Nomogram for Predicting the Risk Factors of Diabetes Mellitus in Middle-Aged Adults: A Secondary Analysis of the 2022 Korea National Health and Nutrition Examination Survey

Article information

J Korean Acad Fundam Nurs. 2025;32(4):421-431

Publication date (electronic) : 2025 November 30

doi : https://doi.org/10.7739/jkafn.2025.32.4.421

Sook Kyoung Park¹⁾, Hye Young Kim^1),, Hyuk Joon Kim²⁾

¹⁾Professor, College of Nursing, Jeonbuk Research Institute of Nursing Science, Jeonbuk National University, Jeonju, Korea

²⁾Assistant Professor, Department of Nursing, Wonkwang Health Science University, Iksan, Korea

Corresponding author: Kim, Hye Young College of Nursing, Jeonbuk National University 567 Baekje-daero, deokjin-gu, Jeonju 54896, Korea Tel: +82-63-270-4618, Fax: +82-63-270-3127, E-mail: tcellkim@jbnu.ac.kr

*This work was supported by the Korean Academy of Fundamentals of Nursing Research Grant of 2023.

Received 2025 April 22; Revised 2025 October 2; Accepted 2025 November 16.

Abstract

INTRODUCTION

Diabetes mellitus (DM) is a metabolic disease with an increasing prevalence worldwide, including in Korea [1]. According to the World Health Organization, approximately 422 million adults globally have DM, and this number is projected to reach 642 million by 2045[1]. In Korea, the DM prevalence among adults aged 30 years and older was 16.3% in 2021 (19.4% in men, 13.4% in women), showing a steady increase [2]. DM-related healthcare costs increased by over 60% from approximately 1.8 trillion KRW in 2015 to 2.9 trillion KRW in 2020, highlighting a substantial eco-nomic burden [3]. Considering the 41.1% prevalence of pre-DM among adults aged 30 years and older (2021∼ 2022), raising awareness and developing proactive measures are urgent [2].

Moreover, DM is a direct or indirect cause of various complications, including cardiovascular and cerebrovas-cular diseases, necessitating continuous management at both the individual and national levels [4]. The Korean Diabetes Association (KDA) emphasizes the importance of early diagnosis by recommending DM screening from the age of 35 years instead of 40 years; however, annual blood screening has limitations [4]. Public health centers provide integrated counseling programs for the early de-tection of hypertension, diabetes, and dyslipidemia. Nevertheless, type 2 diabetes mellitus (T2DM) often remains undiagnosed until complications occur. About one-third of individuals with diabetes are unaware of their condition [4]. Early diagnosis can prevent or delay complications [5], and optimized prediction models can reduce the incidence of DM through lifestyle improvements [6].

Nomograms have recently emerged as effective tools for risk factor identification and disease prediction [7-10]. Nomograms graphically represent clinical data, aiding individualized risk assessment and decision-making by non-experts [7]. Various nomograms have been developed internationally and domestically for cancer survival in children [8], predicting intensive care unit stays for patients with chronic obstructive pulmonary disease [9], and cognitive changes in elderly individuals with mild cognitive impairment [7]. However, in nursing, nomograms have been limited to predicting diabetic foot ulcers [10] and cognitive function changes [7]. Park et al. [11] developed a statistical DM risk prediction model for predicting T2DM risk using logistic regression and a naive Bayesian classifier. While this study contributed to methodological comparisons, it did not provide individualized risk scoring or visualization tools, nor did it undergo external validation in real-world clinical settings. Therefore, the applicability of this model in nursing practice and patient counseling remains limited.

Beyond its statistical advantages, the nomogram has methodological significance in nursing. By converting regression results into individualized risk scores and visual tools, the nomogram enables nurses to identify high-risk individuals, provide tailored lifestyle counseling, and design preventive programs in both community and clinical settings. This methodological approach therefore offers a practical framework for evidence-based nursing interventions and contributes to the development of nursing strategies for diabetes prevention. Although both body mass index (BMI) and waist circumference (WC) are estab-lished measures of obesity, WC has been reported to be more strongly associated with insulin resistance and T2DM in Asian populations [12-14]. In particular, BMI does not adequately reflect visceral fat accumulation, whereas WC is a simple and reliable indicator of abdominal obesity. Therefore, we selected WC over BMI as a predictor in this study.

