Understanding Health-Related Behavioral Protection: A Latent Profile Analysis

Introduction Type 2 diabetes mellitus (T2DM), one of the most prevalent chronic diseases worldwide, is experiencing a continuous rise in incidence and has become a major public health concern. Epidemiological data indicate that T2DM accounts for approximately 90% of all diabetes cases, with an estimated global patient population of 828 million. Of these, China alone accounts for 148 million cases, nearly one-fifth of the global diabetic population.1,2 Due to the long-term, often lifelong, nature of diabetes management—compounded by limited disease awareness and poor adherence to treatment regimens—the treatment and glycemic control rates among Chinese patients have remained low, with glycemic control achieved in only about 50% of cases.3 This situation not only significantly compromises patients’ quality of life and prognosis but also imposes a substantial socioeconomic burden.4 Therefore, identifying effective strategies to promote health behaviors, improve blood glucose and lipid control, and enhance patients’ self-management capabilities has become an urgent priority.5–8 To address these behavioral challenges, the Protection Motivation Theory (PMT) provides a useful framework for understanding the key factors that influence health-related decision-making in chronic disease management.9,10 This theory systematically explains the psychological decision-making process individuals undergo when confronted with health threats.11 It posits that protective motivation is formed through two key cognitive processes: threat appraisal (which includes perceived severity and vulnerability) and coping appraisal (which involves perceived efficacy of the preventive behavior and self-efficacy).12 PMT has been widely applied across various health behavior research domains. For instance, it has successfully increased adults’ willingness to undergo colorectal cancer screening and improved breast cancer screening behaviors among women.13–15 In the context of chronic disease management, studies have demonstrated that PMT significantly influences health behaviors in patients with diabetes. When patients acquire relevant health knowledge and skills, the strength of their protective motivation is positively associated with adherence to health behaviors.16,17 Health education interventions based on PMT have been shown to not only improve glycemic control and body mass index (BMI) but also significantly enhance self-management capabilities and quality of life.18 However, clinical observations reveal considerable individual differences. These differences manifest in both the intensity of patients’ protective motivation and the efficiency of translating that motivation into actual behavior. These differences are evident in both cognitive appraisal processes and preferences for intervention strategies. With the rapid advancement of digital health technologies, such variations may be closely related to patients’ levels of eHealth literacy (eHL) —an area that warrants further investigation. With the advancement of “Internet Plus” technologies, the concept of eHL been increasingly introduced into the field of chronic disease prevention and management.19–21 Significant individual differences have been observed in patients’ acceptance and usage patterns of digital health tools. eHL typically refers to an individual’s ability to seek, acquire, understand, evaluate, and apply health information through electronic resources such as the internet, health management applications, and online communities.22 eHL plays a vital role in improving the medical quality of diabetes patients, reducing gaps, empowering patients, and promoting informed decision-making and self-management.23–26 A cross-sectional survey of undergraduate students in Jordan showed that although most students recognize the usefulness of online health information, their average level of electronic health literacy is only 16.61 ± 4.10 (out of 40 points), which is relatively insufficient, and their level is significantly influenced by school type and professional background.27 This suggests significant group differences in electronic health literacy. Research has shown a moderate positive correlation between eHL and health behaviors. eHL is considered a potential mediating factor through which health-related information leads to behavioral changes,28 and it shows the strongest association with health-promoting behaviors.29 In the field of diabetes, studies have found that patients with higher eHL are more likely to use smart devices to proactively monitor blood glucose, exhibit better adherence to dietary and exercise regimens, and have a clearer understanding of the risks of complications—factors that contribute to improved self-efficacy and self-management behaviors.30,31 Moreover, research has also found that eHL indirectly influences the self-management behaviors of patients with T2DM through self-efficacy and social support. This suggests that eHL not only directly affects health behaviors but also exerts an indirect impact through other psychological and social factors.32 Based on the PMT framework, electronic health literacy is likely to influence health behavior by shaping the core cognitive assessments outlined in this theory. Individuals with higher electronic health literacy may be more capable of accessing, understanding, and critically evaluating online health information. This enhanced ability may lead to a more accurate perception of disease severity and individual susceptibility (threat assessment), as well as a stronger belief in the effectiveness of