Diet Reference Archive: an approach to eating
Clean Eating
Clean eating is a popular label for eating foods seen as whole, natural or unprocessed and avoiding foods seen as processed or impure; there is no agreed definition or single authority for the label. It is promoted as a route to health and wellbeing, and researchers study it mostly for its links with rigid eating and orthorexia.
ON THIS PAGE
- In brief
- Overview
- For the coach
- The evidence (opens on request)
- Key terms
- References and Resources
- Evidence rating and review record
This is a short archive page. The summary and sections 1 and 2 are open when the page loads. Section 3 holds the research detail and opens when you select its heading. Intuitive Eating and Mindful Eating, two approaches that work without food rules, have their own pages, as do Whole30 and Detoxes and Cleanses, which are often mentioned alongside clean eating; this page does not repeat their evidence. A coach avoids moral words about food, including “clean.” Back to the Diet Reference Archive index.
In brief
What is it? A label, not a diet. People use it for eating foods they see as whole, natural, unprocessed or organic, and for avoiding processed foods, additives and sugar; some versions also cut out gluten, grains or dairy without a medical reason. There is no agreed definition or single authority for the label. Individual authors and companies use different definitions, and studies find that people mean different things by it (Ambwani et al., 2019, 2020).
Does it work? The searches for this page did not identify a study that measured heart disease, diabetes, cancer, death or body weight in people following clean eating, so its health effects are unknown. The research measures what people think it means and how it relates to eating behavior instead. In one survey of 762 Australian women, those who said they followed advice from clean eating websites were more likely to meet the national fruit guideline, and had higher dietary restraint scores, than women who never did (Allen et al., 2018). Recipes from popular clean eating blogs had about the same energy, sugar and sodium as matched recipes without the label (Dickinson et al., 2018). Every rating on this page is low or very low.
Is it safe? Eating plenty of vegetables and fruit and choosing minimally processed foods are part of mainstream heart-health guidance (Lichtenstein et al., 2026), and many people who say they eat clean mean just that. The concern is rigidity. Single surveys link following or favoring clean eating with dietary restraint and with disordered eating symptoms (Allen et al., 2018; Ambwani et al., 2019), and “clean” is one of the words people with orthorexia often use for the foods they allow (Donini et al., 2022). These are associations; they do not show that clean eating causes disordered eating. Orthorexia, a proposed name for a harmful fixation on eating only foods a person considers healthy, is not a formal diagnosis in the DSM-5-TR or the ICD-11 (Donini et al., 2022; Lombardo et al., 2026).
Who should not do it without talking to a doctor or dietitian first? These precautions are used on this page; they are not findings of the clean eating studies. A person with a current or past eating disorder; a person whose eating is part of managing a condition such as diabetes, kidney disease, celiac disease or a food allergy; a person who is pregnant or breastfeeding; and a parent planning strict food rules for a child should bring in their clinician or dietitian, or find one, before cutting out foods or food groups.
Related pages: Intuitive Eating, Mindful Eating, Whole30 and Detoxes and Cleanses. Full list on the archive index.
1. Overview
What you do
Clean eating has no agreed rules, because no single authority owns the term. In studies that asked people what it means, the most common answers were eating whole, natural or unprocessed foods, more vegetables and fruit, organic or non-genetically modified food, and lean protein (Ambwani et al., 2019, 2020). One research team, noting that there is no consensus definition, cites a description of choosing natural, wholesome foods free of additives, preservatives and refined or processed ingredients (Allen et al., 2018).
Beyond that core, the term stretches. Some people use it for balanced meals in moderation, others for a strict list of permitted foods, and some versions remove gluten, grains or dairy without a medical reason (Ambwani et al., 2019). In a national text-message survey of young people aged 14 to 24, about 40 percent of those who answered described clean eating in terms of processed or whole foods, about 30 percent in terms of avoiding or restricting foods, and fewer than 1 percent in terms of weight loss (Ambwani et al., 2020).
What you do not do
No food is forbidden by an agreed definition, and nothing has to be counted. In practice, the foods people most often say they avoid are processed foods, fast food and preservatives, fats and fried foods, and added sugar (Ambwani et al., 2019). Because clean eating names no foods, its environmental footprint depends on what each person chooses to eat.
Where it came from
Clean eating has no founder, book or trial behind it. Researchers trace its spread to social media, food blogs and books in which people without nutrition qualifications describe their own diets and lifestyles (Allen et al., 2018; Ambwani et al., 2019). A search of 3 search engines in January 2016 found 13 popular, recently updated clean eating blogs, most of them based in the United States and written by one woman (Dickinson et al., 2018).
No company owns the name, so no single business profits from it. “Clean” dietary products are marketed and sold under it (Ambwani et al., 2019), and food manufacturers have reformulated products in response to demand for “cleaner” processed foods (Allen et al., 2018). ‘Clean’ has no standardized federal nutritional definition. General rules against false or misleading food labeling still apply. A 2022 review documents how differently the food industry uses the word and argues for more federal oversight, while recognizing that existing legal powers can already act against deceptive labels (Negowetti et al., 2022), and the authors of a 2024 menu study note that there is no regulatory guidance on using the word on food packaging in the United States (Hess et al., 2024).
