Increasing obesity and its complications pose a tremendous health burden worldwide. According to the 2020 National Report on Chronic Disease and Nutrition in Chinese Residents, the prevalence of overweight or obesity reached 50·7 %(1). Moreover, our previous research findings indicated a higher prevalence of obesity among Tibetan adults compared with the national average(Reference Li, Zhang and Jian2). Substantial changes resulting from the interaction of environmental and dietary factors may contribute to the increased risk of obesity.
Dietary pattern (DP) analysis, rather than the analysis of single nutrients or food, proves to be valuable in examining the associations between diet and obesity, as well as diet-related diseases(Reference Martinez-Gonzalez, Bes-Rastrollo and Serra-Majem3). Diet plays an important role in modulating inflammatory responses and development(Reference Ahluwalia, Andreeva and Kesse-Guyot4). Although the precise mechanism is not yet fully understood, it is widely accepted that dietary factors and DP alter gut microbiome profiles (e.g. diversity, composition and metabolic activity), which can affect the progression of obesity and related non-communicable diseases (e.g. diabetes and CVD)(Reference Chakaroun, Olsson and Bäckhed5).
Recent evidence has suggested that diets rich in pro-inflammatory components, such as processed meats and high-energy beverages, are associated with an increase in inflammatory profiles, consequently elevating the risks of obesity and related non-communicable diseases(Reference González-Becerra, Ramos-Lopez and Barrón-Cabrera6,Reference Li, Zhan and Huang7) . Healthy diets, such as the Mediterranean diet and plant-based DP identified using an exploration analysis, were characterised by high amounts of vegetables and unrefined cereals and associated with lower inflammation and reduced risk of obesity(Reference Seifu, Fahey and Hailemariam8–Reference Eichelmann, Schwingshackl and Fedirko11). As a result, investigating the inflammatory potential of DP and their effects on obesity has become an area of special interest. However, neither priori (e.g. healthy eating index, recommended food score and diet quality index) nor exploratory approaches (e.g. principal component analysis, factor and cluster analyses) can utilise the information of the biological pathways for diseases and the underlying dietary data. Consequently, novel hybrid approaches, such as the reduced rank regression (RRR) model, have been proposed to identify DP that are more closely related to obesity risk(Reference Weikert and Schulze12,Reference Ocké13) .
The Tibetan Plateau, characterised by its high altitude of over 4000 m above sea level, experiences low temperatures and hypobaric hypoxia. It stands as one of the world’s highest inhabited regions. Since 2005, under initiative by the Chinese government, native pastoralists on the Tibetan Plateau have resettled and moved to urban areas(Reference Bessho14). As a result, there have been substantial and swift changes in their diet and other lifestyle, partially contributing to increased risks of obesity(Reference Peng15). Urbanisation has diversified DP with greater access to Western foods, fruits and vegetables(Reference Kong, Yang and Gong16). While some studies have investigated DP among residents of the Tibetan Plateau using a priori or exploratory approaches(Reference Kong, Yang and Gong16–Reference Peng, Liu and Malowany19), their specific and prospective associations with obesity remain unclear.
This study aimed to identify inflammation-associated DP using inflammatory factors as response variables with RRR methods and explore their associations with obesity in an urbanised Tibetan adult cohort.
Methods and materials
Study population and study design
This community-based prospective open cohort study has been conducted since 2018 in two settled Tibetan communities located in the suburbs of Golmud City, which is the fourth largest city on the Tibetan Plateau, situated at an elevation of 2800 m above sea level. These settlements comprised approximately 4000 individuals who were resettled from their original pastoral areas at an elevation 4000 m above sea level. Since 2007, some have completely abandoned their nomadic lives and settled in towns, while others retain their nomadic habits and travel between pastoral areas and towns. Additionally, certain individuals continue to engage in animal husbandry in pastoral areas.
