Dietary patterns and mortality in clinical obesity: comparative risks of meat-based, pescatarian, and vegetarian diets

Abstract

Background

Plant-based and pescatarian diets are generally considered healthier alternatives to meat-based diets; however, their specific effects on mortality risk among individuals with clinical obesity remain unclear. This study aims to investigate the associations between different dietary patterns and the risks of all-cause and premature mortality in patients with clinical obesity.

Methods

A nationwide cohort study using UK Biobank data included 154,986 participants with clinical obesity. Dietary patterns among these individuals were categorized as regular meat-eaters, occasional meat-eaters, pescatarians, vegetarians, and vegans. Accelerated failure time (AFT) models were used to evaluate the associations between dietary patterns and all-cause and premature mortality. Interaction analyses, subgroup analyses, and sensitivity analyses were further conducted to assess the robustness of the findings.

Results

Among individuals with clinical obesity, the all-cause mortality rates for regular meat-eaters, occasional meat-eaters, pescatarians, and vegetarian/vegans were 12.75%, 13.01%, 7.58%, and 8.15%, respectively. The corresponding rates of premature mortality were 4.31%, 3.95%, 3.02%, and 4.00%. In fully adjusted AFT models, compared to regular meat-eaters, pescatarians had significantly longer time to all-cause mortality (Time Ratio [TR]: 1.14, 95% CI: 1.04–1.26) and premature mortality (TR: 1.30, 95% CI: 1.01–1.69). Subgroup analysis revealed a significant interaction between dietary pattern and sex for premature mortality (p for interaction = 0.01), with the protective effect of the pescatarian diet being more pronounced in men.

Conclusions

Among individuals with clinical obesity, a pescatarian diet was associated with better survival outcomes, whereas vegetarian and vegan diets did not show similar benefits. This finding provides epidemiological evidence for future research and dietary counselling in populations with clinical obesity.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12937-026-01368-w.

Background

Over the past five decades, the global prevalence of obesity has reached pandemic levels, becoming an even more serious public health threat than tobacco use [1]. According to the World Health Organization (WHO) [2, 3], by the year 2000, the number of adults with obesity had already exceeded 300 million worldwide. In 2019, an estimated 5 million deaths from noncommunicable diseases (NCDs) were attributable to higher-than-optimal body mass index (BMI). In recognition of its growing clinical and public health burden, obesity has been formally defined as a chronic and progressive disease. Its impact on health ranges from an increased risk of premature mortality to a substantial reduction in overall quality of life [2–4].

However, BMI alone has limitations in accurately identifying individuals at high risk, as it does not account for body fat distribution or its functional consequences. To resolve this issue, The Lancet Diabetes & Endocrinology Commission introduced the novel concept of clinical obesity in January 2025 [5]. Clinical obesity is characterized by an excess of body fat that directly leads to measurable organ dysfunction and/or symptoms, or significantly impairs the ability to perform normal daily activities. The concept of clinical obesity was introduced to help distinguish individuals whose excess adiposity contributes to disease from those who are metabolically healthy despite increased fat mass, thereby enabling stratified interventions and promoting more efficient allocation of healthcare resources [5].

In recent years, driven by concerns about animal welfare and environmental sustainability [6, 7], plant-based diets and various types of vegetarianism have been gaining increasing popularity. According to the latest data from the World Population Review [8] and the World Animal Foundation [9], there are nearly 1.5 billion vegetarians worldwide, accounting for approximately 22% of the global population. In the United Kingdom, consumer spending on vegetarian and plant-based alternatives increased from £541 million in 2010 to £1.158 billion in 2022 [10]. A growing body of research suggests that diets emphasizing fish or plant-based foods may offer health benefits [11] compared to meat-based diets, including lower risks of cancer, diabetes [12] and ischemic heart disease.

Despite these potential benefits, evidence regarding the association between dietary patterns and mortality among individuals with clinical obesity remains limited. This question is clinically relevant because clinical obesity identifies a high-risk population with established obesity-related impairment, in whom dietary factors may have important implications for long-term survival. Most previous studies [13, 14] have examined dietary patterns in general populations rather than in individuals meeting the newly proposed definition of clinical obesity.

Therefore, we aimed to investigate the associations of different dietary patterns with all-cause and premature mortality among individuals with clinical obesity. By identifying dietary patterns associated with improved survival, this study may contribute to the evidence base for future nutritional studies and dietary recommendations in this high-risk population.

Materials & methods

Study population

The UK Biobank is a large-scale biomedical database and research resource aimed at improving human health, comprising 502,128 volunteer participants aged 40–69 years at baseline. These participants provided extensive genetic, health, and lifestyle information. In the overall cohort, we first excluded 10,750 participants with missing data on body composition indices, 75,155 participants with incomplete socioeconomic information, and 3,307 participants with missing data on other covariates or who were lost to follow-up. We then identified 200,904 individuals with clinical obesity based on the newly released criteria from The Lancet Diabetes & Endocrinology [5] (Supplementary Table 1). Subsequently, we excluded an additional 45,918 participants with missing information on lifestyle factors. Ultimately, 154,986 participants with clinical obesity were included in the final analyses (Supplementary Fig. 1). All participants provided informed consent, and the UK Biobank received ethical approval to ensure adherence to the highest ethical and legal standards (Project ID: 206612).

Dietary assessment and categorization

Dietary intake was assessed at baseline using a touchscreen food frequency questionnaire (FFQ) that asked participants about their consumption frequency of a wide range of foods, including red meat (such as beef, pork, and lamb/mutton), poultry (such as chicken, turkey, and other birds), processed meat products (such as bacon, sausages, and ham), oily and non-oily fish, eggs, and dairy products. Participants selected their intake frequency from several options ranging from “never” to “once or more daily.” Based on the responses, participants were categorized into five dietary groups [15]: (1) regular meat-eaters, defined as those consuming processed meat, red meat, or poultry more than five times per week; (2) low or occasional meat-eaters, defined as those consuming these meats no more than five times per week; (3) fish-eaters, defined as those who did not consume any red meat, processed meat, or poultry but consumed fish; and (4) vegetarians/vegans, defined as those who did not consume any meat or fish. Vegans, who did not consume any meat, fish, dairy products, or eggs, were included in the vegetarian group due to their small sample size.

Definition of clinical obesity

Based on the newly released obesity guidelines from The Lancet Diabetes & Endocrinology Commission, New-criteria obesity [5] is defined as follows: (1) body fat(BF)% measured by bioelectrical impedance analysis: ≥25% for men and ≥ 35% for women [16]. (2) Obesity BMI combined with anthropometric measures: Obesity BMI (≥ 30 kg/m², or ≥ 27.5 kg/m² for Asian populations [16, 17]) combined with at least one of the following: (a) Waist-to-hip ratio (WHR) ≥ 0.90 for men or ≥ 0.85 for women [18]; (b) Waist circumference (WC) ≥ 102 cm for men or ≥ 88 cm for women [19, 20]; (c) Waist-to-height ratio (WHtR) ≥ 0.5 [21]. 3. Multiple anthropometric abnormalities: at least two anthropometric indicators exceeding the above thresholds, regardless of BMI. 4. Severe obesity: BMI ≥ 40 kg/m². Participants with evidence of obesity-related organ or tissue dysfunction were further classified as having clinical obesity.

The organ system dysfunctions included in this study covered disorders of the central nervous system (CNS), respiratory, cardiovascular, metabolic, hepatic, renal, urinary, reproductive, and musculoskeletal systems. Diagnoses were based on self-reported illness codes prior to baseline (UK Biobank field 20002), ICD-10 codes from medical records, and relevant clinical measurements. Detailed definitions are provided in Supplementary Table 1.

The specific conditions considered included migraine, obstructive sleep apnea, chronic obstructive pulmonary disease (COPD), heart failure, atrial fibrillation, pulmonary embolism, hypertension, hyperglycemia, hyperlipidemia, metabolic dysfunction-associated steatotic liver disease (MASLD), diabetes, renal failure, diabetic nephropathy, urinary incontinence, polycystic ovary syndrome (PCOS), erectile dysfunction/impotence in men, male infertility, testicular disorders, osteoarthritis, joint pain, and others. Participants with Cushing’s syndrome or hypothyroidism were excluded.

Outcome ascertainment

The primary outcome was all-cause mortality. Mortality data were obtained through linkage with national death registries. Premature mortality was further defined as death occurring before the age of 70 [22]. Participants were followed from baseline until death, loss to follow-up, or the data processing date (August 2, 2025), whichever occurred first. Mortality data were systematically collected through the National Health Service (NHS) in England (covering England and Wales) and the NHS Central Register, National Records of Scotland, with further details available on the UK Biobank website (https://www.ukbiobank.ac.uk ).