In the present study, independent variables were selected based on prior research and clinical guidelines that have consistently demonstrated their association with T2DM. Age and gender are well-established non-modifi-able risk factors reported in numerous epidemiological studies [2,15,16]. Subjective health status has been shown to predict morbidity and mortality and is increasingly rec-ognized as a proxy for overall health [17,18]. Dyslipidemia contributes to insulin resistance and elevated blood glucose levels [19]. Moderate physical activity is a protective factor that improves insulin sensitivity and reduces the risk of T2DM [20,21]. Abdominal obesity, measured by waist circumference, is a strong predictor of insulin resistance and T2DM, especially in Asian populations [12-14]. Lastly, a family history of DM reflects both genetic and environmental predispositions and has been shown to increase the risk of T2DM by two to four times [22-26]. These variables were therefore included as predictors in our model. While logistic regression provides odds ratios and statistical associations, it is often difficult for clinicians or patients to directly apply these results in practice. The nomogram complements this limitation by transforming regression coefficients into individualized risk scores and visual tools, allowing straightforward estimation of abso-lute risk. This approach enhances the clinical usability of statistical findings and facilitates evidence-based decision-making in nursing practice. Therefore, this study aimed to identify DM risk factors among middle-aged adults using data from the Korea National Health and Nutrition Examination Survey (KNHANES) and to develop a predictive nomogram. The nomogram developed in this study can be used to systematically manage DM risk, provide individualized visual educational materials, and support evi-dence-based DM prevention strategies. In nursing practice, this tool can also help community health nurses and clinical practitioners to identify high-risk individuals at an early stage, deliver tailored lifestyle counseling, and strengthen preventive nursing interventions in primary care and community-based settings.

METHODS

RESULTS

1. Socio-Demographic Characteristics of Participants According to DM Diagnosis

Among the total participants, 227 (10.3%) had DM and 1,977 (89.7%) did not. Significant differences based on DM diagnosis were observed in housing type (t=11.75, p= .001), gender (t=26.05, p<.001), age (F=18.82, p<.001), family income level (F=4.19, p =.016), education level (F=5.05, p=.007), and marital status (t=4.94, p=.028). Compared with the non-DM group, the DM group showed higher proportions of house residents (45.5%) and men (66.4%), with higher proportions in the 50∼59 years (55.6%) and 60∼64 years (27.5%) age groups. Additionally, the DM group had a higher proportion of individuals with a low family income (11.1%), high school graduates (43.4%), college graduates or higher (17.0%), and unmarried individuals (11.9%) (Table 1).

Table 1.

Socio-Demographic, Health-Related, and Health Behavior-Related Characteristics of Participants According to DM Diagnosis (N=2,204)

3. Factors Influencing DM Diagnosis

According to the multiple complex-samples logistic regression analysis, the factors influencing DM diagnosis in this study included gender, age, subjective health status, presence of dyslipidemia, moderate physical activity, abdominal obesity, and family history of DM. Specifically, men had a 2.71 times higher risk of DM than women (95% CI=1.43∼5.16). Compared with those aged 40∼49, individuals aged 50∼59 had a 4.88 times higher risk (95% CI=2.53∼9.43), and those aged 60∼64 had a 5.29 times higher risk (95% CI=2.58∼10.82). Participants with bad subjective health status had a 7.23 times higher risk (95% CI=3.86∼13.52), and those with moderate health had a 3.34 times higher risk (95% CI=1.82∼6.14), compared with those with good subjective health. Those diagnosed with dyslipidemia had a 4.55 times higher risk (95% CI=2.98∼6.95), those with abdominal obesity had a 1.56 times higher risk (95% CI=1.04∼2.34), and those with a family history of DM had a 3.15 times higher risk (95% CI=2.14∼4.64). In contrast, individuals who engaged in moderate physical activity had a 0.33 times lower risk of DM than those who did not (95% CI=0.17∼0.61) (Table 2).

Table 2.

Factors Influencing DM Diagnosis (N=2,204)

4. Nomogram for Predicting DM Risk

A nomogram was developed using variables that were significantly associated with DM. It converts the influence of each predictor into points and provides a total score that estimates the probability of the occurrence of DM. Among the predictors, subjective health status had the largest range (0∼100 points), indicating the greatest discrim-inatory power, followed by age (0∼90 points), dyslipidemia (0∼77 points), gender (0∼63 points), family history of DM (0∼61 points), moderate physical activity (0∼47 points), and abdominal obesity (0∼33 points) (Figure 1). To eval-uate how well the predictions from the nomogram aligned with actual DM cases, an ROC curve was plotted, and the AUC was calculated. The AUC value was .86, indicating high predictive accuracy (Figure 2).