The baseline survey was conducted in November and December 2018, and 1003 adults (≥18 years) were enrolled. Between December 2021 and May 2022, a follow-up study was conducted with 1611 adults completing the survey (514 followed up and 1097 newly included) in the two settled Tibetan communities. Participants were excluded if they met the following criteria: 1) age missing or less than 18 years (n 5); 2) non-Tibetan ethnicity (n 31); 3) missing anthropometric data (n 16); 4) missing biomarkers (n 124) and 5) incomplete FFQ (n 98). Finally, 1826 participants were included in this study, of whom 514 attended both waves and provided the necessary data, resulting in 2578 person-years observations.
This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects/patients were approved by the Ethics Committee of the Qinghai University Medical College (No.2021-15). Written informed consent was obtained from all participants.
Outcome variables
Height and weight were measured using a calibrated, fully automated height and weight scale provided by the local health centre (IPR-scale08 model, Improvau Science & Technology Co., China). Waist circumference was measured using a non-stretching soft tape at the midpoint between the rib margin and the iliac bone. The average of the two measurements was used. BMI was calculated as weight (kg)/height(m) 2.
Overweight and obesity were defined based on Chinese national standards (BMI: normal weight, 18·5–23·9 kg/m2; overweight, 24·0–27·9 kg/m2; overweight and obesity, ≥ 24·0 kg/m2; and general obesity, ≥ 28·0 kg/m2)(20). Central obesity was defined as waist circumference ≥ 90 cm in men or ≥ 85 cm in women(20).
Exposure variables
Dietary intakes
Dietary intakes data were collected using a forty-one-item FFQ by trained and qualified investigators at each wave through face-to-face interviews. A second quality control review of the completed FFQ was conducted on the same day of the survey. If necessary, a second round of face-to-face or telephonic interviews was conducted to ensure accuracy. The FFQ included the frequency of consumption of twenty-six food groups, such as tsamba (a kind of dough made with roasted barley flour and butter with water), refined carbohydrates, fried pasta, whole grains, vegetables, fresh fruits, meat, processed foods, Tibetan cheese, butter tea, milk tea, beverages and snacks. The FFQ was modified from that used in the 2015 China Nutrition and Health Survey(Reference Huang, Wang and Wang21) and validated for the Tibetan population(Reference Peng, Liu and Malowany19).
High-sensitivity C-reactive protein levels and prognostic nutritional index
Blood samples were collected from the participants at the Second People’s Hospital of Golmud after an overnight fast of at least 10 h. Plasma samples were collected in heparin-containing tubes, while serum samples were collected in plain tubes without anticoagulants. Serum albumin and lymphocyte levels were assessed using an automated biochemical analyser (Beckman Coulter AU 480) following standard procedures. The lymphocyte count was calculated through a leucogram using the lymphocyte percentage and the value of the lymphocytes (ml). Serum high-sensitivity C-reactive protein (hs-CRP) levels were determined using an immunoturbidimetric assay with a lower limit of detection of 0·1 mg/l. The prognostic nutritional index (PNI) was calculated using the following formula: albumin (g/l) + 5 × total lymphocyte count × 109/l(Reference Kinoshita, Onoda and Imai22).
Covariate assessment
Smoking status was classified as never smoked, past smokers and current smoker. Alcohol drinking status was classified as never drunk, past drinker or current drinker. Physical activity levels were categorised as light, moderate or vigorous, depending on individuals’ intensities in occupational and leisure time physical activity over the past year. Based on the cut-off point of having lived in ultra-high pastoral areas, the altitude levels were categorised into high (≤4 months/year) or ultra-high latitude (>4 months/year).
Statistical analyses
First, we derived inflammation-related DP from twenty-six food groups using RRR. Subsequently, we calculated dietary scores for each participant at both waves and examined their associations of DP with the risks of overweight, general obesity and central obesity.
In the RRR analysis, we used twenty-six food groups as predictor variables and log-transformed hs-CRP levels and PNI as response variables. The number of factors extracted using the RRR is equal to the number of intermediate response variables (i.e. 2 in the current analysis). Food groups with factor loadings ≥|0·30| were considered significant contributors to the identified DP. Pearson’s correlation coefficients were used to assess the associations between DP scores and responses. DP scores for all participants were calculated from a linear combination of standardised intakes for all food groups, which were weighted by factor score coefficients automatically generated by the statistical software. We then categorised the participants into three groups based on tertiles (T1 to T3) of DP scores in ascending order.
