Obesity is a major contributor to global burden of disease and its prevalence continues to increase steadily in most countries, including China. A decade of CKB research has provided important new evidence about the determinants and health consequences of overweight and obesity in China where the mean adiposity levels and main disease patterns differ from those in Western countries. This research also helped to establish what is ‘biologically’ optimal (e.g. average body mass index [BMI] of 20–22 kg/m2 in rural China), as opposed to ‘statistically’ normal (e.g. average BMI of 30 kg/m2 in the USA), which is associated with the lowest disease burden.
Patterns, correlates and determinants
Overall the mean baseline BMI in CKB was 23.6 kg/m2 and mean waist circumference (WC) was 80.0 cm, with only 4% being obese (i.e. BMI ≥30 kg/m2), as opposed to 30%-50% of the population in the UK and USA. Across the 10 CKB study areas the prevalence varied 7-fold from 1.6% to 11.1% (Chen Z, Int J Epidemiol 2011). In both men and women, lack of physical activity and excess sedentary leisure time were independently and jointly associated with greater adiposity (Du HD, Am J Clin Nutr 2013). Different types of physical activities showed somewhat different associations with adiposity, with for example an inverse and apparently linear association between adiposity and levels of both commuting-related and household activities, particularly in men (Du HD, BMC Public Health 2014). In women, adiposity was also significantly associated with many reproductive factors. There was an inverse association with age at menarche and at first birth, and among post-menopausal women a positive association with age at menopause and reproductive years. (Yang L, Int J Epidemiol 2017) Adiposity had a non-linear positive association with having given birth, but no association with breastfeeding duration.
Cross-sectional associations with blood pressure
Among participants who were not taking any blood pressure-lowering drugs, there was a linear positive association of adiposity with blood pressure throughout the BMI range (18-32 kg/m2) examined (Chen Z, Int J Epidemiol 2015). On average each 10 kg/m2 higher BMI was associated with 16 mmHg higher systolic blood pressure (SBP), which was about twice as steep as that reported in Western populations. Among individuals treated with antihypertensive therapies, the association was only about a third as strong (i.e. about 5 mmHg). There was evidence, particularly in rural areas, of stronger associations in older than younger individuals (Chen X, J Human Hypertension 2015). Among the different measures of adiposity, general adiposity (e.g. BMI, body fat percentage) was more strongly associated with SBP compared with measures of central adiposity (e.g. waist circumference, waist/hip ratio).
Prospective associations with cardio-metabolic and other diseases
Detailed prospective analyses have examined the associations of adiposity with a wide range of conditions. Throughout the range examined (18–32 kg/m2), BMI showed a strong positive and log-linear relationship with incident Type 2 diabetes (T2DM), with hazard ratios (HRs) per 1-SD of 1.98 in men and 1.77 in women (P<0.001). However, BMI at age 25 years was only weakly positively associated with T2DM (men 1.09, women 1.04 per 1-SD) and after adjustment for baseline BMI, it was reversed (Fiona B, Diabetes Care 2018). Compared with BMI, the association of waist circumference with T2DM was stronger, but a similar gender difference existed (2.13 in men vs 1.91 in women per 1-SD; P<0.001). Mutual adjustment for BMI and WC attenuated these associations, especially those of BMI, further suggesting better predicative value of central adiposity for risk of T2DM. The observed gender difference in HRs probably reflected a greater propensity of men than women for visceral fat accumulation and for higher levels of insulin resistance.
Among a proportion of so-called metabolically healthy obese individuals (i.e. without elevated blood pressure, abnormal lipid profiles, and low insulin sensitivity), there was a four-fold higher risk of type 2 diabetes compared with those with a metabolically healthy normal weight (Song ZM, Eur J Endocrinology 2022).
Moreover, conversion from metabolically healthy to unhealthy status during follow-up was associated with a four-fold higher risk of developing T2DM, while those with a persistent metabolic unhealthy status had an eight-fold higher risk. These findings indicated that metabolic health is a transient state, and across all BMI categories individuals who converted from a metabolically healthy to an unhealthy status greatly increased the risk of T2DM.
BMI was strongly positively and log-linearly associated with risk of ischaemic stroke (IS), with an HR of 1·30 per 5 kg/m² higher BMI. This risk estimate was generally consistent with that predicted by equivalent differences in SBP associated with 5 kg/m² higher BMI. However, there was no association between BMI and intracerebral haemorrhage (ICH) across the normal range (i.e. BMI<25 kg/m²), with elevated risk only at BMI≥25 kg/m². Consequently, the HR per 5 kg/m² higher BMI was less extreme (1·11), and much weaker than that predicted by the corresponding difference in SBP (1·48). Other adiposity measures showed similar associations with stroke types. After adjustment for SBP, the HR for IS was greatly attenuated (1.30 to 1·05), while for ICH, it was reversed (1.11 to 0·73), suggesting that for ICH, leanness, either per se or through some other factors, might increase risk, offsetting the protective effects of lower blood pressure (Chen Z, Lancet Global Health 2018).