Covariates

Covariates included age, sex (male, female), ethnicity (White, non-White), BMI. Education level was categorized as high-quality education (college or university degree), medium-quality education (including A levels/AS levels or equivalent; O levels/GCSEs or equivalent; CSEs or equivalent; NVQ or HND or HNC or equivalent; other professional qualifications such as nursing, teaching), or low-quality education (none of the above). Income status was divided into < £18,000, £18,000–£51,999, and ≥ £52,000. Health-related covariates included: smoking status (healthy defined as never smoked or smoked < 100 cigarettes in lifetime), alcohol intake (healthy defined as ≤ 2 drinks/day for men and ≤ 1 drink/day for women), and physical activity (defined as ≥ 150 min of moderate-intensity or ≥ 75 min of vigorous-intensity activity per week).

Statistical analysis

Baseline characteristics were described using means (standard deviations) for continuous variables and frequencies (percentages) for categorical variables. The associations between dietary patterns and baseline characteristics were evaluated using nonparametric tests and Chi-square tests.

The proportional hazards assumption for the Cox proportional hazards model was assessed using Schoenfeld residual tests and was not fully satisfied for certain covariates (e.g., age). Accelerated failure time (AFT) models were therefore used to examine the association between dietary patterns and mortality among participants with clinical obesity. The AFT model directly models the log-transformed survival time and provides intuitive Time Ratios (TRs) with corresponding 95% Confidence Intervals (CIs) [23], and it has been used in previous studies [24]. To determine the parametric distribution, four commonly used AFT models, including Weibull, log-logistic, log-normal, and exponential models, were fitted using the same covariate set. Model fit was evaluated using the Akaike information criterion (AIC) and Bayesian information criterion (BIC). The Weibull model had the lowest AIC and BIC values and was selected as the primary AFT model. The model comparison results are provided in Supplementary Fig. 2 and the Supplementary Materials. Three models were constructed: Model 1 was unadjusted; Model 2 was adjusted for age, sex, BMI, and ethnicity; and Model 3 was further adjusted for education level, income level, smoking, alcohol consumption, and physical activity. Regular meat-eaters served as the reference group in all models.

We further evaluated multiplicative interactions between dietary patterns and key covariates, including age group (< 60 vs. ≥60 years), sex, ethnicity, income, education, smoking, alcohol intake, and physical activity, for both all-cause and premature mortality. Statistically significant interaction terms were interpreted as evidence of effect modification. Stratified analyses were conducted within each subgroup.

Several sensitivity analyses were performed to ensure the robustness of our findings. First, we excluded participants with survival time < 3 years to reduce reverse causality. Second, we reanalyzed the data without combining vegetarians and vegans to examine the consistency of results. Finally, given that distributional assumptions can influence effect estimates, we refitted the models using log-normal and log-logistic distributions alongside the main Weibull model to confirm that the main findings were not driven by the choice of distribution.

All statistical analyses were conducted using R version 4.3.3 on the UK Biobank Research Analysis Platform (UKB-RAP). A two-tailed p-value < 0.05 was considered statistically significant. P-values < 0.001 were reported as “<0.001,” and all other p-values were rounded to two decimal places.

Results

Population characteristics

Figure 1 presents the distribution of all-cause and premature mortality across different dietary groups among individuals with clinical obesity. Among the 154,986 participants, 106,121 were regular meat-eaters, 44,763 were occasional meat-eaters, 2,150 were pescatarians, 1,716 were vegetarians, and 236 were vegans. For analytical purposes, vegetarians and vegans were combined into a single group in Table 1, which summarizes the baseline characteristics across dietary categories.

Fig. 1.

Population distribution charts of dietary patterns and mortality outcomes in clinically obese populations

Table 1.

