Insufficient knowledge and moderate attitudes and practices regarding continuous glucose monitoring among patients with diabetes Authors: Feng Zhang, Zhenyuan Tang, Yating Ma, Yunle Zhang, Pingping Liang, Lianfeng Yang, Li Wang Published: 2026 ABSTRACT AbstractContinuous glucose monitoring (CGM) has shown increasing application in diabetes management. This study aimed to explore the knowledge, attitudes, and practices (KAP) of patients with various types of diabetes towards CGM. A cross-sectional study was conducted in Runda Community Health Service Center, Suzhou, between October 23, 2023, and November 27, 2023. A self-administered questionnaire was developed to collect demographic information and KAP. Structural equation modeling (SEM) was employed to analyze the relationships between the KAP dimensions. A total of 576 participants were included, with a mean age of 66.37 ± 7.51 years. The mean KAP scores were 5.55 ± 5.33 (possible range: 0–22), 31.41 ± 4.22 (possible range: 10–50), and 12.85 ± 3.83 (possible range: 5–25), respectively. Positive correlations were observed between knowledge and attitude (r = 0.514, P < 0.001), knowledge and practice (r = 0.486, P < 0.001), and attitude and practice (r = 0.339, P < 0.001). SEM based on standardized coefficients showed that knowledge had a significant direct effect on attitude (β = 0.553, 95% CI 0.484–0.623, P < 0.001) and practice (β = 0.439, 95% CI 0.347–0.532, P < 0.001). In addition, knowledge exerted an indirect effect on practice through attitude (β = 0.160, 95% CI 0.098–0.222, P < 0.001). This study revealed insufficient knowledge, moderate attitude, and practice toward CGM among patients with diabetes in Suzhou, China. Educational and behavioral interventions are recommended, particularly for diabetes patients with advanced age, lower educational levels, and not having education and promotion on dynamic blood glucose monitors.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-57244-4.Keywords: Health knowledge, Attitude, Practice, Diabetes mellitus, Continuous glucose monitoring, Cross-sectional studies, Structural equation modelingSubject terms: Diseases, Endocrinology Introduction Diabetes is a chronic metabolic disorder characterized by elevated blood glucose resulting from inadequate insulin production or impaired insulin action1. The prevalence of diabetes has been steadily increasing, emerging as a major public health concern. In 2019, 463 million adults aged 20–79 years were living with diabetes worldwide, with this number projected to rise to 700 million by 20452. Approximately 116.4 million adults in China were living with diabetes in 2019, which represents the largest diabetic population worldwide3. Diabetes adversely affects various organ systems, leading to complications such as cardiovascular disease4,5, neuropathy6, nephropathy7, and retinopathy8. Furthermore, the management of diabetes and its complications imposes a substantial economic burden on healthcare systems and individuals. Glucose control is required in diabetes management, and there has been a growing emphasis on continuous glucose monitoring (CGM) in the management of both type 1 and type 2 diabetes mellitus, as well as other types of diabetes9. Unlike traditional self-monitoring of blood glucose, CGM offers continuous glucose trends and fluctuations10. The device is particularly beneficial for patients with significant glucose variability, hypoglycemia unawareness, or difficulty achieving glycemic targets11. Moreover, CGM facilitates data-driven decision-making regarding insulin dosing, diet, and physical activity12. Despite its potential to improve diabetes management, most hospitals in China prescribe CGM for less than 10 individuals per month, suggestive of its limited adoption13. The main challenges identified include limited patient knowledge, high costs and lack of insurance coverage, and imperfect CGM management systems13,14. Since understanding of CGM’s principles, its perceived importance, and its integration into daily life can significantly influence its adoption, exploring patients’ knowledge, attitudes, and practices (KAP) is needed to identify barriers in public CGM use. KAP studies assess individuals’ understanding, attitudes, and behaviors regarding a health-related issue, and are frequently used in public health research to inform intervention strategies. Previous KAP evidence has been available regarding glucose control among diabetes patients. For example, a study in Benin revealed that while most patients understood the effects of exercise and diet on glucose control, they lacked knowledge of relevant medical indicators15. Conversely, a study in Ethiopia reported good KAP among diabetes patients16. Despite the growing adoption of CGM worldwide, KAP evidence on CGM remains sparse, particularly among Chinese diabetes patients. However, evidence on patients’ KAP toward CGM in China remains limited in community and primary care settings, where diabetes management is largely delivered and where access to CGM education and device use may differ from tertiary hospitals. Therefore, this community-based study in Suzhou aimed to assess KAP toward CGM among adults with diabetes (type 1, type 2, and other types) and to identify factors associated with KAP. In addition, we used structural equation modeling to quantify the pathways linking knowledge, attitudes, and practices, providing insight into potential targets for interventions in community health services. We hypothesized that knowledge positively influences attitudes and practices toward CGM, and that attitudes are