Mixed-effects models were used to obtain OR and 95 % CI for overweight, general obesity and central obesity, respectively. Four multivariable models were fitted: Model 1 was adjusted for age and sex; Model 2 was further adjusted for marital status, education, insurance, smoking status, drinking status and physical activity; Model 3 was additionally adjusted for altitude; and Model 4 was the same as Model 3 but included those who attended both two waves of the survey. The models were selected to sequentially control for key demographic, socio-economic, lifestyle and environmental factors, enabling a clearer analysis of the association between DP and obesity. Model 4 further strengthens the findings by focusing on participants with complete longitudinal data to generate the longitudinal association. In these models, fixed effects are exposures that do not change throughout the study, such as age, sex, marital status, education, insurance, smoking status, drinking status, physical activity and altitude. Random effects were included to account for individual-level variability, ensuring that within-participant variations were appropriately modelled. In the subgroup analyses, the multiplicative interaction between inflammation-related DP and covariates (sex, age, education, smoking, drinking, physical activity and altitude levels) was included by adding a product term to the regression model, and the OR were reported. We also performed sensitivity analyses to confirm the effect of latitude, using 6 months as the cut-off point, on the associations between diet and overweight and/or obesity.
Statistical analyses were performed using Stata software (version 17.0; Stata Corporation). The RRR analysis was conducted using the user-written program rrr in Stata. For all analyses, two-sided P values of .05 were considered statistically significant.
Results
Dietary patterns and food groups
Two inflammation-related DP were derived using RRR methods. DP-1 explained 0·73 % of the variation in hs-CRP and 5·00 % of the variation in PNI, whereas DP-2 explained 3·37 % of the variation in hs-CRP and 5·38 % of the variation in PNI from 1391 participants in 2022 (online Supplement Table 1). DP-1 was characterised by high intakes of sugar-sweetened beverages, savoury snacks and poultry and a low intake of tsamba. DP-2 was characterised by high intakes of poultry, pork, animal offal, and fresh fruits and a low intake of butter tea (Fig.1).
Fig. 1. Factor loading matrix of dietary patterns derived from reduced rank regression among 1397 Tibetan adults in 2022.
Baseline characteristics across the tertiles of dietary patterns
In 2018 and 2022, a total of 1826 participants were surveyed in our study at baseline and/or follow-up, with 514 participants being followed up for a total of 1542 person-years. Table1 summarised the demographic and lifestyle characteristics of 1826 participants in 2018 and 2022 according to the tertiles of DP. The mean age of participants was 43·1 years with a sd of 14·4 years. Participants with higher DP-1 scores were younger, highly educated, married, more likely to be smokers or drinkers, and less likely to exhibit general obesity (T1 v. T3:40·7 % v. 28·0 %) or central obesity (T1 v. T3:61·6 % v. 43·5 %). Conversely, participants who scored higher on DP-2 were older and more likely to exhibit general obesity (T1 v. T3:29·8 % v. 38·6 %) or central obesity (T1 v. T3:46·2 % v. 61·7 %). Moreover, participants with higher scores on both DP were more likely to live at high altitudes and participate in light or medium physical activity.
Table 1. Baseline characteristics of Tibetan adults according to tertiles of dietary pattern scores in 2018 and 2022 (n 1826)
T, tertiles; NRCMS, New Rural Cooperative Medical Scheme; URBMI, Urban Resident Basic Medical Insurance; UEBMI, Urban Employee Basic Medical Insurance.
Central obesity: waist circumference ≥ 90 cm for men or waist circumference ≥ 85 cm for women. Smoking status was classified as never smoked, past smokers and current smoker. Alcohol drinking status was classified as never drunk, past drinker or current drinker. Physical activity levels were categorised as light, moderate or vigorous, depending on individuals’ intensities in occupational and leisure time physical activity over the past year. The altitude levels were categorised into high (2800 m above sea level) or ultra-high latitude (4000 m above sea level) based on the cut-off point of having lived in ultra-high pastoral areas for less than or at least 4 months every year.