What it claims to do
Because clean eating has no official source, its claims come from the people who promote it. Researchers describe popular sources promising benefits that range from more energy and better skin to weight loss and the cure of illnesses (Allen et al., 2018; Ambwani et al., 2019). Many people simply take it to mean healthy eating: in the national survey of young people, 71 percent of those who answered called clean eating healthy, 6 percent called it harmful, and 18 percent said it could be both (Ambwani et al., 2020). Section 3 reports what the research shows.
What it gets right
The common core of clean eating overlaps with mainstream guidance. Among the features of a heart-healthy eating pattern, the American Heart Association lists eating plenty of vegetables and fruits, choosing foods made mostly with whole grains, choosing minimally processed foods instead of ultraprocessed foods, and minimizing added sugars (Lichtenstein et al., 2026). A client who says they eat clean may be describing exactly these habits, and that is worth hearing first. Versions that remove all grains depart from the same guidance, which lists whole grains as part of a heart-healthy pattern.
Clean eating and orthorexia
Orthorexia nervosa, often shortened to orthorexia, was first described by a family doctor, Steven Bratman, in 1997, for a fixation on “correct” eating he saw among his patients (Donini et al., 2022). It is not a formal diagnosis. It is not recognized in the current Diagnostic and Statistical Manual of Mental Disorders (DSM-5-TR) or the International Classification of Diseases (ICD-11) (Donini et al., 2022), and a 2026 paper describes it as still not recognized as a mental disorder in the DSM-5-TR or other diagnostic systems, with researchers still debating whether it belongs with eating disorders, with obsessive-compulsive disorder, or on its own (Lombardo et al., 2026).
In 2022, 47 eating disorder researchers and clinicians from 14 countries agreed on a proposed definition: a strong preoccupation with eating, governed by rigid, self-imposed rules that take excessive time to plan, obtain and prepare food, that bring distress and guilt when broken, and that lead to a nutritionally unbalanced diet or to harm in work, school or social life. The panel noted that people with orthorexia often describe the foods they allow as pure, clean, organic or natural. It also stated that adopting healthy or pure eating is necessary but not sufficient for orthorexia, because such efforts vary widely and are not always harmful. The proposal rests on expert opinion, not on new studies (Donini et al., 2022).
The difference matters for a coach. Clean eating is a label many people use for ordinary efforts to eat well; orthorexia, as proposed, is a pattern that harms the person. The label alone does not tell a coach which one is in front of them, and deciding that is not the coach’s job (section 2).
How it differs from related approaches
Clean eating is a label for an ongoing eating style, not a short program; the Detoxes and Cleanses page covers juice cleanses and detox kits, and in a 2019 study of college students almost no one defined clean eating as a cleanse (Ambwani et al., 2019). Whole30 is a branded 30-day elimination program with written rules, while clean eating has none. Intuitive Eating and Mindful Eating work in the other direction: they ask a person to notice hunger, fullness and satisfaction rather than sort foods into allowed and forbidden.
How clients describe it
A client may say they eat clean, eat clean most of the time, are trying to clean up their diet, or feel they ate “dirty.” The phrase can mean anything from more vegetables to a long list of forbidden foods, so asking what the client means comes first. A coach treats “clean” as a judgment word and does not adopt it: a client who uses it has offered their own judgment, which the coach can reflect without taking on the word.
2. For the coach
What a coach does with it
The coaching move is the one used for any eating pattern: ask what the client means by clean eating, what drew them to it, and what they hope it will do; describe the evidence with permission; and hand the decision back. Many clients mean something close to mainstream guidance, and the coach can say so without adopting the word “clean,” which belongs, alongside “junk” and “cheating,” among the moral words about food a coach avoids.
Clean eating is often learned online. In the national survey of young people, social media was the source named most often by those who said where they had heard of it (Ambwani et al., 2020). A coach can be curious about where a client’s food rules came from and how the client feels after time with that content, without telling them what to follow.
Money and time matter too. In a 2024 menu study, a typical American menu rebuilt from less-processed foods cost $15.91 per person a day against $9.85 for the original at fall 2023 grocery prices, with no better diet quality score (Hess et al., 2024). It is a comparison of menus, not proof that every version of clean eating costs that amount. A client who says they cannot afford to eat clean is describing their situation, not a failure.
What to ask
- “When you say you eat clean, what does that look like on an ordinary day?”
- “Where did your ideas about clean eating come from, and which of them still fit you?”
- “What happens for you when a meal does not fit your rules?”
When to refer
Some things a client reports are reasons to involve their clinician rather than topics for coaching: unusual tiredness, dizziness or fainting, missed periods, hair loss, or weight loss the client did not intend. They call for a medical assessment, not for a coach’s interpretation, and a client with an existing care plan follows it. These precautions are used on this page; they are not findings of the clean eating studies. A client who reports a severe or rapidly worsening symptom needs emergency help.