positively associated with practices. Methods Study design and participants This cross-sectional study included patients with diabetes at the Runda Community Health Service Center in Suzhou between October 23, 2023, and November 27, 2023. Eligible participants were invited to participate in the study by trained research staff during their routine visits. Ethical standards were adhered to at each stage of the study, consistent with the principles laid out in the 1964 Declaration of Helsinki and its later amendments. Ethical considerations included obtaining informed consent from all participants before data collection, ensuring that participants were fully aware of the purpose of the study, their voluntary involvement, and their right to withdraw at any stage without consequence. Moreover, participant confidentiality was maintained throughout the study, and data was stored securely in accordance with local and international data protection regulations. The study was carried out after the protocol was approved by the Ethics Committee of the Runda Community Health Service Center. The inclusion criteria were the following: (1) Patients diagnosed with type 1, type 2, or other forms of diabetes17, according to clinical diagnosis by trained physicians or self-reported use of antidiabetic medications; (2) Age over 18 years; (3) Normal level of consciousness and ability to independently answer the questionnaire. The exclusion criteria were: (1) Patients with other severe chronic diseases such as cancer or severe liver disease; (2) Patients with severe complications of diabetes such as serious heart disease, kidney disease, or eye disease; (3) Pregnant or lactating women; (4) Logical errors or irrational submission of response. A total of 642 questionnaires were collected. After excluding 30 questionnaires due to excessively short completion times (under 90 s), 1 questionnaire with abnormal age response, 3 questionnaires with abnormal height responses, 1 questionnaire with abnormal weight response, and 31 questionnaires with logical errors, a total of 576 valid questionnaires were retained. Questionnaire design Initially, we reviewed the literature for existing validated CGM-related KAP questionnaires. However, no instrument was identified that was validated for a Chinese community-based population and simultaneously aligned with the guideline-based CGM content and the KAP framework applied in this study. Therefore, we developed and refined a context-specific questionnaire based on national guidelines and expert consultation. The design of the questionnaire was based on the guidelines Guideline for the prevention and treatment of type 2 diabetes mellitus in China (2020 edition)18,19 and Clinical application guidelines for blood glucose monitoring in China (2022 edition)20. After the initial design, the questionnaire underwent modifications based on feedback from three experts (two endocrinology specialists and one general practice specialist). Following this, a pilot study was conducted among 36 participants, yielding a Cronbach’s α value of 0.831, suggesting good internal consistency. To further evaluate construct validity, confirmatory factor analysis (CFA) was performed for the three-domain KAP structure. The final questionnaire was in Chinese and consisted of four sections: (1) demographic information, including gender, age, residence, education, occupation, monthly income, height, body weight, diagnosis of hypertension, regular blood glucose monitoring methods, and Problem Areas in Diabetes-5 (PAID-5); (2) knowledge dimension (such as CGM functions, advantages, and potential challenges), (3) attitude dimension (such as perceptions of CGM’s usefulness, potential barriers to use, and general opinions on its effectiveness in managing diabetes), and (4) practice dimension (such as questions on the frequency of blood glucose monitoring, awareness of CGM availability). Body mass index (BMI) was calculated as BMI = weight (kg)/height (m)2, with classifications of < 18.5 kg/m2 as underweight, 18.5–23.9 kg/m2 as normal weight, 24–27.9 kg/m2 as overweight, and ≥ 28 kg/m2 as obesity21. The PAID-5 scale was used to measure diabetes distress and emotional disorders. It has a total of 5 items with 5 response options, with the score varying from 0 ("not a problem") to 4 (“serious problem”). The overall score, ranging up to 100, is calculated by summing these scores and multiplying by 5. Scores under 40 indicated no significant diabetes distress, while those 40 and above were categorized as significant diabetes distress. The primary outcome measure in this study is the scores of the Knowledge, Attitudes, and Practices dimension. The knowledge dimension comprised 11 items, with scores ranging from 0 to 22, where 2 points were assigned for “very familiar”, 1 point for “heard of”, and 0 points for “unclear”. The attitude dimension included 10 questions on a five-point Likert scale, ranging from “strongly agree” (5 points) to “strongly disagree” (1 point). The total score ranged from 10 to 50 points. The practice dimension comprised 7 questions. Items P6 did not exhibit a positive or negative attitude bias, and they were analyzed descriptively. The remaining questions also used a five-point Likert scale, from always (5 points) to never (1 point). The score range was 5–25 points. For the practice item assessing CGM use experience (P2: “I have used or am currently using a continuous glucose monitor”), responses were collected using the same 5-point format to maintain consistency across the practice section; higher endorsement was interpreted as a