's records to see when patients stop, and who. Their study was published in the Journal of General and Family Medicine on 26 October.

They followed every adult who had a first dietitian session at the clinic between April 2018 and March 2023, for up to two years. People referred for cancer or malnutrition were left out, because the aims of therapy differ, which left 470 of 533 patients. Their median age was 63 and 59.4% were women. Conditions overlapped: 74.0% had abnormal blood fats, 61.1% high blood pressure, 55.7% obesity and 41.5% diabetes.

Stopping was defined strictly. A patient counted as having discontinued when there was no visit within three months of the last one, nothing booked, and no agreed decision to finish. Patients whose therapy ended by agreement, including those who reached their goals, weren't counted as dropouts, and neither were those who moved to another clinic or died.

On that definition, 88.3% were still attending at 3 months, 70.4% at 6, 54.4% at 12 and 43.5% at 24. The median time to stopping was 15 months. The loss slows down: about 30 percentage points went in the first six months, and about 11 across the whole second year (our subtraction). A patient's last actual visit came earlier than these dates suggest, because the rule places the stop three months after it. When the authors allowed for therapy that ended for other reasons, 51.6% had discontinued by 24 months, rather than the 56.5% implied by the simple curve (our subtraction).

Who left? Patients who weren't taking medication, who didn't have obesity (here a BMI of 25 or more, the cut-off the study used) or who didn't have abnormal blood fats were likelier to stop. After adjustment, medication went with a 37% lower rate of stopping, obesity 31% and abnormal blood fats 29% (our conversion of the hazard ratios). The authors suggest that people on tablets are already in and out of the clinic, and that a clearer problem may feel more worth attending for, though they didn't test either idea. In a separate analysis, patients with three or more target conditions had about half the rate of stopping (adjusted hazard ratio 0.52).

The most useful finding may be about readiness. Each patient's stage of change was recorded at the start, and over half (54.9%) were in the preparation stage, meaning they intended to change but hadn't settled into doing it. Those patients stopped at 1.75 times the rate of people already acting on a change (95% CI 1.23 to 2.49). In raw numbers, 57.4% of them discontinued against 40.6% of those already acting (our division). People who hadn't yet decided to change had a result in between (1.35) that wasn't statistically clear. The authors think people in preparation may disengage when the first difficulties arrive, but that too is their interpretation.

The finding on family needs more care. Living with family went with a higher rate of stopping (adjusted hazard ratio 1.50, 95% CI 1.01 to 2.23, P = .046), and the authors point to earlier work on household eating habits getting in the way of dietary change. But the lower end of that range almost touches no effect, only 68 people didn't live with family, and the raw gap (52.6% against 44.1%, our division) wasn't significant. The study didn't look at who was in the household or how they got along. Complications such as kidney or eye disease also went with more stopping after adjustment (1.46), although the raw figures barely differed (52.9% against 49.8%, our division).

Much showed no clear link: age, sex, having a job, living outside the local area, irregular meals, alcohol, smoking and exercise. Nor did having dropped out of nutrition therapy before (adjusted hazard ratio 1.14, 95% CI 0.69 to 1.89), though that range is wide and only 53 patients had done so.

The limits matter here. This is one clinic, looked at through its records, and the authors note its relatively high dietitian staffing, with three dietitians giving about 1,700 sessions a year, so continuation may differ where access is tighter. Therapy was covered by Japan's universal insurance, which won't be true everywhere. Someone recorded as stopping may have carried on elsewhere. Income, health literacy, mental health and the relationship with the dietitian weren't measured, and stage of change was recorded only once. By our count the main model makes 20 comparisons, and the paper describes no multiple-testing correction, so results near the P = .05 line could be chance. Above all, the study didn't look at health outcomes, so it can't say that those who left did worse or that those who stayed did better. The authors call for multicentre studies. They report no specific funding and no competing interests.