Low blood sugar can be a side effect of insulin and of some other diabetes medicines, including sulfonylureas, and not eating enough carbohydrate or skipping or delaying meals makes it more likely (National Institute of Diabetes and Digestive and Kidney Diseases, n.d.). A client on these medicines who plans a substantial change in what or when they eat tells the prescriber before making it. A client who reports shakiness, sweating or confusion may be having a low blood sugar episode right now; the first step is the one in their own diabetes plan for treating it, with emergency help if the symptoms are severe or the episode is one they cannot manage, and telling the prescriber is the separate step that follows.
Eating behaviors are a separate question. A client may describe a growing list of foods they will not eat, strong guilt or anxiety after eating something off the list, skipping meals with friends or family because of the food, or spending much of the day planning what to eat. Many behaviors have ordinary reasons, such as a work schedule, a household’s preferences, food access, cultural or religious practice, or a medical plan, so on its own a behavior tells a coach little, and none of this is a diagnosis. The coach asks what the behavior is for, whether it causes distress, whether it is narrowing what the person eats, and whether it is getting in the way of daily life.
Where the answers raise a real concern about disordered eating, the coach follows their own referral process, and a client who has a clinician or care plan takes the concern to that clinician. Clean eating can also collide with a medical eating plan. A client with celiac disease, a food allergy, diabetes or kidney disease keeps that plan, and any change to it goes through the clinician or dietitian who set it.
Special populations
- Pregnancy and breastfeeding
- This page’s searches, run October 21, 2026, did not identify a study of clean eating in pregnancy or breastfeeding. Orthorexia questionnaires have been used in pregnancy: in 480 pregnant women at one hospital in Türkiye, read only in the abstract, stronger orthorexia-related attitudes went with higher education, persistent nausea and vomiting, stopping dietary supplements, avoiding caffeine and more pregnancy information seeking; the authors treated the questionnaire as a measure of attitudes, not a diagnosis (Bulduk & Güner, 2026). Food choices and supplements in pregnancy come from the prenatal care team.
- Children and adolescents
- In the national survey of young people aged 14 to 24, 55 percent had heard of clean eating and 41 percent said they would probably try it, with willingness highest among cisgender young women (Ambwani et al., 2020). This page’s searches did not identify a study of clean eating practices in children. A young person’s eating concerns go to a parent or guardian and the young person’s clinician, through the coach’s own referral process.
- Older adults
- This page’s searches did not identify a study of clean eating in older adults. In a convenience sample of 313 adults in China with an average age of 68, higher orthorexia questionnaire scores went with more fruit and vegetables, more physical activity, and better body image and life satisfaction (He et al., 2021), a reminder that these questionnaires can pick up a healthy interest in food as well as a harmful one. Whether a restricted diet meets an older adult’s needs is a question for a dietitian.
- Athletes and people with high activity
- This page’s searches did not identify a study of clean eating in athletes. Reviews of orthorexia questionnaire studies report high proportions of positive screens in exercising groups: 55.3 percent across 24 studies of people who exercise (Hafstad et al., 2023), and 34.5 percent among people focused on sports performance or body composition in a broader review (López-Gil et al., 2023), with very large differences between studies. One review found that the estimate depended on which questionnaire was used, and the other attributes part of its figure to the questionnaire’s flaws (section 3). An athlete’s energy and nutrient needs belong to a sports dietitian.
- People taking medication for diabetes, blood pressure, or blood thinning
- This page’s searches did not identify a study of clean eating in people taking these medicines. A client on any of them who plans a substantial change in eating keeps the prescriber informed before making it, and the low blood sugar steps under When to refer apply.
- People with a history of disordered eating
- This is the clearest reason for care. Researchers have warned that a “clean” diet can make restrictive eating look like a healthy choice and so hide it (Ambwani et al., 2020), and in a study of 77 women in eating disorder treatment, read only as a conference abstract, 64 percent reported having followed at least 1 of 4 popular diets, clean eating among them (Shahan et al., 2020). Where clean eating is part of an eating disorder, the treatment team leads; for a coach, the clinician or care plan comes first.
3. The evidence
(select the heading to open or close this section)
Research on clean eating itself is small: studies of what people mean by the term, 2 surveys linking it with dietary restraint or disordered eating symptoms, 1 analysis of recipes, 1 menu cost study and 1 short randomized experiment with social media content. Much of the related research measures orthorexia with questionnaires rather than asking about clean eating. The certainty ratings for each outcome are in the Evidence rating and review record at the end of the page.
Following clean eating advice: diet and dietary restraint
In a 2018 survey, 762 Australian women aged 17 to 55, most recruited online through a university research webpage and clean eating, health and diet social media pages and the rest on a university campus, said how often they followed dietary advice from clean eating websites: 195 did so sometimes or more often, 199 seldom, and 368 never. Women who followed the advice were more likely to meet the national guideline for fruit, about 63 percent against about 51 percent of women who never did, and for meat and alternatives, about 56 percent against about 42 percent. No clear difference was detected for vegetables, dairy, grains or discretionary foods. They also had higher dietary restraint scores, 2.93 against 2.42 on a 1 to 5 scale (Allen et al., 2018).