recommended health actions and one’s own ability to perform these actions (coping assessment). Therefore, electronic health literacy may serve as a key upstream factor in strengthening protective motivation in the PMT model, ultimately promoting better self-management. However, the heterogeneity of how different subgroups of T2DM patients integrate digital health information into their motivational characteristics remains unclear. This study innovatively adopts the potential profile analysis method, aiming to fill the research gap in the heterogeneity of health behavior protection motivation in the group of T2DM patients and its correlation mechanism with electronic health literacy. Methods Design This cross-sectional study was conducted between January and March 2025 at a tertiary hospital in Changsha, China, using convenience sampling. The inclusion criteria were: (1) a clinical diagnosis of T2DM based on the 2020 Chinese Guidelines for the Prevention and Treatment of Diabetes;33(2) voluntary participation with informed consent; and (3) ability to use electronic devices such as smartphones to access online health information. The exclusion criteria were: (1) patients with psychiatric or cognitive impairments; (2) those whose responses contained significant logical inconsistencies. These inconsistencies were assessed by verifying coherence among relevant questionnaire items and through follow-up confirmation. The sample size of this study was estimated based on common empirical rules for Latent Profile Analysis (LPA), with full consideration given to data quality and response rates. Referring to previous studies on T2DM,34,35 it was assumed that the number of latent classes (K) would be between 2 and 3, with 11 continuous indicators included. According to LPA methodology recommendations, model fitting typically requires a sample size of no less than 200 cases. To ensure data quality and control the impact of invalid questionnaires on statistical results, the sample size was expanded by 20% on the basis of calculation, with a target sample size of 240 cases. A total of 260 questionnaires were distributed in the study, and 253 valid questionnaires were collected, with an effective response rate of 97.3%, which met or slightly exceeded the estimated sample size and met the basic requirements for fitting latent category models and subsequent statistical testing. This study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. The survey was administered anonymously. Ethical approval was obtained from the Medical Ethics Committee of the Second Affiliated Hospital of Hunan University of Chinese Medicine in December 2024 (Approval Number: 2024-KY-074-01). Data Collection This study distributed electronic questionnaires through the Internet platform, and all questionnaires were filled in anonymously. To ensure the quality of responses, including those from elderly or patients with limited digital experience, the questionnaire design adopts a concise and intuitive interface with guided filling instructions, and provides telephone assistance options during the recruitment phase. To control data quality, the system automatically filters out the following invalid questionnaires: questionnaires with logically inconsistent answers, identical item ratings, or completion times less than 60 seconds. During data analysis, entries with missing values were addressed using case exclusion. For continuous variables (eg, total scale scores, age), potential outliers were identified through a conservative, multi-step process. First, values falling outside the range of the mean ± 3 standard deviations were flagged as potential extremes. Each flagged value was then individually reviewed based on professional knowledge and the questionnaire context. For example, for records with a “weekly blood glucose monitoring frequency” of 0 times or more than 21 times (ie 3 times or more per day), a reasonable evaluation will be conducted based on the patient’s self-management ability description. Ultimately, only extreme data that are clearly filled in incorrectly or cannot be reasonably explained are excluded to maximize the retention of valid information and ensure the robustness of the analysis results. Measures Health-Related Behavioral Protection Motivation Scale of Patients with T2DM This scale was developed by Peng Lin et al and included 7 dimensions with susceptibility, severity, internal reward, external reward, response efficacy, response cost, and self⁃efficacy, involving 34 items.17 The susceptibility dimension consists of 5 items, such as “If I do not follow a proper diet, it will worsen both acute and chronic complications of diabetes” and “If I do not exercise regularly, I will not be able to control my blood glucose”. The severity dimension includes 5 items, such as “Worsening diabetes will expose me to the risk of various acute and chronic complications (such as blindness, amputation, etc.)” and “Worsening diabetes will make me feel nervous and anxious”. The internal reward dimension consists of 4 items, such as “Eating sweets or other foods I like (which raise blood glucose) makes me feel relaxed and satisfied” and “Exercising less makes me feel comfortable and at ease”. The external reward dimension consists of 5 items, such as “My family and friends suggest folk remedies or medications for treating diabetes and encourage me to try them.” and “My family and friends hope that I control my blood glucose only through diet and exercise, as they are concerned about the side effects of medication”. The response efficacy dimension includes 5 items, such as “I believe that controlling my diet can help me