Zeng Y, Li PY, Chen Z, Yao NF, Ye H, Gui ZH, Chen HL, Liu L, Wan H, Shen J. Associations of black coffee and black coffee supplemented with milk with diabetes in China: A community-based cross-sectional study. World J Diabetes 2026; 17(3): 115304 [DOI: 10.4239/wjd.v17.i3.115304]
Corresponding Author of This Article
Heng Wan, Associate Professor, Postdoc, Department of Endocrinology and Metabolism, The Eighth Affiliated Hospital of Southern Medical University, The First People’s Hospital of Shunde, No. 1 Jiazi Road, Lunjiao, Shunde District, Foshan 528308, Guangdong Province, China. wanhdr@smu.edu.cn
Research Domain of This Article
Endocrinology & Metabolism
Article-Type of This Article
Observational Study
Open-Access Policy of This Article
This article is an open-access article which was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Mar 15, 2026 (publication date) through Mar 15, 2026
Times Cited of This Article
Times Cited (0)
Journal Information of This Article
Publication Name
World Journal of Diabetes
ISSN
1948-9358
Publisher of This Article
Baishideng Publishing Group Inc, 7041 Koll Center Parkway, Suite 160, Pleasanton, CA 94566, USA
Share the Article
Zeng Y, Li PY, Chen Z, Yao NF, Ye H, Gui ZH, Chen HL, Liu L, Wan H, Shen J. Associations of black coffee and black coffee supplemented with milk with diabetes in China: A community-based cross-sectional study. World J Diabetes 2026; 17(3): 115304 [DOI: 10.4239/wjd.v17.i3.115304]
Yi Zeng, Nan-Fang Yao, Hua-Lan Chen, Jie Shen, School of Nursing, Southern Medical University, Guangzhou 510515, Guangdong Province, China
Yi Zeng, Pei-Yi Li, Zhi Chen, Nan-Fang Yao, Hong Ye, Zi-Hao Gui, Hua-Lan Chen, Lan Liu, Heng Wan, Jie Shen, Department of Endocrinology and Metabolism, The Eighth Affiliated Hospital of Southern Medical University, The First People’s Hospital of Shunde, Foshan 528308, Guangdong Province, China
Pei-Yi Li, Department of Nutrition, The Eighth Affiliated Hospital of Southern Medical University, The First People’s Hospital of Shunde, Foshan 528308, Guangdong Province, China
Pei-Yi Li, Zhi Chen, Lan Liu, Heng Wan, Jie Shen, Guangdong Engineering Technology Research Center of Metabolic Disorders Interdisciplinary Precision Prevention and Digital Healthcare, The Eighth Affiliated Hospital of Southern Medical University, The First People’s Hospital of Shunde, Foshan 528308, Guangdong Province, China
Author contributions: Zeng Y and Li PY conducted the data analyses and drafted the manuscript; Chi Z, Yao NF, Ye H, Gui ZH, and Chen HL conducted the data acquisition; Wan H, Li L, and Shen J performed the conceptualization and revised the manuscript; All authors approved the final version to publish. Zeng Y and Li PY contributed equally as first authors. Shen J and Wan H made equal contributions as co-corresponding authors.
Supported by Foshan Self-Funded Science and Technology Innovation Projects, No. 2420001004610.
Institutional review board statement: This study was approved by the Ethics Committee of the Shunde Hospital of Southern Medical University (No. 20211103.)
Informed consent statement: All study participants provided informed written consent prior to study enrollment.
Conflict-of-interest statement: The authors have no conflicts of interest to declare.
STROBE statement: The authors have read the STROBE Statement-checklist of items, and the manuscript was prepared and revised according to the STROBE Statement-checklist of items.
Data sharing statement: The data supporting the study findings are available from the corresponding authors upon reasonable request.
