Rotterdam dock workers on night shift were coached to eat earlier, and they did, by an hour and a half. Their glucose sensors then drifted slightly the wrong way, and the good news in the paper comes from a test nobody else took

About one in five workers in the European Union works nights, according to the figure the authors of this study cite. Night work goes with more heart disease, type 2 diabetes and obesity, and food is one of the suspects. The body handles a meal worse at 3 a.m. than at 3 p.m., and a night shift pushes eating into exactly those hours. Sleep laboratories have shown that moving food out of the night helps blood sugar over a few days. What almost nobody has tested is whether real workers, on a real roster, can change when they eat, and whether anything measurable follows.

A Dutch team from Erasmus MC in Rotterdam, the research organisation TNO and the national public health institute RIVM tried it. Their paper came out on 8 October 2026 in Frontiers in Nutrition. They recruited 57 healthy men from one company in the Port of Rotterdam, where the work involves short bursts of heavy lifting during docking operations. Everyone worked 12-hour shifts, 06:00 to 18:00 or 18:00 to 06:00, with about seven night shifts a month. The men then chose for themselves what they wanted: 22 picked nutrition coaching, 25 picked sleep coaching and 10 agreed to change nothing. The researchers let them choose to help them stick with it. It also means the groups were different before anything started. The nutrition group was older (38 on average, against 31 in the comparison group), 81% had a BMI over 25 against 38%, and 68% had high total cholesterol against 13%.

The advice applied only to night-shift days and the morning after. Calories were split into four meals: 40% at dinner between 18:00 and 20:00, 20% as a light meal between midnight and 02:00, 25% at breakfast after the shift, and 15% as an afternoon snack after waking. Breakfast was to be mostly carbohydrate and the afternoon snack mostly fat. Each man's plan was adjusted using his results on a test drink and on a glucose sensor, and then to his own tastes. The total calories were meant to stay the same.

On the main behaviour, it worked. The researchers worked out the clock time by which each man had eaten half of his day's calories on a night-shift day. Before the coaching that point came at 22:20. In the first weeks it moved to 19:50, which is 150 minutes earlier. By the end of the three months part of that had slipped back, and the gain was 88 minutes. Carbohydrate, fat and protein all moved by about the same amount. Reported calories and a Dutch diet quality score did not change, and the sleep and comparison groups did not shift their eating times. An earlier report from the same project found that 95% of the men were still using the advice at the end, and 73% eight months later.

The health results are harder to read. Before and after the three months, the nutrition group drank a 950 kcal test shake of sugar, palm oil and milk protein, and gave blood five times over four hours. The curves for total, LDL and HDL cholesterol after the shake changed shape, and so did those for five inflammation markers. That is the result in the paper's title. When the researchers then compared before and after at each time point, they found no significant difference at any of them for cholesterol and few for inflammation, and they describe the lower curves as something seen by "visual inspection". Blood sugar, insulin and triglycerides after the shake did not change at all, which the authors call unexpected. Seventeen markers were tested with no correction for testing so many. And only the nutrition group took the shake test. With no one to compare against, the authors say, they cannot rule out the season, a general drift over time or the tendency of unusual first readings to come back towards average.

Then there are the glucose sensors, which everyone wore for 14 days at a time. In the nutrition group the estimate of average blood sugar (the glucose management index) rose by about 0.14 to 0.15 percentage points by the end, on both day-shift and night-shift days. Two measures of high readings and of swings rose too. Time spent in the healthy range went up by about 1 point on night-shift days, but only compared with the first weeks of the coaching, not with the start. In the sleep-coaching group the same averages went down. The authors suggest that eating earlier bunched the men's glucose peaks into fewer hours, and they say more work is needed to tell whether this is a harmless reshuffle or a real downside. These are small movements. They are still not the direction the study hoped for, and "change in 24-hour glucose" is the one outcome in the trial's registration that the food advice was aimed at. The cholesterol and inflammation curves are not among the registered main outcomes.

Two smaller things are worth knowing. The summary at the top of the paper lists E-selectin among the inflammation markers that fell, while the results section says E-selectin did not change. And the comparison group is given as 10 men, but its percentages in the baseline table (13% and 38%) are eighths, which fits 8. Our reading: this study shows something useful that the laboratory work could not, which is that men doing heavy shift work will move their main meal earlier when the advice is personal and practical, and most keep doing it. It does not show that this improved their health. The benefits it reports rest on one group measured before and after with no comparison, and the measure with a comparison points slightly the other way. If you work nights, a proper dinner before the shift and a light meal during it is a reasonable habit, and earlier laboratory studies linked that pattern to less sleepiness on shift. Don't expect it to lower your cholesterol on this evidence.