The survey measured each woman once, so it cannot say which came first, and the authors note that women with high restraint may seek out clean eating sites. Food intake came from 10 short questions, and many participants were recruited through the kind of pages the survey asked about. Dietary restraint, a deliberate effort to limit how much or what one eats, is a known risk factor for disordered eating but is not disordered eating itself. Asked their opinion, about 40 percent of the women who answered were positive about clean eating, 40 percent negative and 20 percent neutral; negative answers called it a fad, unrealistic, or damaging because it sorts food into good and bad (Allen et al., 2018).
Views of clean eating and disordered eating symptoms
In a 2019 study at a small U.S. college, 148 undergraduates, 70 percent of them women, defined clean eating in their own words and then read 5 short stories about a student on a “clean” diet: alkaline, vegan, gluten-free, meal replacement, or a balanced diet based on federal guidance. Every story described the diet bringing guilt and anxiety, getting in the way of schoolwork and straining friendships. Students rated the balanced story as the healthiest and cleanest and the one they would most consider adopting, despite the distress it described. Students who rated the diets as healthier, cleaner or more worth adopting had higher scores on an eating disorder screening questionnaire (correlations of 0.30 to 0.41) and on an orthorexia questionnaire (0.33 to 0.61) (Ambwani et al., 2019).
After 43 students were removed for finishing too quickly or failing checks on whether they had understood the stories, at most 105 remained, and the paper does not state how many entered each analysis. The design measures attitudes toward stories, not what students ate, and a correlation measured once cannot show which came first (Ambwani et al., 2019).
Three more studies were read only in their abstracts. Among 77 women in eating disorder treatment programs, 64 percent reported having followed at least 1 of 4 popular diets, clean eating among them; clean eating was not associated with how severe the eating disorder was, and the study had no comparison group without an eating disorder (Shahan et al., 2020). In 463 female undergraduates, a story character described as following clean eating was viewed more positively than characters on other diets or not dieting, while a character with orthorexia was among those viewed most negatively (Weinstock & Mazzeo, 2022). A survey of 269 women aged 18 to 30 related viewing fitspiration and clean eating images on Instagram to thin-ideal internalization and disordered eating symptoms, but its abstract states the direction of that result in 2 conflicting ways, so this page does not rely on it (Wu et al., 2022).
Clean eating content on social media
One randomized experiment has tested clean eating content directly. At a German university, 129 female students were randomly assigned to scroll for 5 minutes through an Instagram profile built for the study. One profile showed clean eating posts, with captions about nutrients, avoiding “bad” foods and time spent preparing food; the other showed “foodie” posts about taste, enjoyment and eating with friends; neither showed people. Positive mood fell in both groups, with no difference between them. Negative mood fell slightly after the foodie profile, from 1.33 to 1.23 on a 1 to 5 scale, and stayed level after the clean eating profile, from 1.44 to 1.43; the difference between groups was statistically significant (p = 0.04) but small (Cossu et al., 2025).
The groups did not differ in which snack they chose afterward, and existing eating disorder, orthorexia or body dissatisfaction scores did not predict who reacted more. The paper describes its body satisfaction scale in 2 opposite directions and has other internal reporting problems, so this page reports no direction for the body satisfaction result. Five minutes with 1 profile is far from months of daily use, and the authors call their findings tentative (Cossu et al., 2025).
Surveys point the same way but cannot show cause. In an online survey of 680 women who followed health food accounts, more Instagram use went with higher orthorexia questionnaire scores; 49.0 percent screened positive at a cutoff of below 35 and 90.6 percent at below 40, 2 cutoffs in use in the research (Turner & Lefevre, 2017). A 2026 scoping review of 31 studies of orthorexia and social media found them mostly cross-sectional and describes the relationship as running in both directions (Awad et al., 2026).
Orthorexia: how it is measured and how common it is
Without agreed criteria, orthorexia is measured with questionnaires, and the most used, the ORTO-15, has well-known weaknesses. A 2023 review of 75 studies with 30,476 people in 18 countries found that 27.5 percent scored positive on the ORTO-15 at a cutoff of below 35, with very large differences between studies; its authors attribute part of that high figure to the questionnaire’s flaws and note that no reliable study of how common orthorexia is exists (López-Gil et al., 2023). The survey of health food followers above shows the problem: moving the cutoff changed the positive share from 49 to 91 percent (Turner & Lefevre, 2017). A positive screen is not a diagnosis.
Orthorexia scores overlap with eating disorder symptoms. A 2022 meta-analysis of 36 studies found a moderate correlation between orthorexia and eating disorder symptoms (r = 0.36) and a smaller one with obsessive-compulsive symptoms (r = 0.21); neither association shows that orthorexia, or clean eating, causes either kind of symptom (Zagaria et al., 2022). In a convenience sample of older adults in China, by contrast, higher scores went with healthier habits and better wellbeing, and the authors themselves questioned what the scale was capturing (He et al., 2021). Together these show how hard it is for a questionnaire to separate a healthy interest in food from a harmful fixation, which is the same distinction a coach cannot make from the word “clean.”