Figure 1.

Nomogram for predicting diabetes mellitus risk.

Figure 2.

Predictive accuracy of the nomogram for Diabetes Mellitus.

DISCUSSION

This study aimed to identify the risk factors for DM and construct a predictive nomogram using data from the 2022 KNHANES, focusing on middle-aged adults (aged 40∼64 years). The findings revealed that 10.3% of individuals in this age group were diagnosed with DM by a physician. This prevalence is lower than the 14.8% reported by the KDA for adults aged 30 years and older in 2022[27], likely due to differences in the diagnostic criteria and the target age range. While the KDA includes individuals diagnosed by a physician, those on antidiabetic medications, and those with a fasting glucose ≥126 mg/dL or HbA1c ≥6.5%, the present study considered only physician-con-firmed diagnoses among individuals aged 40 to 64 years. Nevertheless, when prediabetes was included, the prevalence among adults increased to 41.1% [2], highlighting the urgent need for effective strategies to prevent and manage DM. Numerous DM risk prediction tools have been developed globally. In the United States, key predictors include age, gender, family history, obesity, and physical activity [28], while Chinese models emphasize waist circumference, age, and family history [29]. In Korea, a self-assess-ment model includes age, family history of DM, hypertension history, waist circumference, smoking, and alcohol use [30].

This study identified seven significant predictors of DM onset: gender, age, subjective health status, dyslipidemia, moderate physical activity, abdominal obesity, and family history of DM. These predictors were first identified through multiple logistic regression (Table 2). Statistically significant predictors were then re-entered into the final regression model, and the resulting coefficients were used to construct the nomogram (Figures 1 and 2). Therefore, the odds ratios reported in the following discussion reflect the regression results presented in Table 2, while the nomogram provides a visual translation of these predictors for individualized risk estimation. These predictors were integrated into a nomogram, with subjective health status and dyslipidemia emerging as novel risk factors. Notably, subjective health status demonstrated the strongest dis-criminative power in predicting DM, distinguishing this model from previous ones. Subjective health status, which reflects an individual's perceived overall health [17], showed a strong association with DM risk. Participants who rated their health as "moderate" had a 3.34-fold increased risk, and those with "bad" health had a 7.23-fold increased risk compared with those with "good" health. Despite its subjective nature, this variable reflects physical, mental, and social well-being and correlates well with objective health outcomes, making it a reliable proxy [17]. Nam and Nam [18] emphasized the value of subjective health status in assessing the health of older adults, particularly those with chronic conditions such as DM. Additionally, individuals with a positive health perception are more likely to engage in health-promoting behaviors such as physical activity [17], supporting the present findings.

Age was the second most influential factor, after subjective health status. Compared with those in their 40s, individuals in their 50s had a 4.88-fold increased risk, and those aged 60∼64 years had a 5.29-fold increased risk of developing DM, consistent with the findings of Kim and Kang [15]. Age-related decline in pancreatic β-cell function —at a rate of approximately 0.7% per year [31]—combined with increased visceral fat, comorbidities, medication use, and reduced physical activity, contributes to higher DM risk, establishing type 2 DM as a common disease of aging [5]. Therefore, age-specific preventive strategies, including tailored diet and exercise programs, are essential. Gender also played a significant role, with men having a 2.71-fold higher risk than women, consistent with previous studies [2,15,16]. This difference may be partly ex-plained by higher exposure to behavioral risk factors such as smoking and obesity among men [32]. In addition, postmenopausal women may experience increased risk due to hormonal changes that promote central obesity and insulin resistance. These physiological changes contribute to the higher incidence of DM and cardiovascular disease observed in older women, highlighting the need for gender- and age-specific preventive strategies and interventions. Conversely, postmenopausal women are at an increased risk of type 2 DM and cardiovascular disease because of elevated body fat [32]. Therefore, gender-specific risk as-sessments and interventions are required.