Data were presented as mean± sd for continuous variables and as n (%) for categorical variables.
Furthermore, we compared differences among the individuals who completed two waves, those who attended only once and those who dropped out in 2022. We found that individuals who attended both waves were older, had lower education levels, were less likely to drink alcohol and had a higher BMI than the other two groups (online Supplemental Tables 2–3). Due to the unique characteristics of the study subjects (e.g. residing at high altitudes and actively participating in pastoral activities), the follow-up rate was 51·4 %.
Associations of dietary patterns with overweight and/or obesity
The associations between DP and overweight and/or obesity risks are presented in Table2. After adjusting for sociodemographic factors, lifestyle factors and altitude, higher DP scores were associated with a higher risk of being overweight and obesity. The OR (95 % CI) across tertiles for DP-1 were 1·00, 1·28 (1·03, 1·58), 1·37 (1·07, 1·77), and for DP-2 were 1·00, 1·19 (0·96, 1·48), and 1·48 (1·18, 1·85), respectively. A higher DP-2 score was associated with an increased risk of overweight and central obesity. The OR (95 % CI) across the tertiles were 1·00, 1·72 (0·93, 3·17) and 2·83 (1·47, 5·46) for overweight, and for central obesity were 1·00, 1·18 (0·81, 1·72) and 2·25 (1·49, 3·39), respectively. Additionally, the association of overweight and central obesity with DP-2 was consistent among those who had data from the two waves; the OR (95 % CI) across the tertiles for overweight were 1·00, 1·07 (0·36, 3·17) and 4·17 (1·20, 14·46), and for central obesity were 1·00, 1·36 (0·74, 2·48) and 2·74 (1·45, 5·19), respectively, while the associations between DP1 and overweight or overweight and obesity, as well as DP2 and overweight and obesity disappeared.
Table 2. Associations between inflammatory-related dietary patterns and weight status among Tibetan adults (n 1826) *
Values were OR and 95 % CI from mixed-effect logistic models using the data collected during 2018–2022.
* n 514. T, tertile. Model 1 was adjusted for sex and age. Model 2 was further adjusted for marital, insurance, education, smoking status, drinking status and physical activity. Model 3 was further adjusted for altitude levels based on the cut-off point of having lived in pastoral areas for at least 4 months. Model 4 was the same as Model 3 but included those who attended both two waves of the survey. Pfor trend was based on a logistic regression analysis for the categorical variables, assigning median values to the tertile categories of each dietary pattern. Overweight and obesity were defined as BMI ≥ 24.0 kg/m2, overweight as 24.0–27.9 kg/m2 and obesity as ≥ 28.0 kg/m2. Central obesity was defined as waist circumference ≥ 90 cm in men or waist circumference ≥ 85 cm in women. Smoking status was classified as never smoked (never smoked and not currently smoking), past smokers (formerly smoked in their lifetime and currently a non-smoker) and current smoker (currently smoked). Alcohol drinking status was classified as never drunk (never drunk and not currently drinking), past drinker (formerly drunk in their lifetime and currently a non-drinker) or current drinker (currently drunk). Physical activity levels were categorised as light, moderate or vigorous, depending on individuals’ regular activity over the past year. The altitude levels were categorised into high (2800 m above sea level) or ultra-high latitude (4000 m above sea level) based on the cut-off point of having lived in pastoral areas for at least 4 months.
To examine the influence of altitude levels, we conducted sensitivity analyses by categorising altitude levels into high and ultra-high altitude using 6 months of those living in the patrol area as the cut-off point. The associations were similar to those using 4 months as the cut-off point (data not shown).
Subgroup analyses observed a significant interaction between DP-1 and altitude levels in relation to the risks of being overweight (P for interaction = .007) and overweight and obesity (P for interaction = .006) (Fig.2). Specifically, the positive associations of DP-1 with the risks of overweight (OR 95 % CI 3·48 (1·60, 7·58)) and overweight and obesity (OR 95 % CI 3·21 (1·80, 5·73)) were only significant at high altitude level, respectively. However, no such effects were observed for DP-2 (Table3).