higher likelihood of any CGM use experience (current or previous). For descriptive reporting, P2 was summarized as the proportion of participants selecting affirmative responses, reflecting any CGM use experience. Sufficient knowledge, positive attitude, and proactive practice were defined as > 70% scores in each dimension22. The threshold value was chosen according to previous patient-centered research studies in China23,24. Sample size The calculation of sample size was as previous studies25,26. This study primarily aimed to examine associations among KAP domains and to construct a structural equation model (SEM). Given that no universally accepted closed-form sample size formula exists for SEM, sample size considerations were guided by methodological recommendations rather than formal precision-based calculations. Previous studies using Monte Carlo simulations have suggested that a sample size of approximately 200–400 is generally adequate for SEM analyses25,26, depending on model complexity. The final sample size in this study exceeded these recommendations, indicating that it was sufficient to support the planned analyses. Questionnaire distribution and quality control An electronic questionnaire was created using the Sojump website (https://www.wjx.cn/), and a QR code for the electronic questionnaire was generated. Questionnaires were distributed to study participants through scanning QR codes in clinic settings and WeChat groups. To ensure the quality and integrity of the questionnaire results, each IP address was limited to one submission, and all items were mandatory. Five research assistants from the center were trained to assist with participant recruitment, questionnaire administration, and data collection. The assistants played a crucial role in ensuring that the data collection was completed without any ambiguity. The survey was administered face-to-face to ensure that each participant’s responses were recorded accurately, and any questions or difficulties were addressed immediately by the research team. Quality control was mainly conducted through data cleaning in the backend. During the formal questionnaire distribution and data collection process, real-time data checking and field corrections were not conducted to avoid any influence on patients’ self-reports due to the research assistants’ attitudes. After all data were collected, the data were exported from the questionnaire platform and underwent unified data cleaning and quality control. Questionnaires with excessively short response times, obvious numerical errors, logical inconsistencies, or highly consistent answers across the KAP dimensions were excluded. Statistical analysis Continuous variables are presented as mean ± standard deviation (SD) and were compared using Student’s t-test or one-way ANOVA, as appropriate. Categorical variables are presented as n (%) and were compared using the chi-square test. Spearman’s correlation coefficients were used to assess interrelationships among Knowledge (K), Attitude (A), and Practice (P) scores. K, A, and P analyzed as separate continuous scale scores in all regression and SEM analyses. For descriptive reporting only, each KAP score was dichotomized using a cutoff of 70% of the total possible score27–30. To identify factors associated with each KAP scale, we performed univariable and multivariable linear regression analyses. Variables considered clinically relevant and/or associated at P < 0.05 in univariable analyses were entered into multivariable linear regression models; results are presented as unstandardized regression coefficients (β) with 95% confidence intervals (CIs) and P values. To evaluate hypothesized pathways among K, A, and P, we fitted a covariate-adjusted structural equation model (SEM) in which K, A, and P were specified as continuous endogenous variables. The SEM tested the hypotheses that: (H1) K is associated with A and P; (H2) A is associated with P; and (H3) K is indirectly associated with P through A. To reduce confounding and align with the regression analyses, key participant characteristics were included as exogenous covariates with paths to K, A, and P (binary predictors were dummy coded; multicategory predictors were represented using indicator variables or collapsed where appropriate for model stability). Model fit was evaluated using the Root Mean Square Error of Approximation (RMSEA), Standardized Root Mean Square Residual (SRMR), Tucker–Lewis Index (TLI), and Comparative Fit Index (CFI). Consistent with common recommendations, acceptable model fit was defined as RMSEA < 0.08 (preferably < 0.06), SRMR < 0.08, and CFI/TLI > 0.90 (preferably > 0.95) (references). SEM was conducted using Stata version 18.0 (StataCorp LLC, College Station, TX, USA). All other statistical analyses were performed using R software. A two-sided P < 0.05 was considered statistically significant. Results Baseline information A total of 642 questionnaires were initially collected. Following data cleaning, questionnaires with completion time < 90 s (n = 30), implausible responses (n = 5), and logically inconsistent responses (n = 31) were excluded. A final sample of 576 valid questionnaires was included in the analysis. The mean age of participants was 66.37 ± 7.51 years, with 57.47% being female. Most participants resided in urban areas (89.58%) and worked in non-healthcare-related industries (97.05%). The majority engaged in regular blood glucose monitoring (79.51%), had lifestyle interventions (80.56%), and had not received education or promotion on CGM (53.99%). Based on the PAID-5 results, 94.62% of