Corresponding author: Heng Wan, Associate Professor, Postdoc, Department of Endocrinology and Metabolism, The Eighth Affiliated Hospital of Southern Medical University, The First People’s Hospital of Shunde, No. 1 Jiazi Road, Lunjiao, Shunde District, Foshan 528308, Guangdong Province, China. wanhdr@smu.edu.cn
Received: October 15, 2025 Revised: November 24, 2025 Accepted: February 4, 2026 Published online: March 15, 2026 Processing time: 149 Days and 0.5 Hours
Abstract
BACKGROUND
Coffee consumption exhibits significant geographical variation between coffee and diabetes between China and Western countries. Furthermore, existing research on the relationship between coffee consumption and diabetes yields inconsistent results, with limited studies focusing on the Chinese population.
AIM
To investigate the association among coffee consumption, diabetes, and glucose indices among Chinese adults.
METHODS
A cross-sectional survey, involving 6876 individuals from the communities, was conducted in Guangdong, China. Participants provided detailed records of their coffee consumption, including any additives such as milk, sugar, or artificial sweeteners. Diabetic indicators measured included glycated hemoglobin (HbA1c), fasting plasma glucose, and 2-hour postprandial glucose (2h-PG). Statistical analyses were performed using multivariable logistic regression, adjusting for three categories of covariates: Demographic and anthropometric characteristics, lifestyle factors, and clinical and family history indicators.
RESULTS
In the adjusted model, coffee consumption was found to be inversely associated with the prevalence of diabetes, with an odds ratio (OR) of 0.79 and a 95% confidence interval (CI) of 0.64 to 0.98 (P = 0.029). Additionally, higher coffee consumption was inversely associated with elevated 2h-PG levels (OR = 0.83, 95%CI: 0.73-0.95; P = 0.005). Among individuals who consumed coffee with milk, the odds of elevated 2h-PG levels were reduced by 24% (OR = 0.76, 95%CI: 0.66-0.88; P < 0.001), and the odds of elevated HbA1c levels were reduced by 28% (OR = 0.72, 95%CI: 0.63-0.83; P < 0.001).
CONCLUSION
Coffee consumption, particularly the intake of coffee with milk, is inversely associated with the prevalence of diabetes, elevated 2h-PG, and increased HbA1c levels. The intake of coffee with milk is a factor of interest in the context of diabetes prevention or glycemic control; however, this observation necessitates further validation through additional randomized controlled trials.
Core Tip: This community-based cross-sectional study in Canton, China, enrolled 6876 participants to explore the association between coffee consumption (including milk-added coffee) and diabetes prevalence as well as glucose metabolism. With relatively few relevant studies in China compared to European and American countries, results showed coffee and milk-added coffee were negatively correlated with diabetes, elevated 2-hour postprandial blood glucose, and elevated glycated hemoglobin levels.
Citation: Zeng Y, Li PY, Chen Z, Yao NF, Ye H, Gui ZH, Chen HL, Liu L, Wan H, Shen J. Associations of black coffee and black coffee supplemented with milk with diabetes in China: A community-based cross-sectional study. World J Diabetes 2026; 17(3): 115304
Impaired glucose metabolism, the hallmark of diabetes, contributes to systemic complications and increased premature mortality[1,2]. Globally in 2022, diabetes affected 13.9% of women and 14.3% of men[3]. China’s diabetes-related expenditures are projected to rise to 460.4 billion dollars by 2030, nearly doubling from a decade ago[4]. Of great significance is the search for factors in diabetes prevention and intervention. The role of lifestyle changes such as dietary interventions is crucial in controlling blood sugar levels[5-7]. Coffee intake links to type 2 diabetes mellitus (T2DM) prevalence. Prior research has shown that higher intake levels correlate with a lower disease prevalence[8,9].
Globally, coffee stands as one of the most commonly consumed drinks[10,11]. Since 2012, global coffee consumption has been growing at an annual rate of 2%. It is worth noting that the average consumption rate in the Asian market stands at 5.2%, with the most significant growth observed in recent years[12]. The potential glucose-lowering effects of coffee may be attributed to its rich content of bioactive compounds, such as chlorogenic acids and caffeine[13-15]. These compounds have been shown to modulate glucose metabolism through antioxidant and anti-inflammatory pathways[16-18]. Specifically, it enhances hepatic fat oxidation[19]. Given the well-established link among hepatic steatosis, metabolic dysfunction, and T2DM, coffee exerts a regulatory role in maintaining metabolic homeostasis[20]. Additionally, coffee elevates core body temperature, a process that enhances glycemic control by activating brown adipose tissue to augment energy expenditure[21,22]. Prior research demonstrates that milk fortification of coffee boosts its bioaccessibility[23]. Furthermore, consumption of milk-containing coffee, such as latte or cappuccino, has been associated with a lower prevalence of metabolic syndrome and high blood pressure, whereas espresso consumption showed no such association[24]. Conversely, numerous studies indicate that the intake of beverages containing sugar or artificial sweeteners is positively associated with increased diabetes prevalence[25,26].
Notably, coffee consumption exhibits notable regional disparities, and a substantial disparity in coffee intake exists between China and Western countries. However, studies examining the association between coffee (including milk-added varieties) and diabetes remain scarce, and most of these have been conducted in Western countries. Thus, further investigations are warranted to explore the link among coffee intake, coffee type, diabetes, and glucose metabolism from multiple perspectives.
The present cross-sectional study examined the link between coffee intake (including black coffee and milk-added coffee) and diabetes, as well as elevated glycated hemoglobin (HbA1c), fasting plasma glucose (FPG), and 2-hour postprandial glucose (2h-PG) in Chinese community populations.
MATERIALS AND METHODS
Study design and subjects
Between November 2021 and September 2022, a cross-sectional analysis of baseline data from a community-based cohort called the “Shunde Metabolic Diseases and Risk Factors Epidemiological Study” was conducted in China. Stratified multistage random sampling was adopted to secure representativeness of the study sample[27]. All study participants submitted provided written informed consent before enrollment in the study. First, 3412 participants were excluded for missing blood test data. Furthermore, participants with incomplete coffee consumption data or those who consumed coffee containing added sugar or artificial sweeteners were excluded from the analysis (n = 2350), as were those with a previous diagnosis of diabetes mellitus (n = 602). Ultimately, 6876 participants were included in the study per the above inclusion and exclusion criteria (Figure 1).
Figure 1 Study design.
SEPPD-Shunde: Shunde Metabolic Diseases and Risk Factors Epidemiological Study.
Data collection
Structured questionnaires were completed by trained medical personnel, collecting information on demographic characteristics (including age, educational attainment, and sex), lifestyle factors (including smoking behavior and work-related physical activity), as well as medication use and medical histories. Subsequently, a physical examination was conducted to measure weight and height; body mass index (BMI) was computed as BMI = weight (kg)/height2 (m²)[28].
A single venous blood sample (approximately 30 mL) was drawn from each participant after a 10-hour overnight fast, during the morning hours (7:00-10:00 AM). The anticoagulant-treated vacuum tube samples were promptly transferred to a College of American Pathologists-certified icebox and processed within 2 hours of collection, undergoing centrifugation, aliquoting, and storage at -20 °C. Additionally, if participants self-reported a diabetes diagnosis during the survey, further tests, including HbA1c and FPG, were conducted. Plasma glucose was measured 2 hours after 75-g oral glucose tolerance test for those who did not have a clear diagnosis of diabetes[29].
HbA1c was detected by high performance liquid chromatography (instrument model HLC-723G8, Tosoh Corporation, Japan). For clinical biochemical testing, venous blood samples were collected after subjects had fasted for at least 10 hours. Samples were immediately centrifuged and sent to a central laboratory certified by external quality assessment. An automatic biochemical analyzer (AU5831; Beckman Coulter, Inc., Brea, CA, United States) was used to detect FPG, 2h-PG, total cholesterol (TC), triglycerides (TGs), the concentration of high-density lipoprotein cholesterol (HDL-C), and the concentration of low-density lipoprotein cholesterol (LDL-C).
According to the 2024 American Diabetes Association guidelines, an individual was diagnosed with diabetes if they met at least one of the following criteria: FPG ≥ 7.0 mmol/L, 2h-PG ≥ 11.1 mmol/L, HbA1c ≥ 6.5%, or self-reported use of antidiabetic medications. Accordingly, elevated FPG was defined as ≥ 5.6 mmol/L, elevated 2h-PG as ≥ 7.8 mmol/L, and elevated HbA1c as ≥ 5.7%[30]. Hypertension was defined in accordance with the 2020 International Society of Hypertension guidelines as systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, a prior diagnosis of hypertension, or current use of antihypertensive drugs[31]. Dyslipidemia was defined as any of the following: TC ≥ 6.22 mmol/L, TG ≥ 2.26 mmol/L, LDL-C ≥ 4.14 mmol/L, HDL-C < 1.04 mmol/L, physician-diagnosed dyslipidemia, or current lipid-lowering agent use[27].
Educational attainment, a key demographic characteristic, was stratified into three categories: Below high school, high school diploma or equivalent, and post-high school education. Age was categorized into two groups: < 60 years and ≥ 60 years[32,33]. BMI was categorized into two strata: < 24.0 kg/m2 and ≥ 24.0 kg/m2[29]. Smoking status was categorized into three groups: Non-smokers, former smokers (individuals who previously smoked and had abstained for more than 6 months), and current smokers (individuals who currently smoke and have consumed a total of over 100 cigarettes)[34]. Alcohol consumption was categorized into two groups: Drinkers and non-drinkers. Occupational physical activity was assessed using a closed-ended questionnaire consisting of four predefined categories: Sedentary (primarily seated), moderate (repetitive movements while seated or standing, typical of manufacturing occupations), vigorous (primarily standing or walking, common to construction and agricultural work), and heavy (involving lifting, loading, or moving heavy objects)[35]. A family history of diabetes in first-degree relatives was obtained via self-report using a family health history questionnaire and categorized into two binary groups: Present and absent.
Coffee consumption
Basic information on coffee consumption was collected by trained medical personnel. A validated 24-hour dietary recall questionnaire was administered in this study to ascertain coffee intake volume and the types of additives in coffee[36]. Participants were required to report their 24-hour coffee intake volume and specify on the questionnaire whether any additives (e.g., milk, sugar, artificial sweeteners) had been added to their coffee. One serving of coffee was defined as 150 mL. For the analysis, non-coffee drinkers were classified as individuals who reported consuming less than one coffee serving. Considering the generally low coffee consumption among the Chinese population, the classification criteria were simplified, and participants were stratified into two groups: Non-consumers and daily consumers (≥ 1 serving). Coffee consumers additionally reported whether milk was added to their coffee. Per this classification, participants were stratified into three groups: Non-consumers, exclusive black coffee drinkers, and milk-added coffee drinkers[37].
Statistical analyses
This cross-sectional observation and analysis showed that the total data loss rate of all variables in our model was 4.29%. Considering that there were few missing data (≤ 5%), and in order to avoid the possible deviation caused by estimation method, we directly analyzed the complete data. Only those participants who had all of the information were included in the final analysis. Baseline comparisons were performed between the pre-exclusion group (including participants with missing coffee intake data, missing covariate data, or those consuming coffee containing added sugar) and the complete case group after exclusion (Supplementary Table 1). No significant differences were observed across key variables, indicating minimal selection bias in the present study.
A study was conducted, and a method called binary logistic regression was used. The aim was to see the difference between people who drank and did not drink coffee in terms of the prevalence of diabetes and their blood sugar levels. Lastly, the odds ratio (OR) and 95% confidence interval (CI) were estimated to quantify the strength of this association. Two models were built to adjust different things. Model 1 included age, sex and BMI; in model 2, in addition to those in model 1, education level, smoking or not, exercise or not, drinking or not, hypertension, TC, TG, LDL-C, HDL-C, and whether there was diabetes in the family were added, which are variables used for examination. This survey also studied the relationship between coffee drinking habits, diabetes and glucose metabolism.
Additional exploratory stratified analyses were also conducted, which were classified according to age divided into two groups: < 60 years and ≥ 60 years, BMI (less than 24 and over 24), sex (female, male), education level (primary school, middle school, university), physical activity (no exercise, low-intensity exercise, moderate-intensity exercise, high-intensity exercise) and hypertension (with or without). A more careful analysis was made by excluding people with chronic diseases (hypertension and hyperlipidemia). For the sensitivity analysis, the same statistical models (model 1 and model 2) as the primary analysis were employed, with the only modification being the exclusion of the “hypertension” variable from the covariate list. P < 0.05 was defined as a significant statistical difference. All analyses were conducted using R statistical software (version 4.4.3).
RESULTS
Characteristics of participants
Of the 6876 participants included in the final analysis, 25.00% reported daily coffee consumption. Among these daily coffee consumers, 11.94% were diagnosed with diabetes and 86.65% were non-smokers; 62.65% of all included participants were female. Baseline characteristics of coffee drinkers and non-drinkers are summarized in Table 1. Between-group comparisons indicated that the coffee consumer group had a lower proportion of smokers, a lower proportion of individuals with less than high school education, and a lower proportion of participants diagnosed with hypertension.
Table 1 Characteristics of study participants not drink coffee and drink at least one cup of coffee per day, n (%)/median (interquartile range).
Coffee consumption amount, prevalence of diabetes, and indicators of blood glucose regulation
ORs for diabetes and glucose metabolism parameters, adjusted for coffee consumption status, are presented in Table 2. Model 2 analysis revealed that coffee consumption was negatively associated with diabetes (OR = 0.79, 95%CI: 0.64-0.98; P = 0.029), elevated 2h-PG (OR = 0.83, 95%CI: 0.73-0.95; P = 0.005), and elevated HbA1c (OR = 0.76, 95%CI: 0.67-0.86; P < 0.001).
Table 2 Relationship between coffee consumption and diabetes as well as glucose metabolism parameters.
The results of the correlation analysis for different coffee types are summarized in Table 3. In model 2, we found that compared with participants who did not drink coffee, those who drank coffee with milk had a 24% lower prevalence of high 2h-PG (OR = 0.76, 95%CI: 0.66-0.88; P < 0.001), a 28% lower prevalence of high HbA1c levels (OR = 0.72, 95%CI: 0.63-0.83; P < 0.001) and a 25% reduction in the prevalence of diabetes (OR = 0.75, 95%CI: 0.59-0.95; P = 0.017).
In all the groups, we did not find any connection between drinking coffee and diabetes (Table 4), including age, BMI, sex, educational attainment, and physical activity, and hypertension (interaction P > 0.05).
Sensitivity analysis was conducted by excluding hypertensive and hyperlipidemic participants (Table 5). After excluding participants with hypertension (n = 1708) and hyperlipidemia (n = 1544), the findings remained substantially consistent with those of the primary analysis. Our analysis revealed an inverse association between coffee consumption and elevated high HbA1c levels (OR = 0.76, 95%CI: 0.64-0.91; P = 0.002), diabetes (OR = 0.68, 95%CI: 0.47-0.99; P = 0.045), and elevated 2h-PG levels (OR = 0.83, 95%CI: 0.70-0.99; P = 0.035).
In this community-based cross-sectional study among Chinese adults, we observed a significant inverse association between coffee consumption and the prevalence of diabetes, as well as levels of 2h-PG and HbA1c. Notably, these inverse associations remained robust for milk-added coffee. To the best of our knowledge, this is one of the few studies to specifically examine the relationship of coffee additives with metabolic outcomes within a Chinese population, where coffee consumption patterns are rapidly evolving.
In contrast to classifications based on caffeine content, this study pragmatically categorized coffee as black or milk-added, reflecting the low consumption of decaffeinated coffee in the Chinese population. Notably, milk-added coffee remained inversely associated with diabetes risk in this cohort. While this observational association does not establish causality, it suggests that interactions between coffee components and dairy constituents may modulate bioactivity related to glucose homeostasis. Mechanistically, coffee polyphenols such as caffeic acid and chlorogenic acid have been reported to form stable complexes with milk proteins via covalent bonds[37,38]. These interactions are hypothesized to influence the pharmacokinetics or bioefficacy of polyphenols, offering a potential explanation for the observed association. Consequently, further experimental studies are required to validate this interaction and clarify its physiological relevance to glucose metabolism in vivo.
The strength of our analysis lies in the robustness of the observed associations. Even after adjusting for major lifestyle confounders such as physical activity, smoking, and alcohol consumption, the inverse relationship between coffee intake and diabetes remained significant. This provides a robust epidemiological “bridge”, lending biological plausibility to a potential link. However, because we compared non-consumers vs daily consumers, we could not evaluate a dose-response relationship across finer consumption gradients. Future research with quantitatively assessed consumption is warranted to examine this important aspect of causality and further strengthen causal inference.
Given the relatively low coffee intake in the Chinese population, we used a pragmatic categorization (non-consumers vs daily consumers) rather than multiple cup-based categories commonly applied in Western cohorts. Despite this simplified exposure definition, the inverse association between coffee consumption and diabetes remained evident in the fully adjusted models. Our results are consistent with findings from several large-scale prospective cohorts in Western populations[39,40]. For instance, a major study involving three American cohorts reported an 11% lower risk of diabetes among coffee consumers, and a meta-analysis of 30 studies suggested a dose-response relationship, with each additional cup per day associated with a 6% risk reduction[41,42]. Although the overall level of coffee consumption in our Chinese participants was lower than that typically reported in European or American studies, the persistence of an inverse association suggests that any coffee consumption is associated with a more favorable metabolic profile in this demographic.
Several limitations warrant consideration. First, the cross-sectional design precludes us from establishing a temporal sequence or causal relationship. Second, dietary data were based on self-reported 24-hour recalls, which may introduce recall bias; however, this method is a validated tool for assessing population-level dietary exposures. Third, although we controlled for major confounders, residual confounding from unmeasured factors such as specific genetic predispositions cannot be entirely ruled out. Despite these limitations, the use of multiple standardized glycemic markers (FPG, 2h-PG, and HbA1c) and a large community-based sample strengthens the internal validity of our findings.
CONCLUSION
This cross-sectional study revealed an inverse association between coffee consumption and diabetes risk. After adjustment for demographic, lifestyle, and clinical covariates, consumption of coffee with milk remained significantly associated with a reduced prevalence of diabetes, as well as lower 2h-PG and lower HbA1c levels. Although this suggests a potential benefit of milk coffee in glycemic regulation, the observational design precludes causal inference. Further randomized controlled trials are needed to establish causality and elucidate the underlying mechanisms.
ACKNOWLEDGEMENTS
The authors thank all team members and participants in the study.
Jurema Soares M, de Souza Figueira M, Rodrigues Sampaio G, Aparecida Manólio Soares-Freitas R, Clara da Costa Pinaffi-Langley A, Aparecida Ferraz da Silva Torres E. Coffee simulated inhibition of pancreatic lipase and antioxidant activities: Effect of milk and decaffeination.Food Res Int. 2022;160:111730.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 5][Reference Citation Analysis (0)]
Poojary MM, Hellwig M, Henle T, Lund MN. Covalent bonding between polyphenols and proteins: Synthesis of caffeic acid-cysteine and chlorogenic acid-cysteine adducts and their quantification in dairy beverages.Food Chem. 2023;403:134406.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 31][Reference Citation Analysis (0)]
Scientific quality: Grade A, Grade B, Grade B, Grade B, Grade C, Grade C, Grade D
Novelty: Grade A, Grade A, Grade B, Grade C, Grade D
Creativity or innovation: Grade A, Grade A, Grade B, Grade C, Grade D
Scientific significance: Grade A, Grade B, Grade B, Grade B, Grade E
P-Reviewer: Basu R, MD, Postdoctoral Fellow, India; Guo R, Associate Professor, China; Pappachan JM, MD, FRCP, MRCP, Professor, Senior Researcher, United Kingdom; Tung TH, PhD, Associate Professor, Taiwan; Yang WJ, Researcher, China S-Editor: Wu S L-Editor: Filipodia P-Editor: Lei YY