What clean eating recipes contain and cost
A 2018 study compared 86 breakfast, snack, smoothie, dessert and treat recipes from 13 popular clean eating blogs with 86 matched recipes without the label. Per serving, the clean eating recipes had more protein (8.1 against 5.7 grams), fat (15.8 against 12.4 grams) and fiber (5.0 against 2.8 grams), and no clear difference was detected in energy, sugar (21.1 against 23.2 grams) or sodium. Under the United Kingdom’s front-of-pack traffic-light criteria, both sets averaged moderate to high in fat, saturated fat, salt and sugar. The study analyzed recipes, not what people ate (Dickinson et al., 2018).
A 2024 study rebuilt a 2,000-calorie Western-style menu, used in feeding research, from less-processed ingredients. The new menu took 20 percent of its energy from ultraprocessed foods against 67 percent in the original, yet scored about the same on the Healthy Eating Index (44 against 43 out of 100), spoiled sooner (a median of 35 against 120 days to expiration) and cost more, $15.91 against $9.85 per person a day at fall 2023 grocery prices (Hess et al., 2024). Processing level alone did not decide diet quality in this comparison, which measured menus, not people.
Body weight, as measured
This page’s searches did not identify a study that measured body weight in people following clean eating against people who were not. In the national survey of young people, fewer than 1 percent defined clean eating as losing weight (Ambwani et al., 2020). This archive reports weight as something studies measured, not as a goal or as a sign that an approach worked.
Where the evidence is absent
In searches run October 21, 2026, and rechecked for newly indexed records on November 8, 2026, this page did not identify a study measuring heart disease, diabetes, cancer, death or body weight in people following clean eating; a study following people over time after they take up clean eating to see whether disordered eating or orthorexia develops; a trial of clean eating advice in any group; or a study of clean eating practices in children, pregnancy, older adults, athletes, or people taking diabetes, blood pressure or blood-thinning medication. The studies opened for this page do not provide systematic adverse-event rates for following a clean eating diet. The Instagram experiment measured immediate negative mood and body satisfaction after viewing content; it did not test months of following the diet.
The research that exists comes mostly from young women in Australia, Europe and the United States, relies on self-report, and measures people once. Orthorexia research is limited by the lack of an agreed diagnosis and by questionnaires that classify many people as positive (Donini et al., 2022; López-Gil et al., 2023).
4. Key terms
- Clean eating
- A popular label, with no agreed definition, for eating foods seen as whole, natural or unprocessed and avoiding foods seen as processed or impure.
- Dietary restraint
- A deliberate effort to limit how much or what one eats; a known risk factor for disordered eating, not a disorder in itself.
- Ultraprocessed food
- The most processed of the 4 groups in the Nova system that many researchers use, often made with ingredients not used in home cooking.
- Orthorexia nervosa
- A proposed term, not a formal diagnosis, for a harmful fixation on eating only foods a person considers healthy or pure, governed by rigid rules.
- Healthy orthorexia
- A proposed name for an interest in healthy eating that causes no harm, used to separate it from orthorexia nervosa.
- DSM-5-TR
- The current edition of the Diagnostic and Statistical Manual of Mental Disorders, a main manual for diagnosing mental disorders.
- ICD-11
- The current edition of the International Classification of Diseases, published by the World Health Organization.
- Cross-sectional survey
- A study that measures people once, so it can show an association but not which came first.
- Correlation
- A number from -1 to 1 describing how closely 2 measures rise and fall together; it does not show cause.
- Randomized experiment
- A study that assigns people to conditions by chance, so that differences afterward can be attributed to the condition.
- Positive screen
- A questionnaire score past a set cutoff, which suggests a closer look but is not a diagnosis.