the participants reported experiencing no significant diabetes-related distress in terms of emotional and behavioral challenges (Table 1). Table 1Demographic information of participants and KAP score.N (%)KnowledgeAttitudePracticeScorePScorePScorePTotal576 (100%)5.55 ± 5.3331.41 ± 4.2212.85 ± 3.83Gender0.7530.6670.986 Male245 (42.53)5.72 ± 5.4131.62 ± 4.1712.96 ± 4.01 Female331 (57.47)5.42 ± 5.2731.26 ± 4.2512.76 ± 3.69Age (years)66.37 ± 7.51 < 0.001 < 0.001 < 0.001 < 6069(11.98)8.13 ± 6.1033.08 ± 4.2113.79 ± 4.32 60–69327(56.77)5.76 ± 5.1831.60 ± 4.1613.16 ± 0.65 ≥ 70180(31.25)4.16 ± 4.8330.42 ± 4.0911.9 ± 0.77Residence0.008 < 0.001 < 0.001 Rural60 (10.42)3.93 ± 5.0329.20 ± 4.8511.20 ± 4.98 Urban516 (89.58)5.73 ± 5.3331.67 ± 4.0713.04 ± 3.63Education < 0.001 < 0.001 < 0.001 Junior high school or below277 (48.09)3.77 ± 4.2230.12 ± 4.0912.04 ± 3.52 High school /technical school212 (36.81)6.36 ± 5.0132.06 ± 3.6813.29 ± 3.75 College and above87 (15.10)9.23 ± 6.7433.95 ± 4.3914.34 ± 4.33Occupation < 0.001 < 0.001 < 0.001 Healthcare-related industry17 (2.95)16.18 ± 7.7234.76 ± 3.6017.71 ± 4.41 Non-healthcare-related industry559 (97.05)5.22 ± 4.9031.31 ± 4.2012.70 ± 3.71Monthly Income (CNY) < 0.001 < 0.001 < 0.001 < 200095 (16.49)3.59 ± 3.8230.48 ± 3.8611.81 ± 3.65 2000–5000291 (50.52)4.95 ± 4.7130.86 ± 4.0812.46 ± 3.56 > 5000190 (32.99)7.45 ± 6.2432.73 ± 4.3113.96 ± 4.07BMI (kg m−2)0.2840.7040.019 < 23.9296 (51.39)5.73 ± 5.2431.66 ± 4.2913.21 ± 3.55 24 ≤ BMI < 28224 (38.89)5.47 ± 5.4231.21 ± 4.1012.55 ± 4.18 ≥ 2856 (9.72)4.88 ± 5.4330.96 ± 4.2612.13 ± 3.62Hypertension0.8120.9010.056 Yes402 (69.79)5.57 ± 5.4731.40 ± 4.1712.66 ± 3.88 No174 (30.21)5.50 ± 4.9931.44 ± 4.3313.28 ± 3.69Regular blood glucose monitoring < 0.001 < 0.001 < 0.001 Yes458 (79.51)6.29 ± 5.5331.81 ± 4.1513.68 ± 3.53 No118 (20.49)2.65 ± 3.0429.88 ± 4.159.63 ± 3.17Lifestyle intervention0.9760.0220.368 Yes464 (80.56)5.58 ± 5.3731.58 ± 4.2412.92 ± 3.78 No112 (19.44)5.42 ± 5.1530.72 ± 4.0712.54 ± 4.04Oral hypoglycemic agents0.0590.978 < 0.001 Yes445 (77.26)5.78 ± 5.3831.45 ± 4.1413.18 ± 3.79 No131 (22.74)4.76 ± 5.0731.27 ± 4.5011.72 ± 3.74Insulin injection < 0.0010.571 < 0.001 Yes109 (18.92)7.61 ± 6.5231.75 ± 4.6514.26 ± 3.78 No467 (81.08)5.07 ± 4.8931.33 ± 4.1112.52 ± 3.77Having education and promotion on CGM < 0.001 < 0.001 < 0.001 Yes265 (46.01)9.19 ± 5.2333.55 ± 3.8414.45 ± 3.82 No311 (53.99)2.45 ± 2.8929.59 ± 3.6411.48 ± 3.27PAID-5 < 0.0010.415 < 0.001 Significant diabetes distress31 (5.38)8.97 ± 6.1731.94 ± 2.7115.10 ± 3.61 No significant diabetes distress545 (94.62)5.35 ± 5.2131.38 ± 4.2912.72 ± 3.80Quantitative variables are represented as M ± SD, and categorical variables are represented as N (%). Abbreviations: CNY: Chinese Yuan; BMI: body mass index; PAID-5: the Problem Areas in Diabetes Scale 5. Demographic information of participants and KAP score. Quantitative variables are represented as M ± SD, and categorical variables are represented as N (%). Abbreviations: CNY: Chinese Yuan; BMI: body mass index; PAID-5: the Problem Areas in Diabetes Scale 5. KAP comparison The mean KAP scores were 5.55 ± 5.33 (possible range: 0–22), 31.41 ± 4.22 (possible range: 10–50), and 12.85 ± 3.83 (possible range: 5–25), respectively. Significant differences in knowledge scores were noted across variables such as age, residence, education level, CGM education and promotion, and PAID-5 score (all P < 0.05). Similarly, attitude scores varied significantly by age, residence, education level, and CGM promotion (all P < 0.05). Practice scores also demonstrated significant differences based on age, residence, education level, CGM promotion, and PAID-5 score (all P < 0.05) (Table 1). Knowledge distribution Accuracy rates in the knowledge dimension (encompassing “Familiar” and “Aware” responses) ranged from 26.21% to 83.16%. Among all the items, the item with the largest proportion choosing “Unclear” was K10 (73.78% unclear; CGM is appropriate for patients with type 1 diabetes, type 2 diabetes treated with insulin injections, gestational diabetes, and those experiencing frequent unexplained high or low blood sugar levels) and K11 (71.53% unclear; CGM measures glucose levels in body fluids, not blood, and cannot fully replace traditional blood glucose meters) (Table S1). Attitude distribution In the attitudes dimension, positivity rates varied from 8.50% to 75.70%. The majority (75.70%) acknowledged the importance of regular blood glucose monitoring for adjusting treatment plans and controlling blood sugar (A1). And 66.14% strongly agree or agree that the effectiveness of CGM varies from person to person and may not be suitable for all diabetes patients (A7) (Table S2). Practice distribution Practice adherence rates ranged from 7.99% to 65.45%. Only 7.99% agreed/strongly agreed that they had used or were currently using CGM (P2), indicating a very low prevalence of any CGM use experience in this sample. Additionally, merely 12.50% expressed willingness to recommend CGM to other diabetes patients (P5) (Table S3). A majority of participants obtained knowledge about new hypoglycemic drugs or related information through medical education by hospitals or medical staff (91.49%) and recommendations from friends and relatives (65.10%) (Table S3, Figure S1). Correlations of KAP Spearman’s correlation analysis revealed positive correlations between knowledge and attitude (r = 0.514, P < 0.001), knowledge and practice (r = 0.486, P < 0.001), and attitude and practice (r = 0.339, P < 0.001) (Table 2). Table 2Correlation analysis of KAP scores.KnowledgeAttitudePracticeKnowledge1.000Attitude0.514 (P < 0.001)1.000Practice0.486 (P < 0.001)0.339 (P < 0.001)1.000 Correlation analysis of KAP scores. Questionnaire validation CFA supported the three-factor KAP measurement model with acceptable fit indices (RMSEA = 0.074, SRMR = 0.077, TLI = 0.884, CFI = 0.899) (Figure S2). Factors associated with K, A and P scores Table 3, 4 and 5 summarizes univariable and multivariable linear regression analyses of factors associated with K, A, and P scores. In multivariable models, higher knowledge scores were associated with younger age (β = − 0.056, 95% CI − 0.101 to − 0.010), higher education (college or above vs junior high or below: β = 2.822, 95% CI 1.630 to 4.015), regular blood glucose monitoring (β = 0.976, 95% CI 0.109 to 1.843), insulin injection (β = 1.271, 95% CI 0.448 to 2.094), prior CGM education/promotion (no vs yes: β = − 5.775, 95% CI − 6.455 to − 5.096), and the presence of significant diabetes distress (PAID-5; β = 3.935, 95% CI 2.536 to 5.334). For attitude, higher knowledge was independently associated with higher attitude scores (β = 0.254, 95% CI 0.182 to 0.325). Urban residence (β = 1.638, 95% CI 0.658 to 2.619) and higher education (college or above: β = 1.509, 95% CI 0.419 to 2.598) were positively associated with attitude, whereas lack of CGM education/promotion was negatively associated (β = − 1.892, 95% CI − 2.631 to − 1.154). For practice, both knowledge (β = 0.262, 95% CI 0.195 to 0.329) and attitude (β = 0.132, 95% CI 0.060 to 0.204) were independently associated with higher practice scores. Regular blood glucose monitoring was also positively associated with practice (β = 2.554, 95% CI 1.816 to 3.291). Table 3Linear regression analysis of factors associated with knowledge scores.VariableCategory/comparisonUnivariable β (95% CI)PMultivariable β (95% CI)PSexMaleRef.Female− 0.298 (− 1.180, 0.584)0.507AgePer 1-year increase− 0.172 (− 0.228, − 0.115) < 0.001− 0.056 (− 0.101, − 0.010)0.017ResidenceRuralRef.Urban1.801 (0.381, 3.221)0.0130.233 (− 0.856, 1.322)0.675Education levelJunior high or belowRef.High school/technical secondary school2.590 (1.700, 3.479) < 0.0010.710 (− 0.105, 1.525)0.088College or above5.461 (4.263, 6.659) < 0.0012.822 (1.630, 4.015) < 0.001Monthly income (CNY) < 2000Ref.2000–50001.356 (0.162, 2.549)0.026− 0.019 (− 0.957, 0.920)0.969 > 50003.858 (2.589, 5.127) < 0.0010.172 (− 0.997, 1.342)0.773BMI (kg/m2^22) < 23.9Ref.24– < 28− 0.218 (− 1.148, 0.712)0.646 ≥ 28− 0.836 (− 2.359, 0.687)0.282HypertensionYesRef.No− 0.067 (− 1.017, 0.883)0.890Regular blood glucose monitoringYes3.640 (2.601, 4.679) < 0.0010.976 (0.109, 1.843)0.028NoRef.Glucose control method: lifestyle interventionYes0.158 (− 0.944, 1.260)0.778NoRef.Glucose control method: oral hypoglycemic agentsYes1.014 (− 0.023, 2.052)0.055NoRef.Glucose control method: insulin injectionYes2.539 (1.445, 3.633) < 0.0011.271 (0.448, 2.094)0.003NoRef.Received CGM education/promotion (hospital/other institutions)YesRef.Ref.No− 6.738 (− 7.417, − 6.059) < 0.001− 5.775 (− 6.455, − 5.096) < 0.001PAID-5 (diabetes distress)Significant distress3.615 (1.705, 5.526) < 0.0013.935 (2.536, 5.334) < 0.001No significant distressRef.Ref.β indicates the unstandardized regression coefficient. Ref. indicates the reference category. PAID-5, Problem Areas in Diabetes–5. Linear regression analysis of factors associated with knowledge scores. β indicates the unstandardized regression coefficient. Ref. indicates the reference category. PAID-5, Problem Areas in Diabetes–5. Table 4Linear regression analysis of factors associated with attitude scores.VariableCategory/comparisonUnivariable β (95% CI)PMultivariable β (95% CI)PKnowledge score (K)Per 1-unit increase0.415 (0.360, 0.470) < 0.0010.254 (0.182, 0.325) < 0.001SexMaleRef.Female− 0.368 (− 1.066, 0.331)0.301AgePer 1-year increase− 0.120 (− 0.165, − 0.075) < 0.001− 0.026 (− 0.067, 0.015)0.215ResidenceRuralRef.Ref.Urban2.471 (1.358, 3.583) < 0.0011.638 (0.658, 2.619)0.001Education levelJunior high or belowRef.Ref.High school/technical secondary school1.942 (1.227, 2.657) < 0.0010.704 (− 0.031, 1.438)0.061College or above3.835 (2.872, 4.798) < 0.0011.509 (0.419, 2.598)0.007Monthly income (CNY) < 2000Ref.Ref.2000–50000.372 (− 0.585, 1.328)0.446− 0.753 (− 1.597, 0.090)0.081 > 50002.247 (1.230, 3.264) < 0.001− 0.523 (− 1.575, 0.529)0.330BMI (kg/m2^22) < 23.9Ref.24– < 28− 0.458 (− 1.194, 0.278)0.222 ≥ 28− 0.690 (− 1.896, 0.515)0.261HypertensionYesRef.No0.042 (− 0.711, 0.795)0.913Regular blood glucose monitoringYes1.927 (1.085, 2.768) < 0.001− 0.136 (− 0.909, 0.637)0.730NoRef.Glucose control method: lifestyle interventionYes0.857 (− 0.014, 1.727)0.054NoRef.Glucose control method: oral hypoglycemic agentsYes0.179 (− 0.645, 1.003)0.670NoRef.Glucose control method: insulin injectionYes0.418 (− 0.463, 1.300)0.352NoRef.Received CGM education/promotion (hospital/other institutions)YesRef.Ref.No− 3.959 (− 4.572, − 3.347) < 0.001− 1.892 (− 2.631, − 1.154) < 0.001PAID-5 (diabetes distress)Significant distress0.552 (− 0.979, 2.083)0.479No significant distressRef.