Shift work is associated with higher rates of overweight and obesity, and no study has shown that it causes them. The evidence is entirely observational, and where a pooled analysis separates study designs, the cohort studies — the ones that follow the same people forward — report roughly half the excess that the cross-sectional ones do. This article covers what the numbers actually say, why the design difference is the finding rather than a footnote, what the two night-eating experiments did and did not measure, and what is worth acting on if your working hours move.
Cohort study: research that follows the same people forward in time, so it can say what came first. A cross-sectional study measures everyone at a single moment, so it can only say what occurs together. Most of the shift-work evidence is the second kind.
What the pooled evidence actually says
Two meta-analyses own this question, and they agree on the direction while disagreeing usefully on the size.
The larger one pooled 26 studies covering 311,334 participants — 7 cohort studies, 18 cross-sectional and one case-control. Shift work versus non-shift work gave a relative risk of 1.25 for overweight (95% CI 1.08–1.44) and 1.17 for obesity (95% CI 1.12–1.22) (Liu 2018).
The second pooled 28 observational studies of night-shift work. Overall odds ratio for overweight or obesity: 1.23 (95% CI 1.17–1.29). Split by design, cross-sectional studies gave 1.26 and cohort studies gave 1.10. Permanent night workers showed a 29% higher risk than rotating shift workers, 1.43 against 1.14. Abdominal obesity showed the strongest signal of the obesity types examined, at 1.35 (Sun 2018).
Those abdominal-obesity odds are worth naming and then leaving alone. They describe where excess weight tended to sit in these populations; they are not a measurement of visceral fat and they do not support a claim about where any individual's fat is or what any treatment does to it.
Why the design difference is the finding
Because 1.10 and 1.26 are answers to different questions.
A cross-sectional study photographs a workforce once. If the people currently working nights are heavier than the people currently working days, that is consistent with nights causing weight gain — and equally consistent with heavier people being more likely to take, or to stay in, night work, or with a third factor driving both. A cohort study follows people forward, which is the only one of the two designs that can put events in order. When the analysis separated them, the cohort-only estimate came out at less than half the cross-sectional excess (Sun 2018).
There is a second reason to hold these numbers loosely, and the authors of the larger analysis flag it themselves: "the cut-off points of overweight and obesity varied greatly, so the heterogeneity was substantial". Studies conducted in different countries classified people by different thresholds and were then pooled together. For a Singapore reader that matters directly, because Singapore's own obesity guidelines classify overweight from a BMI of 23 and obese from 27.5, while the international classification used in much of the underlying research runs from 25 and 30 (HPB–MOH obesity guidelines, 2016). Pooled estimates built across both scales are not measuring one consistent thing.
None of that makes the association fictitious. It makes the honest sentence a narrower one: shift work travels with higher rates of overweight and obesity, more so with permanent nights than rotating ones, and nobody has shown that a roster is the cause of any particular person's weight.
The diabetes association, and where the line sits
Pooling 12 observational studies covering 226,652 people and 14,595 cases, shift work was associated with an adjusted odds ratio of 1.09 for diabetes (95% CI 1.05–1.12). The association was significantly stronger in men (1.37, 95% CI 1.20–1.56) than in women (1.09), with a p for interaction of 0.01, and rotating shifts carried a higher figure than mixed or evening ones (Gan 2015).
Nine percent is a small effect in observational data, and the sex difference is a subgroup comparison rather than a separate trial.
That statistic describes a population's risk profile and nothing about any individual working nights. Diabetes is a named disease and it belongs with a person's own doctor. GetLean is a weight-management service, and diabetes care sits outside it.
The two night-eating experiments, and what they measured
Both are real, both are small, and neither measured what most people think they measured.
In a 14-day randomised laboratory protocol, 19 participants underwent identical simulated night work with identical sleep disruption; only meal timing differed. Nighttime eating produced misalignment between central and peripheral circadian rhythms and impaired glucose tolerance, while daytime eating prevented it. The authors' own framing is that this "offer[s] a behavioral approach to preventing glucose intolerance in shift workers" (Chellappa 2021).
A separate controlled study — not randomised, 11 healthy men, four simulated night shifts — found increased glucose area-under-the-curve only in the night-eating condition, with no significant change when the same shifts were worked without eating at night. Its own authors call it a small healthy sample and say further study is needed (Grant 2017).
Here is the part that gets dropped: neither study measured body weight, fat mass or lean mass. There is no evidence that shifting meals to daytime hours during night work changes body composition, because the question has not been asked. What has been shown is a glucose response, in a laboratory, in 19 and 11 people respectively.
Two boundaries follow. A glucose finding is not a blood-sugar claim about anything we do, and it is not advice about diabetes. And skipping meals is a poor default for anyone whose appetite is already suppressed by medication — that is a conversation for the prescribing doctor, not an inference from a 14-day laboratory protocol.