Fig. 2. OR and 95 % CI of weight status (overweight; overweight and obesity; obesity; central obesity) according to the dietary pattern-1 from stratified analyses by altitude among 1826 Tibetan adults mixed-effect logistic models adjusted for sex, age, marital status, insurance, education, smoking, drinking and physical activity.
Table 3. OR and 95 % CI of overweight, obesity and central obesity with dietary pattern-2 from stratified analyses among Tibetan adults (n 1826)
Overweight and obesity were defined as BMI ≥ 24.0 kg/m2, overweight as 24.0–27.9 kg/m2 and obesity as ≥28.0 kg/m2. Central obesity was defined as waist circumference ≥90 cm in men or waist circumference ≥85 cm in women. Smoking status was classified as never smoked, past smokers and current smoker. Alcohol drinking status was classified as never drunk, past drinker or current drinker. Physical activity levels were categorised as light, moderate or vigorous, depending on individuals’ intensities in occupational and leisure time physical activity over the past year. The altitude levels were categorised into high (2800 m above sea level) or ultra-high latitude (4000 m above sea level) based on the cut-off point of having lived in ultra-high pastoral areas for less than or at least 4 months every year.
Discussion
Principal results
The present study is the first to derive two inflammation-related DP using RRR methods in a cohort study among urbanised Tibetan adults. The findings highlighted the significant role of DP associated with inflammation in obesity risk, and that altitude levels may impact these associations. Specifically, DP-1 predicted a higher PNI and was characterised by higher intakes of sugar-sweetened beverages, savoury snacks, and poultry and a low intake of tsamba. DP-2 predicted higher levels of circulating hs-CRP and PNI, characterised by high intakes of poultry, pork, animal offal, and fresh fruits and a low intake of butter tea. The hs-CRP is a well-established marker of chronic low-grade inflammation, closely linked to CVD and metabolic diseases, while PNI serves as a combined marker of nutritional and immune status, frequently used to assess prognosis in inflammatory and cancer-related conditions(Reference Li, Zhong and Cheng23,Reference Luan, Tsai and Yang24) . Participants who scored higher on either DP were more likely to have an increased risk of being overweight and/or obesity. Furthermore, participants with higher DP-2 scores were more likely to have central obesity. Finally, altitude might modify the associations between DP-1 and overweight and/or obesity. This study provides valuable insights into the complex interplay among diet, inflammation and weight status in Tibetans.
Compared with research conducted on DP in the general population, there is limited evidence available for the Tibetan population. One study in particular identified four DP using principal component analysis among Tibetan residents(Reference Zhou, Li and Liu17). Two other studies identified three DP using principal component analysis among an urbanised Tibetan population(Reference Peng, Liu and Malowany19,Reference Li, Tang and Liu25) . However, these three studies relied on cross-sectional data and applied principal component analysis.
Supportively, our previous study also identified three DP in urbanised Tibetan population. RRR is an innovative approach in nutrition epidemiology that incorporates prior knowledge of diseases and their pathways. This approach maximises the variability in response variables and bolsters the evidence of causality between diet and diseases. Our study identified two distinct DP associated with inflammation using the RRR with the response variables hs-CRP and PNI. Elevated hs-CRP, which indicates low-grade inflammation, is considered a potential risk factor for obesity and visceral adiposity(Reference Ridker26), and the PNI, which reflects the interplay among inflammation, immunological status and nutrition, is regarded as a promising biomarker with predictive and prognostic value for obesity(Reference Zhang, Li and Zhang27). In our study, both DP explained the greater variation in PNI than in hs-CRP, and DP-2 explained the greater variation in hs-CRP than DP-1.