β indicates the unstandardized regression coefficient. Ref. indicates the reference category. PAID-5, Problem Areas in Diabetes–5. Linear regression analysis of factors associated with attitude scores. β indicates the unstandardized regression coefficient. Ref. indicates the reference category. PAID-5, Problem Areas in Diabetes–5. Table 5Linear regression analysis of factors associated with practice scores.VariableCategory/comparisonUnivariable β (95% CI)PMultivariable β (95% CI)PKnowledge score (K)Per 1-unit increase0.394 (0.345, 0.443) < 0.0010.262 (0.195, 0.329) < 0.001Attitude score (A)Per 1-unit increase0.362 (0.294, 0.430) < 0.0010.132 (0.060, 0.204) < 0.001SexMaleRef.Female− 0.195 (− 0.829, 0.439)0.546AgePer 1-year increase− 0.100 (− 0.141, − 0.060) < 0.001− 0.035 (− 0.070, 0.001)0.057ResidenceRuralRef.Ref.Urban1.839 (0.824, 2.854) < 0.001− 0.142 (− 0.999, 0.715)0.745Education levelJunior high or belowRef.Ref.High school/technical secondary school1.256 (0.587, 1.926) < 0.001− 0.108 (− 0.745, 0.529)0.740College or above2.309 (1.407, 3.211) < 0.001− 0.514 (− 1.465, 0.438)0.291Monthly income (CNY) < 2000Ref.Ref.2000–50000.650 (− 0.220, 1.520)0.1430.038 (− 0.694, 0.770)0.918 > 50002.147 (1.223, 3.072) < 0.0010.316 (− 0.595, 1.227)0.497BMI (kg/m2^22) < 23.9Ref.24– < 28− 0.509 (− 1.176, 0.157)0.134 ≥ 28− 1.015 (− 2.106, 0.077)0.069HypertensionYesRef.No0.614 (− 0.067, 1.295)0.077Regular blood glucose monitoringYes4.050 (3.348, 4.752) < 0.0012.554 (1.816, 3.291) < 0.001NoRef.Glucose control method: lifestyle interventionYes0.376 (− 0.416, 1.167)0.352NoRef.Glucose control method: oral hypoglycemic agentsYes1.462 (0.724, 2.200) < 0.0010.355 (− 0.324, 1.034)0.306NoRef.Glucose control method: insulin injectionYes1.739 (0.951, 2.526) < 0.0010.513 (− 0.177, 1.202)0.145NoRef.Received CGM education/promotion (hospital/other institutions)YesRef.Ref.No− 2.967 (− 3.547, − 2.387) < 0.001− 0.029 (− 0.688, 0.630)0.932PAID-5 (diabetes distress)Significant distress2.378 (1.002, 3.753)0.0011.006 (− 0.116, 2.127)0.079No significant distressRef.Ref.β indicates the unstandardized regression coefficient. Ref. indicates the reference category. PAID-5, Problem Areas in Diabetes-5. Linear regression analysis of factors associated with practice scores. β indicates the unstandardized regression coefficient. Ref. indicates the reference category. PAID-5, Problem Areas in Diabetes-5. Direct relationships of KAP SEM analysis Structural equation modeling based on standardized coefficients showed that knowledge had a significant direct effect on attitude (β = 0.553, 95% CI 0.484–0.623, P < 0.001) and practice (β = 0.439, 95% CI 0.347–0.532, P < 0.001). Attitude also had a significant direct effect on practice (β = 0.289, 95% CI 0.184–0.393, P < 0.001). In addition, knowledge exerted a significant indirect effect on practice through attitude (β = 0.160, 95% CI 0.098–0.222, P < 0.001), resulting in a total effect of β = 0.599 (95% CI 0.536–0.663, P < 0.001) (Fig. 1, Table S4). The SEM showed favorable model fit (RMSEA = 0.072, SRMR = 0.071, TLI = 0.893, CFI = 0.907), indicating that the proposed model corresponds well to the observed data (Table S5). The SEM shows the associations between KAP scores. Rectangles show observed variables, and circles represent residual terms. The direction of causality is indicated by single-headed arrows. The standardized path coefficients are presented alongside the arrows. Discussion In the study, patients with diabetes showed insufficient knowledge, moderate attitude and practice toward CGM. Positive and direct relationships between KAP scores were consistently identified. Healthcare providers should consider implementing educational programs to enhance patients’ perception of CGM usage and optimize their glycemic control. Notably, our community-based sample included adults with heterogeneous clinical profiles, and not all participants would be immediate candidates for CGM. However, contemporary guideline recommendations and real-world practice have gradually broadened CGM use beyond type 1 diabetes and intensive insulin regimens, and selected older adults and patients with type 2 diabetes may benefit in the presence of hypoglycemia risk, substantial glycemic variability, or difficulty achieving glycemic targets. Therefore, evaluating CGM-related KAP in a broad diabetes population remains important to identify knowledge gaps and barriers that may affect equitable access and future uptake as indications and reimbursement policies evolve. These findings should be interpreted in this context. Similarly, a cross-sectional study in Palestine reported limited knowledge regarding diabetes management among patients with type 2 diabetes mellitus (T2DM)31. One possible explanation for the low knowledge levels could be inadequate patient education initiatives in both regions. The findings underscore the importance of enhancing patient awareness regarding diabetes management and the role of CGM. In agreement with inappropriate practices among Chinese diabetes patients, Nguyen, Pham32 reported that only 8.5% of participants reported regular self-monitoring of blood glucose at home in Vietnam. Accordingly, efforts to improve access to CGM devices and promote patient engagement in self-care are essential for enhancing diabetes care. In the knowledge dimension, only 26.21% accurately acknowledged the appropriate use of CGM for various patient groups. The finding reflected the knowledge gap regarding the indications for CGM, which hinders CGM use in diabetes management. Besides, 28.48% accurately recognized that CGM measures glucose levels in body fluids, not blood, and cannot fully replace traditional blood glucose meters. The findings of Liao and Schembre33 were inconsistent with ours and suggested better understanding of CGM functionality and its differences from traditional blood glucose monitoring methods. Given that the study population had a low proportion of CGM users, it was possible that the non-users had little to no direct experience with the device, which could have skewed the knowledge scores. Notably, the literature is not fully consistent with our findings. Some studies