Baseline characteristics of participants with clinical obesity by dietary pattern

Clinical Obesity
Characteristic	Overall, N = 154,986	Regular meat-eater, N = 106,121	Occasional meat-eater, N = 44,763	Pescatarian, N = 2,150	Vegetarian/Vegan, N = 1,952	p-value*
Age, Mean (SD)	57.35(7.78)	57.07(7.83)	58.25(7.53)	55.41(7.75)	54.01(7.88)	< 0.001
BMI, Mean (SD)	29.82(4.38)	30.00(4.43)	29.45(4.24)	28.94(4.16)	28.92(4.55)	< 0.001
Sex						< 0.001
Female	52,332(33.77)	32,103(30.25)	18,242(40.75)	1,102(51.26)	885 (45.34)
Male	102,654(66.23)	74,018(69.75)	26,521(59.25)	1,048(48.74)	1,067(54.66)
Race						< 0.001
White	147,904(95.43)	102,030(96.14)	42,359(94.63)	1,991(92.60)	1,524(78.07)
Non-White	7,082(4.57)	4,091(3.86)	2,404(5.37)	159(7.40)	428(21.93)
Education levels						< 0.001
High	24,029(15.50)	16,171(15.24)	7,548(16.86)	159(7.40)	151(7.74)
Medium	79,966(51.60)	56,423(53.17)	21,909(48.94)	839(39.02)	795(40.73)
Low	50,991(32.90)	33,527(31.59)	15,306(34.19)	1,152(53.58)	1,006(51.54)
Income levels						< 0.001
<£18,000	35,416(22.85)	23,014(21.69)	11,520(25.74)	472(21.95)	410(21.00)
£18,000~£51,999	81,147(52.36)	55,759(52.54)	23,293(52.04)	1,061(49.35)	1,034(52.97)
≥£52,000	38,423(24.79)	27,348(25.77)	9,950(22.23)	617 (28.70)	508(26.02)
Health Smoking	94,082(60.70)	63,197(59.55)	28,118(62.82)	1,403(65.26)	1,364(69.88)	< 0.001
Healthy drinking	74,522(48.08)	48,149(45.37)	23,879(53.35)	1,214(56.47)	1,280(65.57)	< 0.001
Physical active	120,027(77.44)	82,155(77.42)	34,626(77.35)	1,745(81.16)	1,501(76.90)	< 0.001
Mortality	19,679(12.70)	13,534(12.75)	5,823(13.01)	163(7.58)	159(8.15)	< 0.001
Premature mortality	6,483(4.18)	4,574(4.31)	1,766(3.95)	65(3.02)	78 (4.00)	< 0.001

For continuous variables, we represent them as mean (SD), while categorical variables are expressed as counts (percentages)

Abbreviation: BMI Body Mass Index, SD Standard deviation

*The p-values for continuous variables were compared using the Kruskal- Wallis rank sum test, while categorical variables were assessed with the Chi-squared test

Of the total sample, 33.77% were women (n = 52,332) and 66.23% were men (n = 102,654), with the vast majority being of White ethnicity (95.43%, n = 147,904). More than half of the participants had a medium level of education (51.60%) and a moderate household income (£18,000-£51,999; 52.36%).

The median follow-up time of this study was 16.30 years. The overall all-cause mortality rate was 12.70% (n = 19,679), and the premature mortality rate was 4.18% (n = 6,483). All-cause mortality rates across dietary groups were 12.75% (n = 13,534) for regular meat-eaters, 13.01% (n = 5,823) for occasional meat-eaters, 7.58% (n = 163) for pescatarians, and 8.15% (n = 159) for vegetarians/vegans. Corresponding rates for premature mortality were 4.31% (n = 4,574), 3.95% (n = 1,766), 3.02% (n = 65), and 4.00% (n = 78), respectively.

AFT analyses

According to the AFT analysis results presented in Table 2, pescatarians had significantly prolonged survival compared with regular meat-eaters. In the unadjusted model (Model 1), the TR was 1.42 (95% CI: 1.28–1.57), indicating that the survival time was 42% longer. After adjusting for age, sex, BMI, and ethnicity (Model 2), the TR decreased to 1.16 (95% CI: 1.06–1.28), and remained significant in the fully adjusted model (Model 3: TR = 1.14, 95% CI: 1.04–1.26).

Table 2.