Poultry emerged as the second-highest positively loading item in both DP identified in our study. Interestingly, previous research has reported that high intakes of chicken or pork proteins for 12 weeks led to a significant increase in systemic inflammatory factors in rats(Reference Zhang, Song and Zhao28). These findings were further supported by the results from the UK Biobank Study in population(Reference Papier, Hartman and Tong29). Moreover, inflammation-related DP in other studies have been characterised by high intakes of refined grains, processed meat, sugar-sweetened beverages and sweet snacks, which were consistent with our findings(Reference Vermeulen, Brouwer and Stronks30–Reference Yan, Ren and Lin32). Nevertheless, DP with lower intakes of vegetables and fruits were positively associated with inflammation(Reference Vermeulen, Brouwer and Stronks30,Reference Jacobs, Kroeger and Schulze31) , which was contrasting to the findings in DP-2. Our study identified DP-2 as a DP associated with a high intake of fresh fruits. This inconsistency may be explained by the cooking habits of vegetables in the Tibetan population. The primary cooking methods for vegetables are frying or stir-frying, which typically involves a greater amount of cooking oil and may result in excessive energy intake. These practices have been found to be closely associated with inflammation(Reference Moreno-Franco, Rodríguez-Ayala and Donat-Vargas33). However, our dietary surveys did not consider oil intake, and further studies are warranted to clarify this. Finally, both DP identified foods with Tibetan characteristics such as low intakes of tsamba and butter tea. Tsamba is made from barley and rich in dietary fibre and phytochemicals such as β-glucan; butter tea, another popular Tibetan beverage, has been found to have the highest levels of conjugated linoleic acids and n-3 fatty acids, which are known for their anti-inflammatory properties(Reference Agyare and Liang34,Reference Lin, Guo and Lu35) .
Participants with higher scores in both DP were more likely to be overweight and obesity. Higher DP-2 scores were also associated with an increased risk of central obesity. These findings were consistent with the Melbourne Collaborative Cohort Study, a population-based study conducted in Australia, which found that a higher dietary inflammatory score at baseline predicted general and central obesity(Reference Hodge, Karim and Hébert36). Similar results were reported in a Korean Genome and Epidemiology study using a dietary inflammatory index to evaluate dietary characteristics(Reference Khan, Kwon and Shivappa37). However, it is worth noting that dietary intakes in these previous studies collected solely at a single time point.
Interestingly, we observed that the association between overweight and obesity and DP-1 was evident only among individuals living in high altitude but not in ultra-high altitude levels. This phenomenon may be attributed to an additional increase in energy expenditure that compensates for heat loss and maintains body temperature in an ultra-high altitude environment. Additionally, the adaptation to the low-pressure, low-oxygen environment at ultra-high altitude may have altered their gut microbiota and energy metabolism, potentially leading to negative results at ultra-high altitude level. However, it is important to note that the sample size at ultra-high altitude was limited, comprising only 22·9 % of the total study population. As a result, further studies are necessary to elucidate the underlying mechanisms behind these observations. It is noteworthy that when dividing the altitude into two groups using a 6-month living in the ultra-high altitude as the cut-off point, there were no significant changes in the associations between DP and weight status. To ensure a relatively balanced sample size between groups, we used a 4-month cut-off point; however, these results should be interpreted with caution.
The characteristics of our DP and their associations with overweight and obesity suggest that changes in the living environment pose new challenges. Numerous studies have indicated an elevated risk of non-communicable diseases, particularly obesity and other metabolic disorders, during the urbanisation process(Reference Peng, Liu and Malowany19,Reference Li, Tang and Liu25,Reference Peng38) . Dietary patters, as a critical part of lifestyle, have undergone profound changes, especially in a transitioning economy such as China, where consumers are switching from traditional Chinese food, largely characterised by grains and vegetable, to processed or prepared food that contain condensed energy and Na. Urbanised Tibetans migrating from pastoral areas (ultra-high altitude) to urban areas (high altitude) have greater access to a diverse range of foods; however, this shift may contribute to an increased prevalence of overweight and obesity. Urbanisation, accompanied by changes in living altitude levels, may interactively lead to a substantial decline in the consumption of traditional foods (e.g. low intakes of tsamba and butter tea) and an increase in the consumption of industrial foods (e.g. high intakes of higher intakes of sugar-sweetened beverages and savoury snacks), leading to unhealthy DP.