Dyslipidemia was associated with a 4.55-fold increase in the risk of DM. It promotes insulin resistance, raises blood glucose levels, and increases triglyceride synthesis [19]. According to Feingold and Grunfeld [19], dyslipidemia results from excessive caloric and fat intake and insufficient physical activity. If left untreated, it significantly increases the risk of DM and cardiovascular disease. This highlights the need for accessible interventions that promote regular exercise, smoking cessation, and the moder-ation of alcohol intake.

Moderate physical activity reduced DM risk by approximately 67%(OR=0.33). A meta-analysis demonstrated that vigorous activity, moderate activity, and walking reduced the risk of DM by approximately 39%, 32%, and 30%, respectively [20]. The World Health Organization has identified insufficient physical activity as a leading cause of non-communicable diseases, including DM [21]. In Korea, adult men show high rates of obesity and low physical activity, particularly from their 40s onwards [33], indicating an urgent need for targeted interventions. Aerobic and resistance exercises tailored to age and physical condition can improve insulin sensitivity and lipid metabolism [31]. Moreover, physical inactivity contributes to obesity, ex-acerbating insulin resistance and β-cell dysfunction [34]. Therefore, the significance of physical activity in preventing DM must be emphasized. Abdominal obesity increased the risk of DM by 1.56-fold. Central obesity and waist circumference are well-established predictors of DM, and a 1 kg weight loss is associated with a 16% reduction in the risk [5]. Seiglie et al. [13] and Yun et al. [14] con-firmed that abdominal obesity increases the risk of impaired fasting glucose and type 2 DM. Seiglie et al. [13] also found that obesity increased the risk of DM by 1.54 to 2.64 times depending on income level, underscoring the importance of managing abdominal obesity.

A family history of DM increased the risk of developing the disease by 3.15-fold. Even after controlling for con-founders, such as age, gender, BMI, and lifestyle factors, family history remained a significant predictor of prediabetes [22]. Meigs et al. [23] also emphasized its influence. Studies using the KNHANES data have found that family history affects insulin secretion, and when combined with obesity, it significantly increases the risk of early onset DM [30,32]. Both genetic and environmental components play a role; monozygotic twins have an approximately 60% life-time concordance rate, and individuals with a first-degree relative with DM have a 2∼4 times higher risk of hyper-glycemia [25,26]. Active management of modifiable risk factors is essential, especially in patients with a family history.

In the regression analysis, moderate depression was significantly associated with DM risk, whereas mild and severe depression were not. Because nomograms require consistent scaling across categories to generate meaningful risk scores, including depression would have in-troduced instability and potential misinterpretation in the visual risk assessment. Therefore, depression was ex-cluded from the final predictive model, and only predictors that demonstrated consistent and independent associations with DM were incorporated into the nomogram.

Since 2007, Korea has implemented a national registration and management program for hypertension and DM to identify and manage adults aged 30 years and over and reduce complications and healthcare costs. However, par-ticipation among those aged 30∼64 years remains sub-optimal [35]. To enhance engagement, tailored strategies that address barriers such as limited awareness, low moti-vation, and time constraints are needed. This study is meaningful because it utilized a nationally representative data set to identify DM-related risk factors and construct a nomogram to predict the risk of DM among middle-aged adults. These findings provide a scientific basis for the development of targeted strategies and health interventions. However, given the cross-sectional nature of the KNHANES, causality between variables cannot be determined. Fur-thermore, this study did not include genetic or environmental factors. Future research should incorporate these variables to identify the risk factors for DM and prediabetes more comprehensively.

CONCLUSION

In conclusion, this study identified seven significant predictors of DM among middle-aged Korean adults (gender, age, subjective health status, dyslipidemia, physical activity, abdominal obesity, and family history) and developed a nomogram based on nationally representative data to estimate the individual DM risk. Among these, subjective health status emerged as the strongest predictor, highlighting the importance of perceived health for disease prevention and suggesting its potential use as a simple early screening tool in nursing practice. Individu-als with poor or fair health perceptions can be prioritized for tailored lifestyle counseling, structured exercise and diet programs, and motivational interviewing, and community health screenings could integrate subjective health assessment to facilitate early preventive interventions. These findings emphasize the need for age- and gender-specific strategies and behavioral interventions, particularly those targeting individuals with poor lifestyle hab-its or genetic predispositions. This study offers valuable insights into individualized DM risk management in Korea.

Article information Continued