conducted in settings with broader CGM availability (e.g., specialty diabetes clinics, populations with higher proportions of insulin-treated patients, or regions with stronger reimbursement support) have reported substantially higher CGM awareness, more favorable perceptions, and higher uptake than observed in our community-based sample34,35. Conversely, other studies have found that even when CGM access is improved, sustained use may remain limited due to device burden (e.g., discomfort, alarm fatigue), perceived complexity, or limited confidence in interpreting CGM data36,37. These discrepancies may reflect differences in participant characteristics (age distribution, diabetes type, treatment regimen, digital health literacy), study setting (tertiary hospitals vs community clinics), and healthcare financing (insurance coverage and out-of-pocket costs), as well as differences in how CGM-related knowledge and “use” were measured across studies. The majority (75.70%) held positive attitudes towards regular blood glucose monitoring in adjusting treatment plans and controlling blood sugar. Our results are consistent with another research from China demonstrating the significance of regular blood glucose monitoring among T2DM patients38. However, merely 8.50% expressed disbelief regarding the high cost of CGM and the related economic burden. The result may stem from various factors, including limited awareness of alternative payment options and insurance coverage39. In China, the out-of-pocket cost of CGM can be substantial because it is driven by recurrent expenditures (sensors that must be replaced regularly, and in some systems additional hardware), whereas SMBG primarily requires purchase of a glucose meter (often a one-time cost) and ongoing test strips/lancets, with total cost varying by testing frequency. Moreover, reimbursement policies for CGM are still limited or inconsistent across regions and insurance schemes, meaning that many patients must self-pay for CGM devices and consumables. This cost structure differs from SMBG and may partly explain why relatively few participants perceived CGM as affordable and why actual CGM uptake remained low in our community sample. Additionally, patients may underestimate the long-term benefits and cost-effectiveness of CGM technology in improving diabetes management and reducing related healthcare costs. Besides, only 81 participants showed no concern about the complexity of using CGM. Addressing concerns about CGM complexity requires comprehensive education on CGM device operation, data interpretation, troubleshooting, and integration into diabetes self-management40. Additionally, manufacturers can contribute by designing user-friendly interfaces, and intuitive features, and providing accessible support materials for patients41. It was concerning that only 7.99% indicated current or past use of CGM, which highlighted the discrepancy between the perceived importance of CGM and its actual utilization among patients. The low utilization rate can also account for the low-to-moderate KAP scores in the study. Barriers such as cost, discomfort, inconvenience, and lack of insurance coverage are determined as significant factors contributing to low CGM utilization9. Several possible recommendations can be made. First, targeted educational programs should be developed to increase awareness of CGM, particularly among patients with limited access. Second, healthcare providers should be trained to offer guidance and recommendations regarding CGM as a part of diabetes management. Third, addressing the cost issues improved insurance coverage and subsidies can significantly enhance CGM uptake. In addition, only 12.50% expressed willingness to recommend CGM to other diabetes patients, suggestive of the low level of endorsement and advocacy for CGM. The reluctance may result from psychological concerns, insurance coverage, and the perceived limited effectiveness of CGM42. To address the challenges, multidisciplinary cooperations are needed to improve CGM affordability, expand insurance coverage, enhance user education and training, and develop user-friendly CGM technologies. Social media platforms were recognized as a major source of health information, reflecting the increasing reliance on online communities for accessing medical knowledge43. Healthcare providers and policymakers should recognize the growing influence of social media in shaping patients’ health behaviors and develop reliable digital health resources tailored to the needs of patients with diabetes44. Patients with higher levels of knowledge about CGM tend to have more positive attitudes and proactive practices. Accordingly, interventions aimed at improving KAP towards CGM should be evaluated to determine their effectiveness in improving patient outcomes. Several influential factors of KAP were determined. First, older individuals may have lower levels of understanding regarding CGM technology, possibly due to limited access to digital health tools and age-related cognitive decline45. Second, negative associations between the absence of education and the promotion of CGM and both knowledge and attitude were reported. Empowering individuals with accurate information and fostering positive attitudes, education, and promotion can facilitate the adoption of CGM. Third, positive impacts of regular blood glucose monitoring on knowledge and practice were identified. The benefits of regular monitoring to patients should be highlighted, and efforts are also needed to address barriers, including cost, access to testing supplies, and patient