Supportively, our previous study also identified three DP in urbanised Tibetan population. We found that individuals with pastoral DP (higher intakes of Tibetan cheese, tsamba and butter/milk tea) were less likely to have central obesity, and modern DP (pulses, poultry, offal and processed meat) was positively associated with elevated blood pressure and elevated TAG; additionally, the associations may be modified by altitude levels(Reference Wang, Wang and Shi39). These findings, along with our own, highlight the potentially adverse effects of urbanisation on our population. Although our research focused on Tibetan adults, their diets exhibited similar overall changes from traditional to modern as observed in other indigenous or rural populations experiencing urbanisation. Hence, we suggested that individuals who have undergone lifestyle changes to maintain a high intake of their traditional foods and a low intake of processed foods to stay healthy. Further studies are warranted to address this phenomenon and related mechanisms more comprehensively.
Limitations and strengths
This study has some limitations. First, the subjects were enrolled via voluntary participation rather than random sampling, whereas age and sex distributions were similar to those who did not participate. Second, the dietary survey did not capture specific food intake amounts. Nevertheless, previous studies have shown that food portion measurement is usually poorly assessed by FFQ and that food intake frequency, rather than portion size, matters most for individual differences(Reference Thompson, Subar and Brown40). Third, the follow-up period was relatively short, and the cohort sample size was limited. However, the associations of DP2 with both overweight and central obesity were consistently observed, even when the analysis was restricted to individuals who attended both waves. Fourth, their nomadic status was not captured, although the altitude levels, to some extent, indicated the status. Finally, the follow-up rate is 51·4 %. Those individuals who were followed up were older, had lower education levels, were less likely to be current drinker, and had a higher BMI. Thus, the findings may not be generalisable to the larger urbanised Tibetan population, which might have introduced selection bias. According to our knowledge, our study is the first open cohort among Tibetan population. Further cohort studies are encouraged, and cautions are needed when interpreting the results.
Despite these limitations, this cohort study of a special study population is a valuable contribution to the field. This is the first prospective cohort study to identify inflammation-related DP using RRR and to explore their associations with obesity. The findings have important implications for public health policies and interventions.
Conclusions
In conclusion, the urbanised Tibetan population is currently experiencing a nutritional transition that has heightened their vulnerability to overweight and/or obesity. These findings underscore the importance of implementing dietary interventions tailored to the specific ethnic context. Inflammation-related DP, particularly those characterised by high consumptions of animal and ultra-processed foods, in combination with living at high altitudes, may further exacerbate the risks. This emphasises the significance of adopting a balanced diet that takes into account environmental changes to mitigate these risks. It also emphasises the importance of promoting the quality and quantity of plant-based (e.g. diversified grain like whole grain and vegetables) and traditional Tibetan food (e.g. tsamba and butter tea) in dietary recommendations.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114524003337
Acknowledgements
The authors express our gratitude to the medical personnel from the Golmud Disease Prevention and Control Center, Tanggula Mountain Town Health Center, and Golmud Children’s Hospital for their assistance in this study.
This work was supported in part by the Key Program of Regional Innovative Development Joint Funds of the Chinese National Natural Science Foundation of China (grant number: U24A20774), Key R&D and Transformation Program of Qinghai (grant number: 2023-QY-204), Key Science and Technology Project of Qinghai Province (grant number: 2021-NK-A3), Science Promotion and Communication Fund, Chinese Nutrition Society (grant number: CNS-SCP2020–040) and the National Key R&D Program of China (2017YFC0907200, 2017YFC0907201).
X. S. was responsible for idea article conceptualisation. W. J. was responsible for data analysis. Y. C. was responsible for first draft writing. X. T., R. L., B. Z., H. W., L. Z. and Y. Z. were responsible for rewriting. T. K. and Z. S. were responsible for rewriting and checking the reasonableness of the statistical methods. Y. W. and W. P. were responsible for overseeing the whole process of writing the article and finalising the rewrite.
None declared.