avoid complications of diabetes” and “I believe that increasing physical activity is key to managing my blood glucose”. The response cost dimension includes 4 items, such as “Eating less or not eating sweets/my favorite foods reduces my enjoyment of life” and “Due to my busy work schedule, I find it difficult to maintain an exercise routine”. The self-efficacy dimension includes 6 items, such as “I am confident that I can eat three meals a day at regular times and in fixed portions” and “I am confident that I can exercise for 15–30 minutes each time, 4–5 times a week”. The scale uses a 5-point Likert scale, where “1” to “5” represents a range from “strongly disagree” to “strongly agree”. For the severity, susceptibility, response efficacy, and self-efficacy dimensions, scoring is positive, meaning higher scores indicate stronger protection motivation. For the internal reward, external reward, and response cost dimensions, scoring is reversed, meaning higher scores indicate weaker protection motivation. The Cronbach’s alpha coefficient range for the seven dimensions of the scale is 0.772~0.88. In this study, the Cronbach’s alpha coefficient range for the seven dimensions of the scale is 0.772~0.887. The scale level content validity index (S-CVI) is 0.917, and the item level content validity index (I-CVI) is 0. 833~1.000. Exploratory factor analysis extracted 7 common factors with a cumulative variance contribution rate of 62.078%. Confirmatory factor analysis shows that the overall fit of the model is good. eHL Self-Rating Scale for Patients with Diabetes Mellitus This scale was developed by Jiang Feng et al36 and is primarily used to assess eHL among patients with diabetes. It consisted of 4 dimensions and 27 items. The 4 dimensions were functional eHL, communicative eHL, critical eHL and transformational eHL. The functional eHL dimension consists of 7 items, such as “I often read health information related to diabetes on the internet” and “I know what diabetes-related health information I need”. The communicative eHL dimension includes 5 items, such as “I communicate with healthcare providers about the diabetes-related information I obtain from the internet” and “I share the diabetes-related information I obtain from the internet with family, friends, or other patients”. The critical eHL dimension consists of 6 items, such as “I can filter out the diabetes-related information on the internet that is useful to me” and “I can assess the credibility of diabetes-related advertisements on the internet”. The transformational eHL dimension consists of 9 items, such as “I can use diabetes-related electronic applications for self-management” and “I can use the information obtained from the internet about blood glucose monitoring to perform self-monitoring and record my blood glucose levels”. The scale uses a 5-point Likert scale, where “1” to “5” represents a range from “strongly disagree” to “strongly agree”. In this study, the Cronbach’s alpha coefficient of the scale was 0.906, and the exploratory factor analysis Kmo value was 0.954. Four common factors were extracted, and the cumulative variance contribution rate was 60.968%, indicating good reliability and validity. Sociodemographic Information Baseline demographic and clinical information was collected using a structured questionnaire designed by the researchers based on relevant literature. A total of 21 items were included: Gender (male, female); Age (years); Ethnicity (Han, minority); Residence (urban, rural); Educational level (primary school or below, junior high school, high school or vocational school, college diploma, bachelor’s degree or above); Marital status (unmarried, married, divorced, widowed); Occupation (public institution employee, worker, farmer, self-employed, other); Current employment status (employed, unemployed or retired); Medical expense payment method (public insurance, basic medical insurance, new rural cooperative medical scheme, commercial insurance, self-pay); Monthly household income per capita (RMB): (<3000; 3000–6000; 6001–9000; >9000); Primary caregiver (spouse, children, parents, siblings, other); Self-rated health status (very good, good, fair, poor, very poor); Exercise habits (regularly, occasionally, none); Self-care ability (Barthel Index): (severe dependence, moderate dependence, mild dependence, independent); Duration of diabetes (years); Number of hospitalizations due to diabetes (times); Presence of other chronic diseases (yes, no); Most recent fasting blood glucose level (mmol/L); Presence of diabetes-related complications (yes, no); Frequency of blood glucose monitoring per week (≤3 times, >3 times); Missed medication status (Never, occasionally, Frequently). Statistical Analyses Data were analyzed using IBM SPSS Statistics version 27.0 and Mplus version 8.3, and the analysis was conducted in three main steps. First, latent profile analysis (LPA) was performed using Mplus to identify distinct motivational profiles among patients with T2DM. Model fit was evaluated using the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and sample-size-adjusted BIC (aBIC), with lower values indicating better model fit. The Lo-Mendell-Rubin Likelihood Ratio Test (LRT) and the Bootstrap Likelihood Ratio Test (BLRT) were used to compare the k-class model with the (k−1)-class model, and a p-value < 0.05 indicated that the k-class model was superior. Entropy values were used to assess classification accuracy, with values closer to 1 indicating higher precision.37 Second, demographic differences between latent profiles were analyzed using chi-square tests or rank-sum tests in SPSS. A one-way ANOVA (Kruskal–Wallis test) was used to compare eHL scores across the different protective motivation profiles. Finally, variables found to be statistically significant in univariate analysis were included in a multivariate logistic regression model to identify factors associated with profile membership. Categorical variables were described using frequencies and proportions, while non-normally distributed continuous variables were presented as medians and interquartile ranges [M (P25, P75)]. Common method bias and scale reliability were also assessed using SPSS. A two-tailed α level of 0.05 was used to determine statistical significance. Results Sociodemographic Characteristics of the Participants A total of 253 patients with T2DM were included in the final analysis, yielding a valid response rate of 97.3%. The mean age of participants was 49.98 ± 11.70 years, with 152 males (60.1%) and 101 females (39.9%). The age range was 22 to 78 years, with a median age of 49 years (interquartile range: 41–58 years). Age stratification showed that 58 participants (22.9%) were aged ≤40 years, 162 (64.0%) were aged 41–59 years, and 33 (13.0%) were aged ≥60 years. Regarding educational level, 33 participants (13.0%) had completed junior high school or below, 83 (32.8%) had completed high school or technical secondary school, and 137 (54.2%) had attained a college degree or above. Harman’s single-factor test was used to detect common method bias. The results showed that 9 common factors were extracted, with the first factor’s explanatory power being 21.25%, which is below the 40% critical threshold, indicating that there is no significant common method bias in this study’s data. Selection and Naming of the Models The mean total score for health-related behavioral protection motivation among patients with T2DM was 120.14 ± 9.22. Latent profile analysis was conducted using the scores from the seven dimensions of the health-related behavioral protection motivation Scale, and five models were extracted in total. Model fit indices are presented in Table 1. As shown in the table, as the number of model categories increases, the values of AIC, BIC, and aBIC gradually decrease, and the Entropy entropy values are all greater than 0.8. When two potential categories were retained, both LMR and BLRT reached significant levels (both P<0.001), and the category probability distribution of Model 2 was balanced (35.97% vs 64.03%), avoiding the occurrence of extremely small categories in Models 4–5 (such as the existence of only 0.79% of categories in Model 5), which is more in line with the need for identifying and intervening in representative subgroups in clinical practice. Although the three models also showed good fitting trends, the P-value of their LMR test was 0.6488, which did not reach a significant level, indicating that adding a third category did not bring statistically significant improvement. In addition, the minimum category probability among the three models is only 5.45%, and the sample size is relatively limited, which may affect the stability and extrapolation of this category. Therefore, based on the clinical interpretability and practicability of model fitting indicators, statistical test results, and category distribution, this study determined that category 2 models were the best potential profile solutions for type 2 diabetes patients’ health-related behavior protection motivation. According to the profile distribution chart (Figure 1), patients in Class 1 had higher scores in internal reward, external reward, and response cost, and this group was labeled as the “high perceived cost–incentive-dependent group”, comprising 91 participants (35.97%). Patients in Class 2 had higher scores in susceptibility, severity, response efficacy and self-efficacy, and were labeled as the “high sensitivity–high efficacy group”, comprising 162 participants (64.03%). Analysis of eHL in Patients with T2DM The total eHL score of patients with T2DM in the High Sensitivity–High Efficacy group was (110.76 ± 13.78), significantly higher than that of the High Perceived Cost–Incentive-Dependent group, which was (100.35 ± 17.89). Analysis of the differences in eHL between the two groups showed that the scores for functional eHL, communicative eHL, critical eHL, and transformational eHL dimensions, as well as the total score, were all lower in the High Perceived Cost–Incentive-Dependent group compared to the High Sensitivity–High Efficacy group, with statistically significant differences (all P < 0.001). The most significant differences were found in communicative eHL (20.85 ± 2.49 vs 18.93 ± 3.28) and transformational eHL (36.51 ± 5.59 vs 32.74 ± 7.23), as shown in Table 2. Univariate Analysis of Potential Categories of Health-Related Behavioral Protection Motivation in Patients with T2DM The results of univariate analysis showed that patients with different payment methods, weekly blood glucose testing frequencies, and exercise habits had significant differences in the potential categories of health-related behavioral protection motivation (all P < 0.05), as shown in Table 3. Binary Logistic Regression Analysis of Potential Categories of Health-Related Behavioral Protection Motivation in Patients with T2DM A binary logistic regression analysis was conducted with payment method, weekly blood glucose testing frequency, and exercise habits as independent variables, and latent profile categories as the dependent variable. The results showed that, with the High Perceived Cost–Incentive-Dependent group as the reference group, blood glucose testing frequency showed a statistically significant difference (P < 0.001). Patients