5. References and Resources
Allen, M., Dickinson, K. M., & Prichard, I. (2018). The dirt on clean eating: A cross sectional analysis of dietary intake, restrained eating and opinions about clean eating among women. Nutrients, 10(9), Article 1266. https://doi.org/10.3390/nu10091266
Ambwani, S., Shippe, M., Gao, Z., & Austin, S. B. (2019). Is #cleaneating a healthy or harmful dietary strategy? Perceptions of clean eating and associations with disordered eating among young adults. Journal of Eating Disorders, 7, Article 17. https://doi.org/10.1186/s40337-019-0246-2
Ambwani, S., Sellinger, G., Rose, K. L., Richmond, T. K., & Sonneville, K. R. (2020). “It’s healthy because it’s natural.” Perceptions of “clean” eating among U.S. adolescents and emerging adults. Nutrients, 12(6), Article 1708. https://doi.org/10.3390/nu12061708
Awad, E., Alleva, J. M., El Khoury, C., Chamma, N., Martijn, C., & Rizk, R. (2026). Orthorexia nervosa and social media: A mixed-methods scoping review using a systematic methodology. PLOS ONE, 21(3), Article e0340219. https://doi.org/10.1371/journal.pone.0340219
Bulduk, Ö., & Güner, Ö. (2026). Individual and contextual factors associated with orthorexia nervosa tendency during pregnancy. BMC Pregnancy and Childbirth, 26, Article 556. https://doi.org/10.1186/s12884-026-09033-0
Cossu, G., Riefler, A., & Hartmann, A. S. (2025). Food and mood: How clean eating content on social media influences affect and body satisfaction in women. Frontiers in Psychology, 16, Article 1531142. https://doi.org/10.3389/fpsyg.2025.1531142
Dickinson, K. M., Watson, M. S., & Prichard, I. (2018). Are clean eating blogs a source of healthy recipes? A comparative study of the nutrient composition of foods with and without clean eating claims. Nutrients, 10(10), Article 1440. https://doi.org/10.3390/nu10101440
Donini, L. M., Barrada, J. R., Barthels, F., Dunn, T. M., Babeau, C., Brytek-Matera, A., Cena, H., Cerolini, S., Cho, H.-h., Coimbra, M., Cuzzolaro, M., Ferreira, C., Galfano, V., Grammatikopoulou, M. G., Hallit, S., Håman, L., Hay, P., Jimbo, M., Lasson, C., . . . Lombardo, C. (2022). A consensus document on definition and diagnostic criteria for orthorexia nervosa. Eating and Weight Disorders, 27(8), 3695-3711. https://doi.org/10.1007/s40519-022-01512-5
Hafstad, S. M., Bauer, J., Harris, A., & Pallesen, S. (2023). The prevalence of orthorexia in exercising populations: A systematic review and meta-analysis. Journal of Eating Disorders, 11, Article 15. https://doi.org/10.1186/s40337-023-00739-6
He, J., Zhao, Y., Zhang, H., & Lin, Z. (2021). Orthorexia nervosa is associated with positive body image and life satisfaction in Chinese elderly: Evidence for a positive psychology perspective. International Journal of Eating Disorders, 54(2), 212-221. https://doi.org/10.1002/eat.23400
Hess, J. M., Comeau, M. E., Scheett, A. J., Bodensteiner, A., & Levine, A. S. (2024). Using less processed food to mimic a standard American diet does not improve nutrient value and may result in a shorter shelf life at a higher financial cost. Current Developments in Nutrition, 8(11), Article 104471. https://doi.org/10.1016/j.cdnut.2024.104471
Lichtenstein, A. H., Khera, A., Anderson, C. A. M., Appel, L. J., DeSilva, D. M., Gardner, C., Hu, F. B., Jones, D. W., Petersen, K. S., & American Heart Association. (2026). 2026 dietary guidance to improve cardiovascular health: A scientific statement from the American Heart Association. Circulation, 153(18), e1285-e1295. https://doi.org/10.1161/CIR.0000000000001435
Lombardo, C., Cerolini, S., Zagaria, A., Vacca, M., & Rodgers, R. F. (2026). A combined network analysis for orthorexia nervosa, obsessive compulsive, and eating disorder symptoms. Nutrients, 18(8), Article 1179. https://doi.org/10.3390/nu18081179
López-Gil, J. F., Tárraga-López, P. J., Soledad Hershey, M., López-Bueno, R., Gutiérrez-Espinoza, H., Soler-Marín, A., Fernández-Montero, A., & Victoria-Montesinos, D. (2023). Overall proportion of orthorexia nervosa symptoms: A systematic review and meta-analysis including 30 476 individuals from 18 countries. Journal of Global Health, 13, Article 04087. https://doi.org/10.7189/jogh.13.04087
National Institute of Diabetes and Digestive and Kidney Diseases. (n.d.). Low blood glucose (hypoglycemia). niddk.nih.gov/health-information/diabetes/overview/preventing-problems/low-blood-glucose-hypoglycemia
Negowetti, N., Ambwani, S., Karr, S., Rodgers, R. F., & Austin, S. B. (2022). Digging up the dirt on “clean” dietary labels: Public health considerations and opportunities for increased federal oversight. International Journal of Eating Disorders, 55(1), 39-48. https://doi.org/10.1002/eat.23585
Shahan, M., Jordan, K., Siegfried, N., Douglas, C., & Lawrence, J. (2020). Eating disorder risk and severity as a function of following self-administered popular diets [Abstract]. Current Developments in Nutrition, 4(Suppl. 2), 565. https://doi.org/10.1093/cdn/nzaa046_065
Turner, P. G., & Lefevre, C. E. (2017). Instagram use is linked to increased symptoms of orthorexia nervosa. Eating and Weight Disorders, 22(2), 277-284. https://doi.org/10.1007/s40519-017-0364-2
Weinstock, M., & Mazzeo, S. E. (2022). College students’ perceptions of individuals following popular diets and individuals with orthorexia nervosa. Eating Behaviors, 47, Article 101671. https://doi.org/10.1016/j.eatbeh.2022.101671
Wu, Y., Harford, J., Petersen, J., & Prichard, I. (2022). “Eat clean, train mean, get lean”: Body image and health behaviours of women who engage with fitspiration and clean eating imagery on Instagram. Body Image, 42, 25-31. https://doi.org/10.1016/j.bodyim.2022.05.003
Zagaria, A., Vacca, M., Cerolini, S., Ballesio, A., & Lombardo, C. (2022). Associations between orthorexia, disordered eating, and obsessive-compulsive symptoms: A systematic review and meta-analysis. International Journal of Eating Disorders, 55(3), 295-312. https://doi.org/10.1002/eat.23654
Evidence rating and review record
Each row identifies the method, source, outcome or recommendation, and comparison being rated. Ratings from different tools are not interchangeable. The source is either a named published review or this page’s own judgment. The rating in the banner is for dietary restraint, and the banner names the tool that produced it. A high rating means the research is solid; it does not mean clean eating is right for everyone, and it does not mean a coach should recommend it.