What actually transfers to an irregular roster
Three things, none of them shift-specific, all of them well evidenced in general populations.
Training volume can be arranged around the roster. When weekly volume is held constant, "resistance training frequency does not significantly or meaningfully impact muscle hypertrophy" (Schoenfeld 2019). For someone whose days off move every fortnight, that finding is the practical one: fix the weekly total, let the days land where they can. And the floor is genuinely low — one hard set per exercise, two to three times a week, produced significant strength gains over 8 to 12 weeks in resistance-trained men, a dose the authors call suboptimal rather than ideal (Androulakis-Korakakis 2020).
Resistance training is what changes the composition of the loss. Across 114 trials and 4,184 people, lean mass was statistically unchanged where resistance training accompanied caloric restriction (Lopez 2022). Pooling 34 randomised trials, adding exercise to a calorie-restricted diet prevented roughly 46% of the fat-free mass otherwise lost, with mixed training at +1.20 kg (p < 0.001) and strength training at +0.83 kg (p = 0.013) reaching significance while endurance training alone reached +0.51 kg and did not (p = 0.067) — and subgroup testing found no significant differences between the modes (Deller 2026). A second network meta-analysis of 62 trials found dieting without exercise was the only arm whose lean-mass loss was statistically significant (Xie 2025).
Protein has a target that does not care what time it is. Intakes of 1.2 to 1.6 g per kg of body weight per day during energy restriction improve appetite, body-weight management and lean-mass preservation relative to lower-protein diets (Leidy 2015). Note the denominator: body weight, not fat-free mass. Hitting that on hawker food is its own problem, covered in hawker dishes ranked by protein per calorie, and hitting it on a suppressed appetite is covered in protein with no appetite.
What has not been studied
Worth saying plainly, because content in this area tends to imply otherwise.
No trial of any training or eating intervention in Singapore shift workers has been published. The whole shift-work evidence base is observational epidemiology conducted elsewhere, plus the two small laboratory meal-timing studies above. Separately, no trial of resistance training, home training or step targets has been run in people taking a GLP-1 medication at all — every training recommendation for these patients, including ours, is extrapolated from general weight-loss populations.
At GetLean, our philosophy is that GLP-1 medication should act as a catalyst rather than something to depend on indefinitely. On an irregular roster the medication side changes very little; what changes is how hard it is to keep two training sessions and a protein target in a week that never looks the same twice. That is a planning problem, and it is a more tractable one than the roster.
Individual results vary, and clinical-trial figures describe the populations that were studied. Before starting resistance training — particularly with any existing joint, cardiac or metabolic condition — ask a doctor whether it is suitable for you.
Common questions
Does shift work cause weight gain?
No study has shown that. Every pooled analysis in this area is built from observational research, and in the larger of the two, 18 of the 26 included studies were cross-sectional — measuring shift work and body weight at the same moment, which cannot establish which came first (Liu 2018). The association is real and repeatedly found. The causal claim has not been made by the researchers themselves.
How much higher is the risk of obesity for shift workers?
It depends which study design you read, and the difference is large. Across 26 observational studies and 311,334 people, shift workers carried a relative risk of 1.25 for overweight and 1.17 for obesity (Liu 2018). Across 28 studies of night-shift work specifically, the overall odds ratio was 1.23 — but among cohort studies alone it was 1.10, less than half the excess, and permanent night work carried a higher figure than rotating work, 1.43 against 1.14 (Sun 2018). The cohort estimate is the more conservative and the better designed.
Should you avoid eating during a night shift?
The two experiments people cite for this measured blood glucose, not body composition. In a randomised laboratory trial of 19 people, confining meals to daytime hours during simulated night work prevented the glucose intolerance seen when the same people ate at night (Chellappa 2021); a separate controlled study in 11 healthy men, not randomised, pointed the same way over four simulated shifts (Grant 2017). Neither measured weight, fat or lean mass, so no body-composition conclusion follows. Anyone taking appetite-suppressing medication should discuss meal timing with their doctor before removing a meal.
What does the shift-work evidence say about diabetes?
It reports an association, not a cause. Pooling 12 observational studies covering 226,652 people, shift work was associated with about 9% higher odds of diabetes (OR 1.09, 95% CI 1.05–1.12), with a significantly stronger association in men (1.37) than in women (1.09) (Gan 2015). That is background about a population's risk profile. Diabetes is a named disease that belongs with a person's own doctor, and GetLean is a weight-management service rather than a part of anyone's diabetes care.
How do you train on a rotating roster?
Fix the weekly total and let the days float. When weekly training volume is held constant, frequency does not significantly or meaningfully affect muscle growth (Schoenfeld 2019), so the sessions can land wherever the roster allows. The floor is also low: one hard set per exercise, two to three times a week, produced significant strength gains in resistance-trained men over 8 to 12 weeks, a dose those authors describe as suboptimal rather than ideal (Androulakis-Korakakis 2020). Individual results vary.