who tested their blood glucose more than three times per week had 2.95 times higher protection motivation compared to those who tested ≤3 times per week (OR = 2.954, 95% CI: 1.679–5.197), as shown in Table 4. Discussion Protection Motivation Theory (PMT) emphasizes that individuals make behavioral changes in response to perceived threats as a means of self-protection.38 The health-related behavioral protection motivation of patients with T2DM is the willingness to perform and maintain health protection behaviors, which arises from a comprehensive assessment of disease risks.39 In line with the first aim of this study, latent profile analysis successfully identified two distinct subgroups based on their motivational characteristics, confirming the heterogeneous nature of protection motivation within the T2DM population. The motivation level of patients is positively correlated with the adherence to health-related behaviors and directly influences the onset and progression of diabetes.40 The results of this study showed that the health-related behavioral protection motivation score for patients with T2DM was (120.14 ± 9.22), which is at a moderate level and slightly higher than the results of Peng Lin et al (105.77 ± 11.03).39 This difference may be related to recent socioeconomic development and the increasing public health awareness.41 This study found that there is heterogeneity in health-related behavioral protection motivation among patients with T2DM. Based on their characteristics, patients can be divided into two categories: the High Perceived Cost–Incentive-Dependent group and the High Sensitivity–High Efficacy group. The High Perceived Cost–Incentive-Dependent group accounts for 35.97%. Patients in this group scored higher in the internal reward, external reward, and response cost dimensions, indicating weaker health-related behavioral protection motivation. They are more dependent on internal or external rewards, such as “Exercising less and being less active makes me feel comfortable” and “My family and friends hope I avoid insulin due to concerns about dependency”, etc. At the same time, they find it troublesome to adhere to diabetes-related health behaviors, such as “Eating less or avoiding sweets/my favorite foods reduces my enjoyment of life” and “Managing my weight takes too much of my time and energy”. These statements reflect the inhibitory effect of high perceived costs and rewards on health behaviors. In the High Perceived Cost group, although individuals experience higher internal rewards (such as comfort), they also exhibit lower self-efficacy. This suggests that they may perceive the costs of engaging in protective behaviors as high while feeling insufficient in their ability to cope. This contradiction reflects a conflict between cost assessments (such as time costs, financial costs, etc.) and reward perceptions (such as comfort, immediate satisfaction, etc), which is associated with an inhibitory effect on behavior change. Jiang Jiawei, through an intervention model based on PMT, combined offline education with online management to reduce internal and external rewards, and enhance the motivation levels of individuals at risk of prediabetes, thereby encouraging them to adopt health-promoting behaviors and improving health outcomes.42 Zhang promoted the generation of self-protection motivation by providing patients with family and social support, weakening the internal and external rewards of unhealthy behaviors.43 Other studies have found that presenting successful case examples can help improve the response efficacy in patients with T2DM. Personalized guidance can assist patients in reducing response costs. It is also recommended that healthcare providers prioritize motivation strategies for diabetes patients, emphasizing the relationship between health-related behaviors and the onset and progression of the disease to enhance their protection motivation levels. Personalized interventions should be provided in various areas, including dietary management, exercise management, weight management, blood glucose monitoring, medication adherence, and regular follow-up visits.44,45 The High Sensitivity–High Efficacy group accounts for 64.03%, and patients in this group exhibit strong health-related behavioral protection motivation. They score higher in the dimensions of susceptibility, severity, response efficacy, and self-efficacy, and are better able to recognize the potential risks and complications associated with diabetes, such as “If I do not follow a proper diet, it will worsen both acute and chronic complications of diabetes” and “Worsening diabetes will increase the economic burden”. Therefore, their adherence to health-related behaviors is higher, as seen in statements like, “I am confident that I can eat three meals a day at regular times and in fixed portions” and “I have the information to control my weight within a reasonable range”. This is consistent with the “Threat Appraisal–Coping Efficacy” dual-path model proposed by Maddux et al46 which suggests that when individuals perceive the severity of a threat and their own coping abilities as strong, they are more likely to engage in protective behaviors. Both threat appraisal and coping efficacy jointly determine an individual’s behavioral choices. Zhao Xiaoling, by increasing patients’ awareness of the severity and susceptibility of diabetes, stimulated self-protection motivation, identified the reasons for patients’ lack of self-care, reduced internal and external rewards, and provided personalized interventions and continuous health education, thereby improving patients’ self-efficacy.47 