Accelerated Failure Time Model* Results for the Association of Dietary Patterns with All-Cause and Premature Mortality among Participants with Clinical Obesity

	Model 1		Model 2		Model 3
	TR (95%CI)	p	TR (95%CI)	p	TR (95%CI)	p
All-cause mortality
Regular meat-eater	Ref
Occasional meat-eater	0.99(0.97,1.01)	0.17	1.01(0.99,1.03)	0.51	1.01(0.99,1.03)	0.40
Pescatarian	1.42(1.28,1.57)	< 0.001	1.16(1.06,1.28)	0.002	1.14(1.04,1.26)	0.01
Vegetarian	1.34(1.22,1.49)	< 0.001	1.03(0.93,1.14)	0.57	1.02(0.92,1.12)	0.72
Premature mortality
Regular meat-eater	Ref
Occasional meat-eater	1.10(1.04,1.16)	0.002	1.00(0.94,1.06)	0.99	1.01(0.95,1.07)	0.76
Pescatarian	1.49(1.15,1.93)	0.003	1.33(1.03,1.73)	0.03	1.30(1.01,1.69)	0.04
Vegetarian	1.10(0.86,1.39)	0.45	0.99(0.77,1.25)	0.91	0.98(0.77,1.25)	0.87

Model 1: Unadjusted (no covariates)

Model 2: Adjusted for sex, age, race, and BMI

Model 3: Further adjusted for education level, income level, smoking status, alcohol consumption, and physical activity level

Abbreviation: AFT Accelerated failure time, BMI Body mass index, CI Confidence interval, TR Time Ratio

*The association between dietary patterns and all-cause mortality (including premature death) in the clinically obese population was assessed using an AFT model - a parametric survival analysis method where a TR > 1 indicates that covariates prolong event time, while TR < 1 suggests covariates shorten event time

Similarly, for premature mortality, pescatarians exhibited delayed onset compared with regular meat-eaters, with TRs of 1.49 (95% CI: 1.15–1.93), 1.33 (95% CI: 1.03–1.73), and 1.30 (95% CI: 1.01–1.69) in Models 1, 2, and 3, respectively (Table 2).

Interaction, subgroup, and sensitivity analyses

Figure 2 illustrates the interaction effects between dietary patterns and various covariates, including age, sex, ethnicity, education level, income level, healthy smoking status, healthy alcohol consumption, and physical activity. A significant interaction was observed only between sex and dietary pattern in relation to premature mortality (p for interaction = 0.01), where pescatarians showed a more pronounced prolongation of survival time compared to regular meat-eaters (TR = 1.66, 95% CI: 1.12–2.45). Additionally, the survival benefit associated with being a pescatarian was more evident among older adults (aged ≥ 60), men, individuals with lower educational attainment, those with lower income, and those who were physically active (Fig. 2).

Fig. 2.

Interaction and subgroup analyses of the associations between dietary patterns and mortality among participants with clinical obesity. A Interaction and subgroup analyses for all-cause mortality. B Interaction and subgroup analyses for premature mortality. Time ratios (TRs) and 95% confidence intervals (CIs) were estimated using accelerated failure time (AFT) models, with regular meat-eaters serving as the reference group. A TR greater than 1 indicates a longer time to the event, whereas a TR less than 1 indicates a shorter time to the event. P values for interaction were derived from multiplicative interaction terms between dietary patterns and the corresponding subgroup variables. Abbreviations: AFT, accelerated failure time; BMI, body mass index; CI, confidence interval; TR, time ratio.

Supplementary Table 2 presents the results of the four sensitivity analyses. In sensitivity analysis 1, excluding participants with a follow-up time of three years or less, pescatarians continued to show a consistent survival benefit (all-cause mortality: Model 3, TR = 1.12, 95% CI: 1.03–1.21; premature mortality: Model 3, TR = 1.26, 95% CI: 1.03–1.54). In sensitivity analysis 2, vegetarians and vegans were analyzed separately, and the survival benefit remained apparent only among pescatarians (all-cause mortality: Model 3, TR = 1.14, 95% CI: 1.04–1.26; premature mortality: Model 3, TR = 1.31, 95% CI: 1.01–1.69). In sensitivity analyses 3 and 4, we refitted the models using log-normal and log-logistic distributions, respectively. The observed associations remained robust under these alternative parametric assumptions (log-normal: all-cause mortality TR = 1.15, 95% CI: 1.04–1.28; premature mortality TR = 1.32, 95% CI: 1.01–1.74; log-logistic: all-cause mortality TR = 1.14, 95% CI: 1.04–1.26; premature mortality TR = 1.31, 95% CI: 1.01–1.70).

Discussion

In our study, we first identified patients with clinical obesity based on the definition proposed by The Lancet Diabetes & Endocrinology. We then examined all-cause and premature mortality rates across different dietary pattern subgroups within this clinically obese population. Using AFT models, we found that compared to regular meat-eaters, pescatarians demonstrated significantly prolonged time to all-cause and premature mortality among clinically obese individuals. This protective association was consistently confirmed in both subgroup and sensitivity analyses. However, no statistically significant associations were observed between vegetarian/vegan diets and either all-cause or premature mortality when compared to regular meat consumption in this population. These findings suggest that the pescatarian dietary pattern, rather than avoidance of meat alone, was associated with longer survival among individuals with clinical obesity.

