Change of neck circumference in relation to visceral fat area: a Chinese community-based longitudinal cohort study

The study addresses whether longitudinal changes in neck circumference (NC), a simple anthropometric index of upper-body adiposity, are associated with visceral fat area (VFA) and abdominal obesity. While body mass index (BMI) is widely used, it does not capture adipose tissue distribution, and visceral adiposity is a key determinant of cardiometabolic risk. Gold-standard imaging (CT/MRI) quantifies VFA but is impractical for routine use due to radiation, time, and cost. NC is easy to measure, minimally influenced by respiration or meals, and has shown cross-sectional associations with VFA and metabolic risk. The research question is whether change in NC predicts change in VFA and the development of abdominal obesity over time in a Chinese community cohort.

Prior work demonstrates that fat distribution, especially visceral adiposity, is more strongly linked to metabolic and cardiovascular outcomes than overall adiposity. CT and MRI are gold standards for VFA but not feasible routinely. NC has emerged as an indicator of upper-body fat and cardiometabolic risk: cross-sectional studies, including prior work in Chinese communities, identified NC cut-offs correlating with MRI-assessed VFA and metabolic abnormalities. A Japanese cohort of postmenopausal women found NC changes correlated with waist circumference and body fat over three years. Additional cohorts link NC to metabolic syndrome, diabetes, hypertension, dyslipidemia, and fatty liver. Multiple cross-sectional studies across different populations (Chinese with type 2 diabetes, South Koreans, South Africans, and Caucasians) show NC correlates with abdominal obesity metrics. However, longitudinal evidence directly relating NC change to MRI-measured VFA had been lacking, motivating this study.

Design and population: Community-based longitudinal cohort in Shanghai. Baseline assessments were conducted in 2013–2014 with follow-up in 2015–2016 (follow-up 1.1–2.9 years; mean 2.1 ± 0.2 years). Exclusions included malignant tumors, steroid or thyroid hormone treatment, abnormal thyroid function or prior hyper-/hypothyroidism, and cardiovascular/cerebrovascular disease at baseline. Of 1943 with complete baseline data, 1421 with follow-up VFA were analyzed. Ethics approval was obtained, and written informed consent collected. Measurements: Standardized questionnaires captured medical history and medications. Anthropometry included height, weight, waist circumference (WC), and blood pressure. BMI was calculated as kg/m². NC was measured with participants standing, head in the horizontal plane; a tape was placed around the inferior margin of the laryngeal prominence, perpendicular to the neck's long axis. NC change (%) was computed as [(follow-up NC − baseline NC)/baseline NC] × 100. Biochemistry: Fasting plasma glucose (FPG), fasting insulin (FINS), lipid profile (TC, TG, HDL-C, LDL-C), C-reactive protein, liver enzymes, creatinine, and HbA1c were measured after an overnight fast by standard methods. Non-diabetic participants underwent a 75-g OGTT; those with diabetes underwent a 100-g steamed bread meal test; 2-hour plasma glucose was measured. Insulin resistance was estimated by HOMA-IR = FINS (mU/L) × FPG (mmol/L) / 22.5. Imaging: VFA and subcutaneous fat area (SFA) were quantified using a 3.0T MRI scanner (Archiva, Philips Medical Systems) at the L4–L5 intervertebral level in the supine position. Segmentation used Slice-O-Matic v4.2 (Tomovision). Abdominal obesity was defined as VFA ≥ 80 cm². Statistical analysis: SPSS 20.0 was used; two-sided p < 0.05 denoted significance. Normality was assessed. Means ± SD summarized normally distributed variables; medians (IQR) summarized skewed variables; counts (%) for categorical variables. Paired t-tests and Wilcoxon tests compared baseline vs follow-up as appropriate. NC change categories were < −2.5%, −2.5% to < 2.5% (reference, "maintainers"), ≥ 2.5% to < 5%, and ≥ 5%. Linear regression modeled associations of NC change categories with log-transformed VFA and SFA at follow-up; coefficients were back-transformed and reported as ratios (relative changes versus reference). Models were adjusted sequentially: Model 1 (age, sex), Model 2 (Model 1 + baseline SBP, DBP, HbA1c, HOMA-IR, TG, HDL-C, LDL-C), Model 3 (Model 2 + baseline BMI and WC). Logistic regression among participants without abdominal obesity at baseline (n = 683) assessed associations of NC change (%) with incident abdominal obesity at follow-up, adjusting for age, sex, BMI, WC, SBP, DBP, HbA1c, TG, HDL-C, and LDL-C. Subgroup analyses were stratified by sex, BMI (<25 vs ≥25 kg/m²), WC (lower vs higher: <90 cm men/<85 cm women vs ≥90 cm men/≥85 cm women), and age (<65 vs ≥65 years).