Healthcare providers should fully recognize the importance of health-related behavioral protection motivation in patients with T2DM. Through accurate assessments, they can understand the motivation levels behind patients’ health-related behaviors, identify obstacles in behavior change, and develop one-on-one interventions to help patients improve adherence to health-related behaviors. The study found that patients in the High Sensitivity–High Efficacy group scored higher than those in the High Perceived Cost–Incentive-Dependent group across all dimensions of functional eHL, communicative eHL, critical eHL, and transformational eHL, as well as in the total score. This finding directly addresses the second research aim, demonstrating a significant association between motivational profiles and eHL levels. It suggests that the core cognitive and efficacy appraisals which define the high motivation profile may be supported or enhanced by higher digital health competencies. There were significant individual differences in the acceptance and usage patterns of digital health tools, which may be related to their information processing capabilities. In line with existing evidence, patients with diabetes who possess higher eHL are more likely to proactively monitor their blood glucose, control their diet, engage in regular physical exercise, and have a clearer understanding of the risks associated with their condition. Patients with higher eHL are more likely to access diabetes-related health information, such as educational videos, professional websites, and academic research updates. They can search for diabetes-related information online, recognize its dangers, and assess the accuracy of the information, extracting healthier lifestyle habits for self-management and prevention of complications. In contrast, patients with lower eHL lack these abilities, which is consistent with Zhang Xia’s research.31 Self-management of diabetes is closely related to lifestyle, and patients with higher eHL tend to adopt healthier lifestyles.48 This suggests that clinical healthcare providers can establish diabetes information platforms, train patients to search for and evaluate online information, thereby potentially supporting the enhancement of their health-related behavioral protection motivation and improving adherence to diabetes-related health behaviors. Moreover, among the various dimensions, the most significant differences between the two groups were found in communicative eHL (20.85 ± 2.49 vs 18.93 ± 3.28) and transformational eHL (36.51 ± 5.59 vs 32.74 ± 7.23). This may primarily be due to differences in motivational characteristics. The High Sensitivity–High Efficacy group, with stronger self-efficacy (4.5 ± 0.4) and response efficacy (4.3 ± 0.6), is more adept at communicating with healthcare providers and transforming health information into practice. On the other hand, the High Perceived Cost–Incentive-Dependent group faces application barriers due to higher response cost (4.1 ± 0.4). In the health belief model, perceived costs (such as response costs) influence patients’ behavioral choices, and high perceived costs may lead to resistance to interventions.49 This suggests that future clinical practices should focus on improving patients’ ability to apply health information and address their individualized needs regarding intervention methods. Multivariate logistic regression analysis showed that blood glucose testing frequency was statistically significant (P < 0.001), with patients who tested their blood glucose more than three times per week having 2.95 times higher protection motivation compared to those who tested ≤3 times per week (OR = 2.954, 95% CI: 1.679–5.197). This may be related to the self-management reinforcement effect brought about by frequent monitoring. Frequent blood glucose testing may enhance patients’ disease perception by providing immediate physiological feedback, which in turn facilitates the formation of protection motivation. In the multi factor analysis model, only the frequency of blood glucose monitoring maintained independent significance, highlighting the core role of this behavior in driving protective motivation. In contrast, payment methods and exercise habits were significant in univariate analysis, but no longer significant in multivariate analysis. This result suggests that blood glucose monitoring may be a more direct and actionable behavioral intervention target. This suggests that clinical practitioners should pay attention to the positive impact of blood glucose monitoring frequency on patients’ self-management motivation. Related studies also show that nursing interventions based on PMT can increase the frequency of blood glucose monitoring.50,51 For patients with lower protection motivation, personalized monitoring plans could be considered, with the appropriate increase in blood glucose monitoring frequency (eg, more than three times per week), leveraging the positive feedback effect generated by the monitoring behavior itself to help patients establish a virtuous cycle of disease awareness and health behavior. However, it is important to note that high-frequency monitoring may increase the patient’s burden, and it is recommended to optimize monitoring strategies using mobile health technologies (eg, continuous glucose monitoring) or shared decision-making between healthcare providers and patients, achieving a balance between enhancing protection motivation and reducing patient burden.52 The payment method was statistically significant in univariate analysis but did not show a significant effect in multivariate logistic regression (P > 