motivation. Fourth, patients with higher levels of education may have greater access to healthcare resources, including diabetes education programs, which can further enhance their knowledge and attitudes toward CGM46. The study had several limitations. First, the cross-sectional design limits causal inference; therefore, the observed relationships should be interpreted as associative rather than causal. Second, participants were recruited during routine visits using a convenience-based approach without randomization. As individuals who agreed to participate may differ systematically from non-participants (e.g., in motivation, health literacy, socioeconomic status, or disease severity), selection bias may be present, which could weaken internal validity and limit representativeness. Third, regional practices and institutional policies in Suzhou region were taken into the formulation of the questionnaire, thereby constraining the generalizability of our findings. In addition, the relatively high mean age of the study population may further limit generalizability, particularly when comparing our findings with regional and international studies that include populations with different age distributions. Forth, self-reported data introduced the possibility of social desirability bias, potentially leading to inflated scores47. Fifth, we excluded individuals with severe chronic diseases and serious complications of diabetes. Despite the exclusion was to minimize the confounding effects of these conditions, it may limit the generalizability of our findings to diabetes patients with coexisting severe conditions. Sixth, despite the data collection was strictly conducted, the self-developed questionnaire may not fully capture the nuanced KAP levels of CGM. More objective assessment methodology is required to validation our findings. Seventh, the duration of diabetes varied across participants, which may influence diabetes knowledge, technology acceptance, and self-management behaviors; however, diabetes duration was not analyzed as an explanatory factor in this study, and residual confounding cannot be excluded. Eighth, the questionnaire was developed primarily with input from clinical physicians, and did not involve multidisciplinary experts such as health psychologists or questionnaire methodologists. Although the items were based on clinical guidelines, relevant literature, and practical experience in diabetes management, and the instrument demonstrated acceptable reliability and structural validity, the lack of multidisciplinary input may limit the comprehensiveness and theoretical robustness of the questionnaire. Finally, CGM use was uncommon in our sample; since participants who had used CGM or were familiar with it may report higher KAP, the estimates and associations should be interpreted with caution. Conclusions Insufficient knowledge, and only moderate attitudes and practices toward CGM were observed among Chinese patients with diabetes in the Suzhou region, together with positive and direct associations among KAP scores. These findings are important because CGM is increasingly recommended to support timely treatment adjustment and reduce dysglycemia, yet the identified knowledge gaps and very low use suggest that many patients in community settings may be unable to benefit from diabetes technology even when clinically appropriate. The observed KAP relationships indicate that improving CGM-related knowledge may help shape attitudes and encourage adoption and sustained use. Accordingly, targeted educational and behavioral interventions—particularly for older patients, those with lower educational attainment, and those without prior exposure to CGM education/promotion—may help improve technology literacy and promote more equitable access to CGM-supported care. Future multicenter studies with larger and more diverse samples are warranted to confirm these findings and to evaluate the effectiveness of KAP-focused interventions on CGM uptake and glycemic outcomes. Supplementary Information Supplementary Information 1. Supplementary Information 2. Supplementary Information 3. Acknowledgements Not applicable. Abbreviations Continuous glucose monitoring Knowledge, attitudes, and practices Glycated hemoglobin Problem areas in diabetes-5 Body mass index Standard deviations Root mean square error of approximation Standardized root mean square residual Tucker–Lewis index Comparative fit index Author contributions Feng Zhang and Li Wang carried out the studies, participated in collecting data, and drafted the manuscript. Feng Zhang and Li Wang performed the statistical analysis and participated in its design. All authors read and approved the final manuscript. Funding Not applicable. Data availability All data generated or analysed during this study are included in this published article. Declarations Competing interests The authors declare no competing interests. Ethics approval and consent to participate The study was carried out after the protocol was approved by the Ethics Committee of the Runda Community Health Service Center. I confirm that all methods were performed in accordance with the relevant guidelines. All procedures were performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments, and informed consent was obtained from all participants. Consent for publication Not applicable. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Information 1. Supplementary Information 2. Supplementary Information 3. Data Availability Statement All data generated or analysed during this study are included in this published article.