Every rating on this page uses GRADE, the same 4-level scale used in the clinical guidelines you will meet in practice. High means further research is very unlikely to change the conclusion. Moderate means further research may change it. Low means further research is likely to change it. Very low means the conclusion is uncertain. GRADE rates each outcome separately, so an approach can be low for one outcome and very low for another.
The 4 ratings below are this page’s own assessments, assigned from the evidence under the archive’s September 28, 2026 decision; no published GRADE rating of a clean eating outcome was found in the sources opened for this page, and all 4 rest on studies read in full. Each row names its evidence and exactly which comparison it rates, and its reasons say where the rating starts and lower it 1 level for each concern they name. Where a paper reports no confidence interval, the row says so and labels any interval as this page’s approximation. Three outcomes are not rated, which means the evidence opened for this page is not adequate to rate them, not that clean eating has no effect or that it is safe.
On a narrow screen, scroll the table sideways to see every column.
| Outcome | Certainty | Who rated it, and for which comparison | Why it is rated this way |
|---|---|---|---|
| Dietary restraint scores, by following advice from clean eating websites (the banner rating) | Very low | This page’s own judgment, using the GRADE approach, from Allen et al. (2018); the survey came with no published quality rating. Comparison: Australian women aged 17 to 55 who said they followed dietary advice from clean eating websites sometimes, often or very often (195), against women who never did (368), measured once on a 10-item restraint scale scored 1 to 5; mean 2.93 (standard deviation 0.76) against 2.42 (0.80), a difference of 0.51 points. | Observational studies start at low. Lowered 1 level for risk of bias: most women were recruited online, many through clean eating, health and diet social media pages; exposure was 1 self-reported question; the comparison was not adjusted for any other factor; and the authors note that women with high restraint may seek out these sites, so the association may run the other way. That gives very low. Not lowered for imprecision: the paper reports no interval, and this page’s approximation from the reported means, standard deviations and group sizes, treating the 2 groups as independent, is 0.38 to 0.64 points (95% confidence interval); the analysis included 732 women rather than the 762 enrolled, which changes that approximation little: even removing all 30 missing responses from both groups, a deliberately extreme bound rather than the actual analysis count, gives 0.37 to 0.65. The whole interval lies above 0.16 points, a small standardized difference of 0.2 standard deviations on this scale and the threshold this page uses. Inconsistency could not be judged from 1 survey. Not lowered for indirectness: the evidence measures the population and exposure this row names. Publication bias was not assessed: there is too little evidence to judge it. Not raised for the size of the association, because of the risk of bias. Restraint is a risk factor for disordered eating, not disordered eating itself, and the survey cannot show that clean eating causes it. |
| Meeting the national fruit guideline, by following advice from clean eating websites | Very low | This page’s own judgment, using the GRADE approach, from Allen et al. (2018). Comparison: as in the row above; met the Australian guideline for fruit, 123 of 195 women who followed the advice (about 63 percent) against 189 of 368 who never did (about 51 percent), this page’s calculation from the reported counts, a difference of about 12 percentage points. | Observational studies start at low. Lowered 1 level for risk of bias, for the reasons in the row above; in addition, intake came from 10 short self-report questions. That gives very low. Imprecision is also serious: the paper’s own test compared all 3 groups together and gives no interval for this difference; this page’s approximate 95% interval, from the reported counts, runs from about 3 to 20 percentage points, which includes differences both below and above 10 percentage points, the threshold this page uses for a difference that matters. Very low is already the lowest level, so this concern is recorded but cannot lower the rating further. Inconsistency could not be judged from 1 survey. Not lowered for indirectness. Publication bias was not assessed. Meeting a fruit guideline is a measure of intake, not of health, and no clear difference was detected for vegetables. |
| Eating disorder and orthorexia symptom scores, by favorable views of “clean” diets | Very low | This page’s own judgment, using the GRADE approach, from Ambwani et al. (2019); the study came with no published quality rating. Comparison: correlations between U.S. undergraduates’ average ratings of 5 short stories about “clean” diets (how healthy, how clean, how willing to adopt) and their own questionnaire scores, measured once at one college: 0.30 to 0.41 with an eating disorder screening questionnaire and 0.33 to 0.61 with an orthorexia questionnaire, in at most 105 of 148 students after exclusions. | Observational studies start at low. Lowered 1 level for risk of bias: a convenience sample from one college’s psychology participant pool; 43 of 148 students excluded for speed or failed comprehension checks; every story described a diet causing distress and impairment, so the ratings measure tolerance of impairing diets as much as views of clean eating; self-report; and no adjustment for any other factor. That gives very low. Not lowered for imprecision: the paper reports no intervals, and with about 103 to 105 students, a number this page infers from the exclusions and the reported degrees of freedom because the paper does not state it, this page’s approximate 95% interval for the smallest correlation, 0.30, is about 0.11 to 0.47, so even its lower end lies above 0.1, the conventional threshold for a small correlation and the one this page uses. The very low rating does not depend on this precision judgment. Inconsistency could not be judged from 1 study. Not lowered for indirectness: the row names attitudes, which the study measured. Publication bias was not assessed. A correlation measured once cannot show which came first. |