- Sample: 1421 adults (578 men, 843 women), age 24–80 years (mean 57.8 ± 7.1). Mean follow-up 2.1 years. - Anthropometry and labs: At follow-up, median VFA increased from 82.43 (58.27–113.10) to 84.55 (59.83–113.50) cm². Glycemic indices (FPG, 2hPG, HbA1c), FINS, HOMA-IR, TC, and HDL-C were higher at follow-up (p < 0.05). No significant changes in BMI, NC, WC, SFA, SBP, DBP, TG, LDL-C, CRP. - NC change categories: n = 406 (< −2.5%), 560 (−2.5% to <2.5%, reference), 329 (≥2.5% to <5%), 126 (≥5%). Median VFA at follow-up increased across NC change groups: 80.58, 88.50, 97.80, and 101.99 cm², respectively (all pairwise p < 0.05). SFA medians (168.69, 172.02, 170.66, 177.34 cm²) did not differ significantly across groups (all p > 0.05). - Linear regression (VFA ratio vs reference [−2.5% to <2.5% NC change]): • Model 1 (age, sex): < −2.5%: 0.91 (95% CI 0.86–0.97); ≥2.5% to <5%: 1.28 (1.08–1.54); ≥5%: 1.63 (1.24–1.82). • Model 2 (+ SBP, DBP, HbA1c, HOMA-IR, TG, HDL-C, LDL-C): < −2.5%: 0.94 (0.90–0.99); ≥2.5% to <5%: 1.26 (1.12–1.45); ≥5%: 1.42 (1.22–1.60). • Model 3 (+ BMI, WC): < −2.5%: 0.97 (0.94–1.03); ≥2.5% to <5%: 1.10 (1.03–1.25); ≥5%: 1.26 (1.05–1.49). - Linear regression (SFA ratio): After full adjustment (Model 3), no significant differences across NC change groups (ratios ~1.02, 1.03, 1.07; all p > 0.05). - Logistic regression in participants without abdominal obesity at baseline (n = 683): NC change (%) positively associated with incident abdominal obesity at follow-up (OR 1.23; 95% CI 1.09–1.39). A 5% NC increase corresponded to OR 1.36 (95% CI 1.19–1.68), indicating a 36% higher risk. - Subgroup analyses by sex, BMI, WC, and age showed consistent positive associations of NC change with abdominal obesity; no significant interactions were detected.

The study demonstrates that increases in neck circumference over approximately two years are independently associated with higher visceral fat area and greater risk of developing abdominal obesity, even after adjusting for traditional cardiometabolic risk factors and baseline adiposity (BMI and WC). This directly addresses the research question by providing longitudinal evidence that NC change reflects changes in visceral adiposity rather than subcutaneous fat, as SFA did not differ by NC change after adjustment. Given the impracticality of routine CT/MRI for VFA assessment, NC, a simple and reproducible measure, may serve as a practical surrogate marker to monitor visceral fat accumulation and stratify risk in clinical and community settings. The consistency across subgroups (sex, age, BMI, WC) suggests broad applicability within similar populations. These findings align with prior cross-sectional associations and extend them by showing temporal relationships between NC change and visceral adiposity and incident abdominal obesity.

In a Chinese community-based cohort, change in neck circumference was positively and independently associated with MRI-assessed visceral fat area and with incident abdominal obesity. NC measurement can serve as a simple, practical supplementary index for evaluating abdominal obesity in clinical practice. Future studies should validate these findings across multiple centers, diverse regions and ethnicities, and evaluate how NC-guided screening or interventions impact cardiometabolic outcomes. Incorporating imaging or other methods to account for factors that may affect NC measurement (e.g., cervical musculoskeletal changes) could further refine its utility.

Potential selection bias due to the single-center, community-based design in Shanghai limits generalizability to other regions and ethnic groups. Aging or musculoskeletal and other conditions may alter cervical anatomy or musculature and affect NC measurement accuracy. The study did not exclude or image-screen for all conditions that might influence NC, which could introduce measurement error. Follow-up duration was relatively short (~2 years), and residual confounding cannot be excluded despite adjustments.