In our study, we found that patients with clinical obesity had a high mortality rate (12.7%), which may be attributable to the fact that the newly defined criteria for clinical obesity already encompass the presence of chronic conditions such as cardiovascular disease and diabetes. This is consistent with the elevated mortality observed in previous studies among individuals diagnosed with obesity based on BMI or other adiposity indices [4, 22, 25].

In our study, using AFT models, we found that among clinically obese individuals, only pescatarians exhibited a significantly longer time to all-cause and premature mortality relative to regular meat-eaters. Previous studies [26, 27] have also reported inverse associations between fish consumption and conditions such as heart failure and cerebrovascular disease. Given that the clinical obesity definition encompasses individuals with existing cardiovascular disease, this may partly explain our findings.

We did not observe a significant association between vegetarian/vegan diets and lower mortality among individuals with clinical obesity. Previous studies [28–30] have suggested that vegetarian diets are associated with more favorable cardiometabolic risk profiles and lower risks of cardiovascular disease and ischemic heart disease. However, evidence regarding all-cause mortality remains inconsistent. Dinu et al. reported a lower incidence of and mortality from ischemic heart disease among vegetarians, but did not find significant reductions in all-cause mortality, overall cardiovascular mortality, cerebrovascular disease, or cancer mortality [29].

Several factors may explain the lack of a significant mortality benefit among vegetarians/vegans in our study. First, the health effects of plant-based diets may depend on overall dietary quality. Thompson et al. [14] reported that healthy plant-based diets lowered mortality risk, whereas unhealthy ones increased it. Similarly, a review by Boushey et al. [31] linked dietary patterns rich in vegetables, fruits, legumes, nuts, whole grains, unsaturated oils, and fish but low in red and processed meat, high-fat dairy, refined carbohydrates, and sweets to lower all-cause mortality. These findings suggest that the overall quality and nutrient density of plant-based diets may be more important than simply excluding animal foods.

Second, vegetarian and vegan dietary patterns are heterogeneous. In the present study, vegetarian/vegan diets were classified mainly according to avoidance of meat and fish, rather than detailed dietary quality. Therefore, this group may have included individuals consuming diets rich in whole grains, fruits, vegetables, legumes, and nuts, as well as individuals consuming more refined carbohydrates or unhealthy plant foods. In addition, with the increasing availability of plant-based meat alternatives, less healthy plant-based substitutes may increase exposure to unfavorable nutrients while reducing overall nutrient density [32]. Healthy plant-based foods typically encompass whole grains, fruits, vegetables, nuts, and legumes, whereas their unhealthy counterparts include fruit juices, sugar-sweetened beverages, refined grains, potato products, and sweets/desserts [33, 34]. Prior research indicates that the health effects of plant-based diets are contingent on overall dietary quality; specifically, unhealthful plant-based dietary patterns have been consistently linked to elevated risks of type 2 diabetes, coronary heart disease, and all-cause mortality [33–35]. Such heterogeneity within the vegetarian/vegan group may have attenuated the association between vegetarian/vegan diets and mortality in our study.

In contrast, pescatarian diets may combine lower intake of red and processed meat with continued fish consumption. In the present study, pescatarians showed longer time to all-cause and premature mortality among individuals with clinical obesity. This finding is consistent with previous evidence from the UK Biobank, which reported lower risks of several cardiovascular outcomes among fish-eaters compared with meat-eaters, with the strength of these associations varying across population subgroups and social factors such as deprivation level [27]. Previous dietary pattern research has also suggested that pescatarian diets may represent a pattern with relatively high overall diet quality, as reflected by higher Healthy Eating Index scores [36]. Therefore, the longer survival observed among pescatarians in our study may be related to the combined contribution of overall dietary quality, socioeconomic characteristics, and lifestyle-related factors. By contrast, the absence of a significant mortality benefit among vegetarians/vegans should be considered in the context of heterogeneity within plant-based dietary patterns. Although concerns have been raised that some strict vegetarian diets may be associated with lower total protein intake [37–39], less complete amino acid profiles [40, 41], and reduced digestibility of plant proteins [41, 42], such concerns are most relevant when diets are poorly planned, insufficient in total energy, or limited in food variety. Well-planned vegetarian and vegan diets can generally provide adequate total protein and essential amino acids when energy intake is sufficient and a variety of plant foods, including legumes, whole grains, nuts, seeds, vegetables, and fruits, are consumed [38]. In overweight adults, a low-fat vegan dietary intervention has also been shown to improve diet quality, body weight, fat mass, and metabolic outcomes [43]. Therefore, the null association observed among vegetarians/vegans in the present study may need to be interpreted in relation to variation in plant-based diet quality, total energy intake, protein sources, degree of food processing, and other unmeasured nutritional characteristics.