| Negative mood immediately after 5 minutes of viewing clean eating content on Instagram | Low | This page’s own judgment, using the GRADE approach, from Cossu et al. (2025); the experiment came with no published quality rating. Comparison: a study-built Instagram profile of clean eating posts against a study-built profile of “foodie” posts about taste and enjoyment, randomly assigned to 129 female university students in Germany (64 and 65), measured before and right after viewing; negative mood 1.44 to 1.43 against 1.33 to 1.23 on a 1 to 5 scale, a difference in change of about 0.09 points; group by time interaction p = 0.04, partial eta squared 0.03. | Randomized trials start at high. Lowered 1 level for risk of bias: participants saw which content they were given and rated their own mood moments later, so expectations about the study could shape their answers; how the allocation sequence was concealed, beyond block randomization in a spreadsheet, is not described; the clean eating group started with slightly higher negative mood; and the follow-up test the paper reports for the foodie group (an F value of 0.02 with p below 0.01) is internally inconsistent, which leaves the analysis uncertain. Lowered 1 level for imprecision: the paper reports no interval; treating its interaction test, F(1, 127) = 4.32, as a squared t statistic, this page’s conditional approximation for the difference of about 0.09 points, computed from rounded means, is about 0.004 to 0.18 points (95% interval). That interval includes differences both below and above 0.10 points, a small standardized effect of 0.2 baseline standard deviations (about 0.49) and the threshold this page uses, and the same holds across the range the rounding allows. Two levels down from high gives low. Inconsistency could not be judged from 1 trial. Not lowered for indirectness: the row names brief exposure in young women, which is what was tested, though it says nothing about months of daily use. Publication bias was not assessed. The rating describes an immediate mood difference after viewing content, not the effect of following a diet over weeks or months. |
| Body satisfaction immediately after viewing clean eating content on Instagram | Not rated | From Cossu et al. (2025), the same comparison as the row above. | The report describes the direction of its body-satisfaction scale inconsistently and contains internal reporting problems. The direction of the result cannot be established reliably from the report. Not rated is not an inference of no effect from the follow-up tests. Not rated means the evidence opened for this page is not adequate to rate, not that there is no effect. |
| Heart disease, diabetes, cancer, death, or body weight in people following clean eating | Not rated | The searches for this page did not identify a study. Comparison sought: people following clean eating against people who are not, over any period. | The searches run for this page on October 21, 2026, and rechecked for newly indexed records on November 8, 2026, did not identify a trial or cohort study measuring any of these outcomes in people following clean eating; the searches are recorded in the page’s working files. Not rated means there is no adequate evidence body, not that clean eating has no effect or that it is safe. The overlap with heart-health guidance described in section 1 is a comparison of advice, not evidence about people who call their eating clean. |
| Developing disordered eating or orthorexia after taking up clean eating | Not rated | The searches for this page did not identify a study. Comparison sought: people who take up clean eating against people who do not, followed over time. | The searches run for this page did not identify a study following people over time. The surveys in the rows above measured people once, and the study of women in eating disorder treatment was read only as a conference abstract and had no comparison group (Shahan et al., 2020), so none of them can show whether clean eating comes before disordered eating. Not rated means the evidence is not adequate, not that there is no risk. |
How this archive talks about weight. This page reports what studies measured, including body weight, and does not treat weight change as a goal or as a measure of an approach’s success. Weight a person loses without meaning to can be a sign of a health problem, and where it is, the page names it as a reason to see a clinician. People are described the way the study described them.
Last reviewed: Reviewed and approved by James Peth, MS, MPH, PhD, on November 14, 2026, after a combined final review of the sixth batch of 5 archive pages, the repairs it asked for and a targeted recheck. Written October 21, 2026, and repaired November 8, 2026.
What changed: First version, repaired November 8, 2026, after outside review: the definition and labeling wording, the account of harms, the uncertainty calculations behind the ratings (no rating changed), and the sources now read in full. The judgments in the table are this page’s own, assigned from the evidence. Some sources could be read only as abstracts, and the page says so where each is used; none of them carries a rating.
Next scheduled review: August 2027. Pages built on annual guidance are rechecked each year.
Diet Reference Archive | One James
Prepared by James Peth, MS, MPH, PhD
© 2026–2027 One James. All rights reserved.
This page is for education and coaching practice. It is not medical or dietary advice.