Although adiposity is a strong risk factor for cardio-metabolic diseases, many previous studies have reported that among individuals with diabetes, higher adiposity was associated with lower cardiovascular disease (CVD) and overall mortality (i.e. so-called ‘obesity paradox’ phenomenon). CKB found that among those with diabetes, BMI was positively and log-linearly associated with risk of incident CVD. There were, however, U-shaped associations of BMI with CVD and overall mortality, with the lowest mortality risk at BMI 22.5–24.9 kg/m2 and highest at BMI <18.5 kg/m2. In those with self-reported or screen-detected diabetes the excess mortality risk at low BMI was particularly marked for deaths occurring immediately after CVD onset (i.e. short-term mortality) (Iona A, BMJ Open Diabetes Research & Care 2021). The opposing associations of BMI with CVD incidence and mortality (particularly short-term mortality) at low BMI suggest it may be important for individuals with diabetes to maintain normal weight.
Apart from cardio-metabolomic diseases, adiposity has also been associated with a range of other diseases in CKB, including U-shaped associations with risks of sepsis-related mortality (Weng L, Critical Care 2020), incident chronic obstructive pulmonary diseases (Li J, Eur Respir J 2020), positive associations with risks of colorectal cancer (Pang Y, Br J Cancer, 2018), liver cancer (Pang Y, Int J Cancer 2019), non-alcohol fatty liver disease (NALFD), and hepatobiliary diseases. For NALFD, the association was log-linear throughout the BMI range examined, with HR per 5 kg/m2 of 2.81 (Pang Y, Scientific Report 2019). Strong positive associations were also seen for other measures of adiposity. For hepatobiliary diseases, a meta-analysis of genetic data in CKB and UK Biobank supported the cause-effect nature of the association, with HRs per 1-SD higher genetically-derived BMI being 1.55 (1.30-1.84) for chronic liver disease and 1.42 (1.22-1.64) for gallbladder disease (Pang Y, JAMA Network Open 2020).
Novel mechanisms linking adiposity with specific diseases
To explore the novel mechanisms linking adiposity with disease risks, CKB undertook observation and genetic analyses of adiposity with 92 proteins among 628 participants without prior history of cancer. Overall, BMI was associated with 30 proteins, with 27 positive associations (e.g. interleukin-6 and 18, hepatocyte growth factor) and 3 inverse associations (e.g. ligand, carbonic anhydrase-9). Genetic analyses, though with more limited power, showed directionally consistent findings as observation analyses. Of the 30 BMI-associated biomarkers, 10 (including interleukin-6, interleukin-18, and hepatocyte growth factor) were nominally associated with incident CVD (Pang Y, JAMA Cardiol 2021). In a study of 176 people with NAFLD and 180 sub-cohort individuals, we used a Metabolon metabolomic platform that quantifies 1208 metabolites and demonstrated that BMI was significantly associated with 199 metabolites after adjustment for multiple testing, with directionally consistent findings in genetic analyses. Moreover, 35 metabolites were associated with NAFLD risk, of which 15 were also associated with BMI, suggesting these biomarkers may lie on the pathway between adiposity and NAFLD (Pang Y, Am J Clin Nutr 2021).
Impact of research
In China diabetes now affects more than 10% of the adult population. By applying CKB findings, we estimated that the increase in adiposity accounted for about 50% of the increase in T2DM since 1980, even though most people still have a BMI within the so-called normal range (i.e. less than 25 kg/m²). High adiposity (BMI greater than 23 kg/m²) also accounted for 15% of total stroke cases and large proportions of other diseases.
Research based on CKB data has informed policy development in China to promote healthy living in order to combat the rising prevalence of obesity and the associated disease burden. Several key CKB findings were incorporated into prevention guidelines for chronic diseases in China (Chinese Circulation Journal, 2020), including the Healthy China 2030 Plan, in which control of obesity was a key actionable area. In future CKB will continue to provide reliable evidence about the health consequence of adiposity on a much wider range of fatal and non-fatal conditions, to assess the genetic determinants of different measures of adiposity, and to explore the novel biological mechanisms linking adiposity with specific diseases.