Nutrition & DietReview ArticleOpen Access

What Nutrition Strategies Actually Help Shift Workers? A Systematic Review Finds Slim Evidence

A systematic review of 43 studies finds no clear dietary guidelines for shift workers, with time-restricted eating and macronutrient timing showing promise but weak evidence.

Friday, October 2, 2026 1 view
Published in J Hum Nutr Diet
A night-shift nurse in scrubs eating a meal in a hospital break room under fluorescent lighting, with a clock on the wall showing 3am

Summary

Shift work disrupts the body clock and is linked to metabolic disease, yet nobody knows the best way for shift workers to eat. Researchers searched four major databases and screened nearly 7,000 articles. They found 43 studies on diet in shift workers, including only four in first responders such as police and firefighters. The studies tested changes in carbohydrate, protein and fat, time-restricted eating, and meal timing. Some suggested metabolic benefits, but results on alertness and work performance were inconsistent. Most studies were small, short, and at moderate-to-high risk of bias, and many used simulated night shifts in healthy volunteers. The authors conclude there is not enough evidence for formal dietary guidelines for shift workers. They call for larger, better-designed, real-world trials, especially in first responders.

Detailed Summary

Shift work is common, and it is linked to metabolic disorders, gastrointestinal problems, fatigue-related accidents, and poorer diet quality. Eating at the wrong biological time can disrupt peripheral clocks in the liver, muscle, and fat, impair glucose tolerance, and alter leptin and ghrelin. Diet could therefore be a modifiable lever, but no consensus guidance exists. This review asked what nutrition interventions have actually been tested in shift workers, with first responders (police, fire, ambulance) as a pre-specified subgroup.

The authors registered the protocol (PROSPERO CRD42023421400) and followed PRISMA and SwiM reporting guidance. They searched Embase, Cochrane, Web of Science, and PubMed (plus SCOPUS per the methods) from inception to August 2023, re-ran the searches in July 2025, and added grey literature and interviews with UK ambulance and fire services. Of 10,054 database records and 631 grey-literature records, 6,909 abstracts were screened after removing 3,759 duplicates. Two reviewers screened independently, and 43 studies were included. Bias was assessed with Cochrane tools and the Newcastle–Ottawa Scale. Meta-analysis was impossible because the studies were too heterogeneous.

The included studies were 21 crossover trials (49%), 14 observational studies (33%), and 8 parallel RCTs (19%). Samples were small: the mean was 119 overall, but crossover trials averaged only 18 participants and RCTs 45. Mean age was 33 years, mean BMI 25.9, and 79% of participants were healthy. Nearly two-thirds (65%) were free-living, 40% were nursing or healthcare staff, and 35% used simulated shifts. Only four studies (9%) involved first responders. Over 20 different definitions of night shift timing were used, and 40% did not report years of shift experience.

Interventions were macronutrient changes (23%), diet quality or patterns (23%), time-restricted eating (21%), and meal timing (21%), plus one nutritional counseling study. Outcomes were metabolic (29%), cardiometabolic (22%), cognition and performance (22%), and sleep (7%). Macronutrient trials varied widely: high-protein arms ran 30–60% of energy, high-fat 40–45%, and carbohydrate 20–70%. Results were mixed. One crossover trial found lower postprandial glucose (p=0.04) and insulin (p=0.02) with a lower-carbohydrate meal but no triglyceride difference (p=0.72). A high-protein RCT found no differences in glucose, insulin, or triglycerides (p=0.43–0.73). Cognitive findings were similarly inconsistent, with working memory differing in one trial (p<0.001) but vigilance not (p=0.087).

The 2025 search update added six papers that did not change the conclusions. The authors state that time-restricted eating and macronutrient distribution show potential for metabolic outcomes, but evidence on cognitive and occupational performance is inconsistent. Overall, the evidence is insufficient to support evidence-based dietary guidelines for shift workers.

Important caveats apply. Moderate-to-high risk of bias, small samples, short durations, and heterogeneous protocols limit confidence. Many trials were laboratory simulations in healthy, mostly young participants, which may not reflect real shift work. Inclusion was restricted to English-language publications, and caffeine- and fluid-only studies were excluded. Risk-of-bias assessment was performed by a single reviewer.

Key Findings

  • 43 studies met inclusion criteria from 6,909 screened abstracts (10,054 database and 631 grey-literature records before deduplication)
  • Designs were 49% crossover trials, 33% observational, and 19% parallel RCTs; mean sample was 119 overall but only 18 in crossover trials
  • Only 4 studies (9%) involved first responders, while 40% were in nursing and healthcare staff
  • Interventions: macronutrient changes 23%, diet quality 23%, time-restricted eating 21%, meal timing 21%
  • Outcomes: metabolic 29%, cardiometabolic 22%, cognition and performance 22%, sleep only 7%
  • Mixed results: one lower-carbohydrate crossover trial reduced postprandial glucose (p=0.04) and insulin (p=0.02), while a high-protein RCT showed no glucose, insulin, or triglyceride differences (p>0.4)
  • Over 20 different night-shift definitions and moderate-to-high bias prevented meta-analysis and guideline development; 6 papers added in 2025 changed no conclusions

Methodology

Systematic review registered on PROSPERO (CRD42023421400) and reported per PRISMA and SwiM guidelines. Databases were searched from inception to August 2023 and re-run July 2025, with grey literature, snowball searching, and interviews with UK ambulance and fire services. Two reviewers screened independently, with a third resolving disagreements. Risk of bias used Cochrane RoB tools for RCTs and crossover trials and the Newcastle–Ottawa Scale for observational studies, and findings were synthesized narratively by intervention type because meta-analysis was not feasible.

Study Limitations

The authors note substantial methodological heterogeneity, small sample sizes, short interventions, and moderate-to-high risk of bias, with inconsistent reporting of shift type and experience. First responders were underrepresented and many trials used simulated shifts in healthy volunteers, limiting real-world applicability. Only English-language studies were included, risk-of-bias assessment was done by one reviewer, and conflicts of interest were not stated in the text provided.

Enjoyed this summary?

Get the latest longevity research delivered to your inbox every week.

Enter your email to subscribe: