Nutritional intake characteristics and glycemic control in patients with type 2 diabetes mellitus

Abstract

BACKGROUND

Optimal nutritional status is fundamental to the effective management of type 2 diabetes mellitus (T2DM); nonetheless, many patients encounter difficulties in maintaining a balanced diet that aligns with clinical recommendations. Imbalances in energy, macronutrient, and micronutrient intake remain prevalent and may contribute to suboptimal glycemic control.

AIM

To investigate the nutritional intake profile of patients with T2DM, compare dietary and clinical characteristics according to glycemic control status, and identify clinical and demographic factors independently associated with poor glycemic control.

METHODS

This single-center cross-sectional study enrolled 125 patients with T2DM, whose dietary intake was assessed using a 24-hour dietary recall. A three-tier analytical approach was implemented. First, nutrient intake was quantified using standardized food composition tables and compared against the recommended nutrient intakes (RNIs) established by the Chinese Nutrition Society. Patients were then stratified into well-controlled (HbA1c ≤ 7.0%) and poorly controlled (HbA1c > 7.0%) groups based on glycated hemoglobin (HbA1c) levels. Multivariable logistic regression models were constructed to evaluate the associations of total energy intake and carbohydrate energy percentage with poor glycemic control, with adjustment for demographic and clinical covariates, including antidiabetic treatment regimen. A secondary model additionally adjusted for total cholesterol.

RESULTS

Overall nutritional intake was characterized by total energy, protein, and carbohydrate consumption below recommended levels, alongside a relatively high fat intake. The majority of micronutrients—including vitamins A, D, E, B₁, B₂, and C, as well as essential minerals such as calcium, magnesium, zinc, and selenium—were significantly below the RNIs, whereas sodium and iron intakes were excessive. Dietary fiber intake was less than half of the recommended level. In multivariable analysis, each 100-kcal/day increment in total energy intake was independently associated with a greater odds of poor glycemic control after adjustment for age, sex, body mass index (BMI), diabetes duration, educational level, and antidiabetic treatment regimen [odds ratio (OR) = 1.536, 95% confidence interval (CI): 1.176-2.005, P = 0.002]. This association remained significant after further adjustment for total cholesterol (OR = 1.491, 95%CI: 1.134-1.960, P = 0.004). In contrast, carbohydrate energy percentage was not independently associated with poor glycemic control in either adjusted model (P = 0.069 and P = 0.081, respectively). Higher BMI was also associated with greater odds of poor glycemic control, whereas educational level and antidiabetic treatment regimen showed significant associations with glycemic status. Total cholesterol was not independently associated with poor glycemic control in the fully adjusted model (P = 0.110).

CONCLUSION

Patients with T2DM in this cohort exhibited substantial nutritional imbalances, characterized by inadequate dietary fiber and micronutrient intake, along with a relatively high proportion of energy derived from fat. Multivariable analysis further identified higher BMI as a factor associated with greater odds of poor glycemic control, while educational level and antidiabetic treatment regimen were significantly associated with glycemic status. These findings underscore the importance of individualized nutritional assessment integrated within comprehensive clinical management for patients with T2DM.

Key Words: Type 2 diabetes mellitus; Nutritional status; Glycated hemoglobin; Dietary intake; Correlation

Core Tip: This study revealed substantial dietary imbalances among patients with type 2 diabetes mellitus. In unadjusted analyses, patients with poor glycemic control exhibited higher total energy intake and greater absolute carbohydrate-derived energy intake. After adjustment for demographic and clinical covariates—including antidiabetic treatment regimen—higher total energy intake remained independently associated with increased odds of poor glycemic control, whereas the percentage of energy derived from carbohydrates was not independently associated. This association with total energy intake persisted after further adjustment for total cholesterol, indicating robustness to additional metabolic confounding. Higher body mass index was also independently associated with poor glycemic control, while educational level and antidiabetic treatment regimen showed significant overall associations with glycemic status. Collectively, these findings underscore total energy intake as a clinically relevant nutritional factor within the broader context of comprehensive diabetes management, distinct from carbohydrate composition alone.

INTRODUCTION

Type 2 diabetes mellitus (T2DM), the most prevalent form of diabetes mellitus, accounts for approximately 90% of all diabetes cases[1]. In China, the prevalence of T2DM continues to rise, with an increasingly younger age of onset attributable to multiple contributing factors[2]. This trend poses a serious threat to public health and underscores the urgent need for improved prevention and management strategies. Although no definitive cure for T2DM currently exists, adherence to a scientifically balanced diet can effectively alleviate pancreatic burden and help stabilize blood glucose levels. Optimal glycemic control and the prevention of macrovascular and microvascular complications require a combination of pharmacological therapy and healthy lifestyle modifications, including dietary management and regular physical activity. Nevertheless, most diabetic patients in China exhibit poor adherence to dietary recommendations, which impedes effective metabolic control[3]. In particular, patients with T2DM often display excessive total caloric intake and imbalanced macronutrient distribution, characterized by insufficient carbohydrate and protein consumption alongside excessive fat intake—deviations that fall outside the ranges recommended by the Chinese Guidelines for Medical Nutrition Therapy in Diabetes[4].

Blood glucose level serves as a critical indicator of disease severity in diabetic patients, and glycated hemoglobin (HbA1c) is widely regarded as the "gold standard" for assessing glycemic control, as it reflects average blood glucose levels over the preceding 2-3 months[5]. Dietary education has been shown to enhance patients' understanding of diabetic nutrition and contribute to improved glycemic outcomes. Although nutritional status assessment is fundamental to effective dietary therapy, the rate of achieving target blood glucose levels among Chinese diabetic patients remains suboptimal, largely due to substantial deficiencies in dietary nutritional status[6]. To address this gap, the present single-center, cross-sectional study aimed to characterize the intake of energy, macronutrients, micronutrients, and food groups among patients with T2DM, and to compare these dietary profiles according to glycemic control status. A secondary objective was to identify clinical, metabolic, and demographic factors independently associated with poor glycemic control, after adjustment for relevant confounding variables.

MATERIALS AND METHODS

Study participants

This single-center, cross-sectional observational study employed a convenience sampling method to recruit patients with T2DM at our hospital between January 2024 and June 2025. Eligible participants met the following inclusion criteria: (1) Diagnosis of T2DM according to World Health Organization criteria, defined as fasting plasma glucose ≥ 7.0 mmol/L, 2-hour oral glucose tolerance test ≥ 11.1 mmol/L, or HbA1c ≥ 6.5%; (2) Age ≥ 18 years and disease duration ≥ 3 months; and (3) Normal mental and cognitive function with adequate reading and communication abilities. A total of 165 consecutive outpatients were initially screened. After applying predefined eligibility criteria, 40 patients were excluded for the following reasons: (1) Type 1 diabetes or other specific types of diabetes (n = 5); (2) Current insulin therapy (n = 13); (3) Pregnancy or postpartum status (n = 2); (4) Severe diabetic complications (n = 12); and (5) Severe cardiac, hepatic, or renal dysfunction (n = 8). The final analytical sample therefore comprised 125 patients. The study was approved by the Ethics Committee of Tongren Hospital, Shanghai Jiao Tong University School of Medicine (Approval No. K2024-003-01). Written informed consent was waived by the committee, as the analysis was based on anonymized data.

Data collection

General information: A structured questionnaire was administered to collect data on sex (male/female), age, body mass index (BMI), marital status (unmarried/married/divorced or widowed), educational level (primary school or below/junior or senior high school/college or above), and smoking history (current/former/never). Fasting blood glucose, HbA1c, and other relevant biochemical parameters were measured using standardized phlebotomy and laboratory procedures. Venous blood samples were drawn on the morning following the dietary recall assessment, after an overnight fast of 8-12 hours. HbA1c was determined by high-performance liquid chromatography using a TOSOH HLC-723G8 fully automated HbA1c analyzer.

Clinical and medication variables: Diabetes duration was defined as the number of years since the initial clinical diagnosis. Regular use of oral antidiabetic drugs (OADs) was defined as continuous OAD therapy for at least 3 months prior to enrollment. Information on specific OAD classes—including metformin, sodium-glucose cotransporter-2 inhibitors, dipeptidyl peptidase-4 inhibitors, and sulfonylureas—was extracted from electronic health records. Patients were categorized into three treatment groups: Dietary management alone (without OADs), OAD monotherapy, or combination OAD therapy. Dosage information was not consistently available and was therefore not included in the analysis. Patients receiving any form of exogenous insulin were excluded a priori.

Dietary intake records: Dietary intake was assessed using the 24-hour dietary recall method. Participants were instructed to record all foods and beverages consumed during the preceding 24 hours, excluding atypical occasions such as banquets or family gatherings. Records included meal times, food types, cooking methods, and portion sizes. Nutrient and energy intakes were calculated using standardized food composition tables. Daily cooking oil and salt consumption per individual was estimated based on total household usage, using the formula: Individual daily intake = (total household daily consumption × proportion of food consumed by the individual at that meal)/(total number of family members sharing the meal × a correction coefficient). Macronutrient intake is expressed both as absolute energy contribution (kcal/day) and as a percentage of total daily energy derived from carbohydrates, protein, and fat. Micronutrient and dietary fiber intakes, reported as absolute daily amounts (mg/day, μg/day, or g/day), were evaluated against the dietary reference intakes.

Survey methods and evaluation criteria

The questionnaire survey was conducted by trained physicians from the research team. All investigators received standardized training on the study objectives, survey procedures, questionnaire content, and completion requirements prior to data collection. All questionnaires were collected on-site. Dietary and nutrient intake adequacy was assessed against the Chinese Dietary Guidelines, the Chinese Food Guide Pagoda, the Chinese Guidelines for Medical Nutrition Therapy in Diabetes, and the recommended nutrient intakes (RNIs) established by the Chinese Nutrition Society. Participants were divided into well-controlled (HbA1c ≤ 7.0%) and poorly controlled (HbA1c > 7.0%) groups, with the 7.0% cutoff aligned with the Chinese Guidelines for the Prevention and Treatment of Type 2 Diabetes and the American Diabetes Association Standards of Care[7]. Demographic, clinical, and dietary characteristics were compared between the two groups, and factors associated with poor glycemic control were further evaluated.

Derivation of nutrient comparators and hypothesis testing

Individualized daily energy requirements were estimated based on age, sex, ideal body weight, and physical activity level, in accordance with the Chinese dietary reference intakes. For micronutrients and dietary fiber, the RNI reference values were derived from the fixed cutoffs recommended in the dietary reference intakes guidelines for Chinese adults with T2DM or for the general population of the same age group.

Dietary assessment and nutrient calculation

Daily dietary intake was assessed by trained interviewers using the 24-hour dietary recall method. Dietary records were matched and processed against the Chinese Food Composition Table (6^th Edition, Volumes 1 and 2; Peking University Medical Press), which serves as the authoritative national reference for nutrient quantification. To reduce recall-related measurement error, standardized food portion photographs, measuring cups, and three-dimensional food models were used to quantify portion sizes. The weights of raw ingredients in composite dishes and home-cooked recipes were calculated using validated local raw-to-cooked conversion factors. Household cooking oil and salt consumption was estimated using standardized measuring containers. Nutritional composition and daily energy intake were calculated using the Mint dietary analysis software, which is based on the aforementioned food composition tables. Nutrient values reflect intake from food sources only; commercial dietary supplements (e.g., multivitamins, calcium, or omega-3 capsules) were recorded qualitatively at baseline but were not included in the nutrient intake calculations.

Dietary assessment protocol and quality control

Face-to-face interviews were conducted by trained physicians under the direct supervision of senior clinical dietitians. All interviewers underwent standardized training prior to participant enrollment to ensure consistency in dietary interviewing and portion-size estimation. Dietary records were reviewed for completeness and plausibility before inclusion in the analysis; incomplete records were excluded. Caloric inconsistency was predefined as daily energy intake < 500 or > 3500 kcal/day for women and < 600 or > 4000 kcal/day for men. No participants in the final analysis cohort met these exclusion criteria.

Model specification and stability considerations

Given the fixed sample size of this cross-sectional study, a parsimonious approach was adopted for the multivariate logistic regression model to minimize the risk of overfitting. Covariates were selected based on clinical relevance and potential confounding effects, rather than solely on statistical significance in univariate analyses. Given that glycemic control regimens may influence blood glucose levels, particular attention was paid to adjusting for this variable. The number of model parameters was kept limited relative to the number of outcome events, and model fit was assessed using the Hosmer-Lemeshow goodness-of-fit test. To evaluate the robustness of the observed associations, a secondary model was constructed by further adjusting for total cholesterol levels in the primary model.

Statistical analysis

All statistical analyses were performed using SPSS version 25.0 (IBM Corp., Armonk, NY, USA). Categorical variables are presented as frequencies (percentages), and continuous variables as mean ± SD. Between-group comparisons were conducted using independent-samples t-tests for continuous variables and χ² tests or Fisher’s exact tests, as appropriate, for categorical variables. Paired-samples t-tests were used to compare observed nutrient intakes with individualized reference intake values. To control for the increased risk of type I errors due to multiple univariate testing, the Benjamini-Hochberg false discovery rate (FDR) correction was applied. Relationships between variables were examined using univariate linear regression. The goodness-of-fit of the multivariable logistic model was assessed using the Hosmer-Lemeshow test, with P > 0.05 indicating adequate model fit. A two-sided P < 0.05 was considered statistically significant.

RESULTS

Macronutrient intake in the overall cohort

Based on the Dietary Reference Intakes for Chinese Residents (2023 Edition), individualized RNIs were calculated according to age, sex, and physical activity level. In the overall patient cohort, mean daily intakes of total energy, carbohydrates, and protein were significantly lower than the corresponding RNIs, whereas fat intake was significantly higher than the recommended level (all P < 0.05; Table 1).

Table 1 Comparison of macronutrient intake with recommended nutrient intakes in all patients, mean ± SD.

Nutrient	Actual intake	Percentage of total calories (%)	Recommended nutrient intake	Percentage of total calories (%)	t value	P value	FDR-adjusted q value
Energy (kcal/day)	1998.07 ± 308.09		2135.60 ± 396.40	60	2.964	0.004	0.004
Carbohydrates (g/day)	280.40 ± 66.00	55.48 ± 6.32	320.34 ± 59.46	60	4.831	< 0.0001	< 0.0002
Protein (g/day)	62.62 ± 10.21	12.80 ± 2.74	80.09 ± 14.86	15	10.680	< 0.0001	< 0.0002
Fat (g/day)	69.55 ± 11.13	31.73 ± 5.25	59.32 ± 11.01	25	7.367	< 0.0001	< 0.0002

False discovery rate (FDR) adjustment was performed using the Benjamini-Hochberg procedure, with q < 0.05 denoting FDR-adjusted statistical significance. FDR: False discovery rate.

Vitamin intake in the overall cohort

Daily intakes of vitamins A, D, E, B₁, B₂, and C were all significantly below the RNI thresholds (P < 0.05 for each comparison; Table 2).

Table 2 Comparison of intake of vitamins with recommended nutrient intakes in all patients, mean ± SD.

Nutrient	Actual intake	Recommended nutrient intake	t value	P value	FDR-adjusted q value
Vitamin A (μg RAE/day)	389.14 ± 130.82	704.80 ± 49.62	24.24	< 0.0001	< 0.0001
Vitamin D (μg/day)	7.70 ± 1.00	10.88 ± 1.91	16.31	< 0.0001	< 0.0001
Vitamin E (mg α-TE/day)	9.81 ± 3.28	14	33.42	< 0.0001	< 0.0001
Vitamin B1 (mg/day)	0.94 ± 0.20	1.30 ± 0.10	18.38	< 0.0001	< 0.0001
Vitamin B2 (mg/day)	0.70 ± 0.27	1.20	29.05	< 0.0001	< 0.0001
Vitamin C (mg/day)	63.24 ± 17.75	100	39.83	< 0.0001	< 0.0001

False discovery rate (FDR) adjustment was performed using the Benjamini-Hochberg procedure, with q < 0.05 denoting FDR-adjusted statistical significance. RAE: Retinol activity equivalents; α-TE: Alpha-tocopherol equivalents; FDR: False discovery rate.

Mineral and dietary fiber intake in the overall cohort

Daily intakes of calcium, magnesium, zinc, selenium, and dietary fiber were significantly lower than the RNIs, whereas intakes of iron and sodium were significantly higher (P < 0.05 for all). These differences remained statistically significant after false discovery rate (FDR) adjustment (all q < 0.05). No significant difference was observed for daily phosphorus intake (P > 0.05; Table 3).

Table 3 Comparison of mineral and dietary fiber intake with recommended nutrient intakes in all patients, mean ± SD.

Nutrient	Actual intake	Recommended nutrient intake	t value	P value	FDR-adjusted q value
Ca (mg/day)	510.06 ± 173.37	800	32.89	< 0.0001	< 0.0001
P (mg/day)	687.02 ± 228.63	704.96 ± 11.68	0.886	0.377	0.377
Mg (mg/day)	159.70 ± 37.25	318.48 ± 4.22	46.5	< 0.0001	< 0.0002
Fe (mg/day)	15.36 ± 3.52	13.23 ± 3.09	4.785	< 0.0001	< 0.0002
Zn (mg/day)	9.63 ± 2.27	10.24 ± 1.76	2.337	0.021	0.024
Se (μg/day)	48.71 ± 10.39	60	52.44	< 0.0001	< 0.0002
Na (mg/day)	1867.94 ± 61.91	1482.40 ± 38.24	58.46	< 0.0001	< 0.0002
Dietary fiber (g/day)	10.44 ± 4.65	25	25.08	< 0.0001	< 0.0002

False discovery rate (FDR) adjustment was performed using the Benjamini-Hochberg procedure, with q < 0.05 denoting FDR-adjusted statistical significance. Ca: Calcium; P: Phosphorus; Mg: Magnesium; Zn: Zinc; Se: Selenium; Fe: Iron; Na: Sodium; FDR: False discovery rate.

Energy and macronutrient intake by glycemic control status

Patients with poor glycemic control had significantly higher total energy intake and absolute carbohydrate-derived energy intake than those with well-controlled glycemia (P < 0.05). The percentage of total energy derived from carbohydrates was also significantly higher in the poorly controlled group, whereas the percentages derived from protein and fat were significantly lower (P < 0.05). All these differences remained significant after FDR adjustment (q < 0.05). No significant between-group differences were observed in absolute energy intakes from protein or fat (P > 0.05; Table 4).

Table 4 Comparison of energy and macronutrient intake in patients with different blood glucose levels, mean ± SD.

	Carbohydrates		Protein		Fat		Total calories (kcal/day)
	Energy contribution (kcal/day)	Percentage (%)	Energy contribution (kcal/day)	Percentage (%)	Energy contribution (kcal/day)	Percentage (%)
Well-controlled group (n = 44)	917.92 ± 240.47	51.10 ± 7.04	244.53 ± 39.83	14.10 ± 3.08	609.03 ± 97.70	34.81 ± 5.81	1771.48 ± 268.11
Poorly controlled group (n = 81)	1232.24 ± 204.27	57.86 ± 4.36	253.72 ± 41.24	12.09 ± 2.26	635.19 ± 100.84	30.05 ± 4.07	2121.15 ± 254.77
t value	7.713	6.621	1.205	4.161	1.400	5.343	7.195
P value	< 0.0001	< 0.0001	0.231	< 0.0001	0.164	< 0.0001	< 0.0001
FDR-adjusted q value	< 0.0002	< 0.0002	0.231	< 0.0002	0.191	< 0.0002	< 0.0002

False discovery rate (FDR) adjustment was performed using the Benjamini-Hochberg procedure, with q < 0.05 denoting FDR-adjusted statistical significance. FDR: False discovery rate.

Dietary intakes in patients with different blood glucose levels

In unadjusted comparisons, patients with well-controlled glycemia reported significantly higher intakes of vegetables and fish/shrimp, and significantly lower intakes of fruits, meat, soybeans and nuts, cooking oils, and table salt, compared with those with poor glycemic control (P < 0.05 for all; Table 5).

Table 5 Comparison of dietary intakes in patients with different blood glucose levels (g/day), mean ± SD.

	Cereals and tubers	Vegetables	Fruits	Meat	Eggs	Fish/shrimps	Soybeans and nuts	Dairy	Cooking oil	Table salt
Well-controlled group (n = 44)	317.50 ± 97.54	304.34 ± 86.77	18.91 ± 9.26	80.36 ± 38.86	39.14 ± 12.10	72.50 ± 40.19	62.23 ± 23.04	70.84 ± 41.60	30.20 ± 10.28	5.25 ± 2.43
Poorly controlled group (n = 81)	304.05 ± 85.85	265.80 ± 69.77	32.57 ± 14.82	109.68 ± 62.09	37.36 ± 11.32	50.80 ± 24.66	71.65 ± 21.91	73.58 ± 21.45	40.84 ± 11.63	6.30 ± 2.63
t value	0.797	2.702	5.547	2.843	0.819	3.722	2.256	0.486	5.082	2.181
P value	0.427	0.008	< 0.0001	0.005	0.415	0.0003	0.026	0.628	< 0.0001	0.031
FDR-adjusted q value	0.474	0.016	< 0.0005	0.013	0.474	0.001	0.043	0.628	< 0.0005	0.004

All P values in this table represent exploratory univariate comparisons. False discovery rate (FDR) adjustment was performed using the Benjamini-Hochberg procedure, with q < 0.05 denoting FDR-adjusted statistical significance. FDR: False discovery rate.

Demographic and clinical data by glycemic control status

Significant between-group differences were observed in body mass index (BMI), educational level, antidiabetic treatment regimen, and total cholesterol level (P < 0.05 for all; Table 6).

Table 6 Comparison of demographic and clinical data, mean ± SD.

	Well-controlled group (n = 44)	Poorly controlled group (n = 81)	t value	P value
Age (years)	52.36 ± 15.21	53.05 ± 14.84	0.683	0.496
Gender			0.559	0.455
Male	24	38
Female	20	43
BMI (kg/m2)	22.51 ± 3.90	24.44 ± 3.82	2.681	0.008
Duration of diabetes mellitus (years)	7.71 ± 4.43	8.08 ± 5.71	0.373	0.710
Marital status			0.952	0.621
Unmarried	5	12
Married	31	50
Divorced or widowed	8	19
Educational level			12.042	0.002
Elementary school and below	6	35
Junior and senior high school	14	21
High school or above	24	25
Clinical treatment regiment			9.687	0.008
Dietary control alone without OADs	15	51
OAD monotherapy	20	22
Combination OAD therapy	9	8
OAD class
Metformin	24	21	10.141	0.002
SGLT2 inhibitors	6	5		0.193
DPP-4 inhibitors	5	6		0.516
Sulfonylurea	3	6		0.999
Smoking history			0.286	0.593
Present	19	31
Absent	25	50
Total cholesterol (mmol/L)	3.91 ± 0.91	4.60 ± 1.16	3.367	0.001
Triglycerides (mmol/L)	1.64 ± 0.96	2.27 ± 2.53	1.594	0.114
Low density lipoprotein	2.38 ± 0.50	2.54 ± 0.66	1.427	0.156
High density lipoprotein	1.15 ± 0.27	1.08 ± 0.37	1.102	0.272

Oral antidiabetic drug classes were not mutually exclusive because patients receiving combination therapy could be included in more than one medication class. Fisher’s exact test was used when expected cell counts were < 5. SGLT2: Sodium-glucose cotransporter-2; DPP-4: Dipeptidyl peptidase-4; BMI: Body mass index; OAD: Oral antidiabetic drug.

Factors influencing HbA1c in patients with T2DM

Multivariable binary logistic regression analysis using the Enter method was performed to identify factors independently associated with poor glycemic control (defined as HbA1c > 7.0%; Table 7). Total energy intake and carbohydrate energy percentage were included as the primary dietary exposures of interest. Age, sex, BMI, diabetes duration, educational level, and antidiabetic treatment regimen were entered as clinically relevant covariates.

Table 7 Multivariate analysis of factors influencing glycated hemoglobin in patients type 2 diabetes mellitus.

Variable	Value	Model 1			Model 2
		OR	95%CI	P value	OR	95%CI	P value
Total energy intake	Continuous variable, per 100 kcal/day	1.536	1.176-2.005	0.002	1.491	1.134-1.960	0.004
Carbohydrate energy percentage	Continuous variable, %	1.110	0.992-1.241	0.069	1.112	0.987-1.253	0.081
Age	Continuous variable	1.012	0.963-1.065	0.629	1.008	0.957-1.061	0.771
Gender	0 = female, 1 = male	0.410	0.126-1.336	0.139	0.433	0.132-1.419	0.167
Duration of diabetes mellitus	Continuous variable	1.030	0.925-1.146	0.592	1.034	0.929-1.152	0.536
BMI	Continuous variable	1.170	1.020-1.343	0.025	1.193	1.031-1.380	0.018
Educational level	0 = elementary school and below, 1 = junior and senior high school, 2 = high school or above
Elementary school and below				0.009			0.029
Junior and senior high school		0.133	0.025-0.700	0.017	0.154	0.029-0.824	0.029
High school or above		0.092	0.020-0.428	0.002	0.127	0.027-0.605	0.010
Clinical treatment regiment	0 = dietary control alone, 1 = OAD monotherapy, 2 = combination OAD therapy
Dietary control alone				0.029			0.047
OAD monotherapy		0.271	0.078-0.943	0.040	0.311	0.089-1.094	0.069
Combination OAD therapy		0.129	0.023-0.732	0.021	0.144	0.026-0.807	0.028
Total cholesterol	Continuous variable				1.540	0.906-2.618	0.110

Model 1: Hosmer-Lemeshow P = 0.223, Nagelkerke R² = 0.619. Model 2: Hosmer-Lemeshow P = 0.189, Nagelkerke R² = 0.635. BMI: Body mass index; OAD: Oral antidiabetic drug; OR: Odds ratio; CI: Confidence interval.

In Model 1 (adjusted for the above-mentioned covariates), higher total energy intake was independently associated with increased odds of poor glycemic control [odds ratio (OR) = 1.536, 95% confidence interval (CI): 1.176-2.005, P = 0.002]. Carbohydrate energy percentage was not independently associated with poor glycemic control after adjustment (OR = 1.110, 95%CI: 0.992-1.241, P = 0.069). Higher BMI was also significantly associated with greater odds of poor glycemic control (OR = 1.170, 95%CI: 1.020-1.343, P = 0.025). Educational level and antidiabetic treatment regimen showed significant overall associations with glycemic control (P = 0.009 and P = 0.029, respectively).

In Model 2, which additionally adjusted for total cholesterol, the association between total energy intake and poor glycemic control remained materially unchanged (OR = 1.491, 95%CI: 1.134-1.960, P = 0.004). Carbohydrate energy percentage remained non-significant (OR = 1.112, 95%CI: 0.987-1.253, P = 0.081), whereas BMI remained significantly associated with higher odds of poor glycemic control (OR = 1.193, 95%CI: 1.031-1.380, P = 0.018). Educational level and antidiabetic treatment regimen retained their overall significance (P = 0.029 and P = 0.047, respectively). Total cholesterol was not independently associated with poor glycemic control in the fully adjusted model (OR = 1.540, 95%CI: 0.906-2.618, P = 0.110). Both models demonstrated acceptable calibration, as indicated by the Hosmer-Lemeshow goodness-of-fit test (Model 1: P = 0.223; Model 2: P = 0.189).

DISCUSSION

T2DM is a chronic metabolic disorder closely intertwined with dietary habits, necessitating long-term, often lifelong, management[8]. Dietary therapy remains a cornerstone of diabetes care; in patients with mild T2DM, adequate glycemic control may even be achieved through dietary modification alone. Sustained adherence to a scientifically balanced dietary regimen can effectively slow disease progression and is crucial for preventing complications and enhancing patients' quality of life[9]. Nevertheless, many patients with T2DM exhibit imbalanced dietary intakes that are associated with suboptimal glycemic control and may complicate long-term disease management. Current nutritional guidelines for diabetes management advocate for increased carbohydrate and dietary fiber intake alongside reduced fat consumption, within the context of appropriate caloric control[10].

In the present study, we observed a distinct nutritional imbalance characterized by a high proportion of energy derived from fat and pronounced deficiencies in dietary fiber and micronutrients, even when total caloric intake fell below the RNI. These findings suggest that nutritional management should extend beyond simple energy restriction to encompass overall dietary quality and macronutrient distribution. Patients with T2DM often experience nutritional inadequacies, with energy intake substantially below recommended levels. Given that carbohydrates represent a major dietary energy source, grains—predominantly refined staples such as rice and white flour—along with sugars remain the primary contributors to energy intake[11]. Owing to insufficient knowledge about diabetic dietary management, many patients may excessively reduce or eliminate staple foods, potentially contributing to inadequate energy intake. Notably, these observations differ from the dietary patterns frequently reported in Western populations with T2DM, in which excessive caloric intake from ultra-processed foods has been implicated in poor metabolic control[12]. However, our findings align with the "double malnutrition" phenomenon described in East Asian populations, wherein patients may exhibit inadequate intake of certain nutrients alongside an imbalanced macronutrient profile[13]. This underscores that clinical dietary guidance must go beyond simple calorie restriction and emphasize appropriate energy intake, balanced macronutrient distribution, overall dietary quality, and fat reduction. Moreover, the insufficient intake of vitamins, minerals, and dietary fiber observed in our cohort may reflect restrictive or unbalanced food choices made after diabetes diagnosis, rather than being attributable solely to inadequate energy intake[14].

We further stratified patients by glycemic control status based on HbA1c levels. In unadjusted comparisons—with FDR correction applied—patients with well-controlled glycemia reported higher intakes of fresh vegetables and fish/shrimp, and lower intakes of fruits, meat, soybeans and nuts, cooking oils, and table salt, compared with those in the poorly controlled group. Although the well-controlled group exhibited higher vegetable and fish consumption, dietary fiber intake in the overall cohort remained significantly below recommended levels. A recent international meta-analysis emphasized that achieving fiber intake targets depends not only on quantity but also on dietary diversity[15,16]. Increased consumption of dairy products, poultry, eggs, and seafood is also recommended to improve the proportion of high-quality protein. Fruit intake should be individualized, taking into account portion size, carbohydrate content, and overall dietary composition[17]. Furthermore, patients should avoid excessive dietary fat, cooking oil, and salt, while maintaining adequate vegetable intake and overall dietary balance. It is important to note that these food-group differences were derived from unadjusted comparisons and should not be interpreted as independent associations with glycemic control. Nevertheless, they identify potentially relevant dietary characteristics that may inform nutritional assessment and patient education, while their independent relationships with glycemic control warrant further investigation.

We next analyzed factors associated with poor glycemic control. After adjustment for age, sex, BMI, diabetes duration, educational level, and antidiabetic treatment regimen, higher total energy intake remained independently associated with greater odds of poor glycemic control. Each 100-kcal/day increase in total energy intake was associated with a 53.6% higher odds of poor glycemic control (OR = 1.536, 95%CI: 1.176-2.005). This association persisted after further adjustment for total cholesterol (OR = 1.491, 95%CI: 1.134-1.960), indicating that the relationship between total energy intake and glycemic control was robust to additional metabolic adjustment. In contrast, carbohydrate energy percentage was not independently associated with poor glycemic control in the multivariable-adjusted models. Higher BMI was also significantly associated with greater odds of poor glycemic control, whereas higher educational level and specific antidiabetic treatment regimens were significantly associated with better glycemic control. In a case-control study involving 662 patients with diabetes, analyses examining the relationship between BMI and poor glycemic control demonstrated that the proportion of patients with a BMI ≥ 25 kg/m² was significantly higher among those with poor glycemic control than among those with BMI < 25 kg/m²[18,19]. The association between higher BMI and poor glycemic control observed in our study is consistent with the established relationship between excess adiposity, insulin resistance, and impaired glucose metabolism. Importantly, total energy intake remained significantly associated with poor glycemic control even after adjustment for BMI, suggesting that the observed dietary association was not fully explained by differences in body mass.

Educational level may influence patients' knowledge of diabetes dietary management, which in turn may be associated with glycemic levels. Mastery of diabetes-related dietary knowledge is closely related to educational attainment; therefore, enhanced efforts should be directed toward strengthening dietary education for patients with lower educational levels to support adherence to appropriate dietary management. Unlike highly standardized diabetes education systems, our findings suggest that patients with lower educational levels may face greater difficulty translating clinical guidelines into everyday dietary practices. In the adjusted models, antidiabetic treatment regimens were also significantly associated with glycemic control, highlighting the importance of accounting for medication use when evaluating the relationship between dietary intake and HbA1c. Notably, after adjustment for treatment regimens, the association between total energy intake and poor glycemic control persisted; while this adjustment could not eliminate the association entirely, it reduced the likelihood of confounding due to medication. Although total cholesterol was associated with glycemic control in univariate analyses, it was no longer independently associated with poor glycemic control after dietary variables and antidiabetic treatment regimens were included in the fully adjusted model (OR = 1.540, 95%CI: 0.906-2.618, P = 0.110). This attenuation suggests that the previously observed association may have partly reflected shared metabolic, dietary, or treatment-related characteristics rather than an independent association.

This study has several limitations that should be acknowledged. First, owing to its single-center, cross-sectional design and the use of non-probability sampling, causal inferences cannot be drawn, and selection bias cannot be excluded. Second, dietary intake was assessed using a single 24-hour dietary recall, which represents a significant methodological limitation. Although this method provides immediate data on current intake, it may not capture day-to-day dietary variation or long-term nutritional habits. Furthermore, reliance on patient recall may introduce recall bias. To mitigate these effects, standardized food portion aids were used during interviews; however, future studies incorporating repeated recalls on non-consecutive days or food frequency questionnaires[20] are needed to more accurately reflect long-term nutritional status and its association with long-term glycemic control. Third, the exclusion of patients on insulin therapy and those with severe end-organ complications limits the generalizability of our findings to broader or more severely affected T2DM populations. Fourth, although antidiabetic treatment regimens were included in the multivariate models and patients were classified by individual OAD class, information on drug dosages was not available. Moreover, because the number of patients receiving certain OAD classes was relatively small and there was overlap between drug classes among those on combination therapy, detailed adjustment for individual drug classes was not feasible. Therefore, residual confounding related to medication cannot be ruled out. Fifth, the relatively small sample size limited the number of covariates that could be included in the multivariate models and may have reduced the precision of certain estimates, particularly for treatment subgroups. Consequently, a prospective, large-sample, multicenter study incorporating repeated dietary assessments and more detailed medication data is warranted to confirm the associations between the observed dietary intake patterns and glycemic control.

CONCLUSION

In summary, patients with T2DM in this cohort exhibited substantial imbalances in dietary quality, characterized by pronounced dietary fiber deficiency, an excessively high proportion of energy derived from fat, widespread micronutrient inadequacies, and total caloric intake falling below the RNI. After adjustment for demographic and clinical factors, including antidiabetic treatment regimen, higher total energy intake was independently associated with greater odds of poor glycemic control (HbA1c > 7.0%), whereas carbohydrate energy percentage was not independently associated with poor glycemic control. The association with total energy intake remained significant after further adjustment for total cholesterol. Higher BMI was also associated with greater odds of poor glycemic control, while educational level and antidiabetic treatment regimen showed significant associations with glycemic control. These findings provide real-world evidence regarding nutritional intake characteristics and their relationship with glycemic control in patients with T2DM. Collectively, these results suggest that, within the framework of personalized nutrition education and comprehensive clinical management, dietary interventions for patients with T2DM should address overall energy intake and dietary quality, rather than focusing exclusively on carbohydrate restriction.