0.05). This suggests that the impact of payment methods on protective motivation may not exist independently, but rather be explained by other socio-economic or clinical factors such as income level and accessibility of medical resources, or by these more fundamental variables in multivariate analysis. The possible reason for this could be the insufficient sample size of categories such as commercial insurance and out-of-pocket payment methods, suggesting that future research should validate this factor with a larger sample size. Exercise habits were statistically significant in univariate analysis but did not show a significant effect in multivariate logistic regression (P > 0.05). However, a dose-response trend was observed, with protection motivation decreasing in the following order: regular exercise > occasional exercise > no exercise. Although this did not reach statistical significance in multivariate analysis, the dose-response trend (regular exercise > occasional exercise > no exercise) still holds potential clinical significance, reflecting the positive impact of exercise on protection motivation through improved blood glucose control and enhanced self-efficacy, consistent with other studies.28 It indicates a robust correlation pattern between exercise behavior and protective motivation, which has not become an independent predictor. This may suggest that the impact of exercise on motivation is partially mediated through pathways such as improving blood sugar control or enhancing self-efficacy, rather than a direct effect. It is recommended that future studies validate this association by expanding the sample size. At the same time, when formulating intervention strategies, promoting regular exercise can be considered as a potential target for enhancing protection motivation in diabetes patients, particularly for those who are completely inactive, who may require priority intervention. This study is a cross-sectional study. As such, the observed associations, while informative, do not establish causality. This study innovatively applied potential profile analysis to the study of health-related behavior protection motivation of patients with T2DM, and for the first time explored the relationship between different types of protection motivation and e-health literacy, providing new evidence for understanding the internal psychological mechanism of patients’ health behavior in the digital context. It also identified the influencing factors that can be intervened, such as payment methods, blood glucose monitoring frequency, etc, through multi factor analysis, which helps to refine the intervention measures. Due to geographical and time constraints, the sample size included was limited, and it was not possible to dynamically assess the characteristics of health-related behavioral protection motivation at different stages of T2DM. Limitations This study adopts a cross-sectional design, and the convenient sampling method is used to recruit subjects, which may lead to sample selection bias, affect the external validity of the results and the representativeness of the wider type 2 diabetes patients. Due to the limitations of implementation time and geographical scope, the sample size is relatively limited, which may further reduce the confidence of statistical testing and limit the ability to conduct in-depth analysis of important subgroups (such as different disease course or complication status). It is recommended that future research expand the sample size under feasible conditions or engage in multi center collaboration to enhance the robustness and extrapolation of the results. In addition, the two core scales used in this study - diabetes patients’ health-related behavior protection motivation scale and diabetes patients’ electronic health literacy self-assessment scale were independently developed under the Chinese cultural background. Although good reliability and validity have been demonstrated in both this study and previous research, it has not yet been widely validated and applied internationally. This to some extent limits the direct comparability of research results with similar international studies. It is suggested that in future research, internationally validated scales can be used for cross-cultural comparisons, or the scales used in this study can be adapted and validated across cultures to further confirm their universality. Conclusion This study performed a characteristic analysis of health-related behavior protection motivation in patients with type 2 diabetes, identifying two distinct profiles: a high-perceived-cost incentive-dependent group and a high-sensitivity high-efficacy group. The former demonstrated weaker motivation and poorer behavioral adherence, representing a high-risk population that requires focused attention in clinical practice. Weekly frequency of blood glucose monitoring was identified as one of the influencing factors of this protection motivation, suggesting the need for enhanced early assessment and targeted interventions for such patients. Furthermore, protection motivation was significantly correlated with e-health literacy, indicating that improving patients’ ability to access and evaluate online health information can enhance their self-management and improve clinical outcomes. The findings of this study provide a theoretical basis for the development of precision-based digital diabetes management strategies. Future studies could consider employing a longitudinal time-series design across multiple centers to further explore the dynamic trajectory of health-related behavioral protection motivation in patients with T2DM over time.