Moreover, our study found a significant interaction between dietary pattern and sex in relation to premature mortality among individuals with clinical obesity, with the protective effect of a pescatarian diet being more pronounced in men. Although the interaction did not reach statistical significance for all-cause mortality (p for interaction = 0.07), a stronger protective trend among men was also observed for this outcome. These findings may be partially explained by previous research suggesting that men, compared to women, are less likely to seek weight-loss interventions and more susceptible to the metabolic consequences of obesity [44]. Moreover, studies have shown that, compared to women, men are more likely to accumulate visceral adipose tissue (VAT) in the abdominal region [45, 46], a fat distribution pattern strongly associated with increased risk of metabolic diseases [47]. In this context, the stronger association observed among men may reflect the greater influence of overall dietary patterns in individuals with higher metabolic vulnerability. Pescatarian diets may represent a broader dietary pattern characterized by lower red and processed meat intake, continued fish consumption, and potentially higher overall dietary quality, rather than the effect of a single dietary component. Further studies with detailed dietary, body composition, and metabolic assessments are needed to clarify the mechanisms underlying these sex-specific associations.

Strengths and limitations

This study has several notable strengths. First, we analyzed a large-scale cohort containing comprehensive data on dietary patterns, along with extensive measurements of potential confounders and health outcomes. Second, the application of the novel clinical definition of obesity may provide more clinically relevant identification of obese populations compared to relying solely on BMI criteria.

However, several limitations should be acknowledged. First, dietary assessment was only performed at baseline, and dietary patterns may change over time. Second, due to the limited number of participants following a vegan diet, we combined vegans with vegetarians in the primary analysis, although we conducted sensitivity analyses separating these groups. Third, the baseline touchscreen dietary questionnaire used to define dietary patterns collected only limited information on selected food types and intake frequency. Therefore, we were unable to assess total energy intake, macronutrient intake such as protein, or the degree of food processing, and residual confounding related to these factors cannot be excluded. Fourth, the main analyses were based on baseline dietary data and therefore could not capture changes in dietary habits during follow-up. This may limit causal inference. Future studies incorporating repeated dietary assessments are needed to validate our findings. Fifth, several conditions, such as diabetes and hypertension, may lie on the causal pathway linking dietary patterns to mortality risk, and further mechanistic studies are needed. Finally, our findings apply specifically to participants with clinical obesity in the UK Biobank, and their generalizability requires validation in other populations, national cohorts, and more ethnically diverse settings.

Conclusion

Our study found that, compared to regular meat-eating individuals with clinical obesity, pescatarians had a significantly longer time to all-cause and premature mortality, while no significant association was observed between vegetarian diets and mortality. Notably, the protective effect of the pescatarian diet was more pronounced in men, suggesting that sex differences should be considered when designing personalized nutritional interventions.

Authors’ contributions

Conceptualization, Validation, and Writing – review and editing : YZL, TTL, BXY and SYL; Data curation, Writing – original draft : YZL and SYL; Formal analysis, Methodology, Software, and Visualization : SYL and TTL; Funding acquisition : YZL; Supervision : YZL and TTL. YZL is the guarantor and attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

Funding

This work is supported by the National Natural Science Foundation of China (Grant No. 82470826). The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Data availability

The data for this study were sourced from the UK Biobank (Project ID: 206612). UK Biobank data can be accessed upon reasonable request, and the official website is: [https://www.ukbiobank.ac.uk].

Declarations

Ethics approval and consent to participate

All UK Biobank participants provided informed consent, and the project underwent rigorous ethical review to ensure compliance with the highest ethical and legal standards (Project IDs: 206612).

Participants in the dietary surveys from which the data were drawn had already provided informed consent for the original studies.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.