This study is a cross-sectional analysis and has been reported in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines (https://www.strobe-statement.org/).

KNHANES is a nationwide health and nutrition survey conducted since 1998 under the National Health Promotion Act [28]. This large-scale cross-sectional study was designed to assess the health status, health behaviors, and food and nutritional intakes of a non-institutionalized civilian population in South Korea. The KNHANES data collection process encompasses household surveys, health questionnaires, physical examinations, and nutritional surveys. The survey was conducted in various phases, including the first (1998), second (2001), third (2005), fourth (2007–2009), fifth (2010–2012), sixth (2013–2015), seventh (2016–2018), and eighth phases (2019–2021) [28]. This study employed survey data from 2008 to 2020 that encompassed HRQoL assessments using the Euro Quality of Life-5 Dimensions (EQ-5D) instrument [28].

From a total of 108,497 participants in the KNHANES between 2008 and 2020, the following were excluded from the analysis: 1) individuals under 65 years of age (n = 87,549); 2) those with missing relevance analysis weights (n = 2,357); 3) individuals who were not diagnosed with or were unaware of having T2DM (n = 14,892); and 4) respondents who did not complete the EQ-5D questionnaire (n = 146). Consequently, 3,553 individuals were included in the primary analysis (Fig. 1). Participants whose blood glucose levels met the diagnostic criteria for diabetes but who had not been diagnosed or were unaware of their condition were excluded, as the study aimed to assess the quality of life in individuals who were both diagnosed and aware of having T2DM, potentially leading to lifestyle changes.

All data included in this study were collected with informed consent from participants and this study was approved by the Ethics Committee of the KDCA (approval numbers: 2008-04EXP-01-C, 2009-01CON-03-2C, 2010-02CON-21-C, 2011-02CON-06-C, 2012-01EXP-01-2C, 2013-07CON-03-4C, 2013-12EXP-03-5C, 2018-01-03-P-A, 2018-01-03-C-A, 2018-01-03-2C-A, and 2018-01-03-3C-A). According to Article 2, Paragraph 1 of the Bioethics Law, and Paragraph 1 of Article 2 of the Enforcement Rule of the same law, the KNHANES is classified as research conducted directly by the state for public welfare. Consequently, from 2015 to 2017, the study was conducted without a separate ethical review as per the advice of the Ethics Committee of the KDCA [28].

Data on age, sex, education level, monthly household income, employment status, and household composition were obtained through interviews with trained investigators. Data on body mass index (BMI), physical activity, smoking status, alcohol consumption, type of diabetes treatment, and duration of diabetes were collected using self-reported health questionnaires [28]. Educational level was categorized as less than elementary school, middle school graduate, high school graduate, or above. Income levels were analyzed based on equivalized household income, which was calculated by dividing the average monthly household income by the number of household members. It was then categorized into quartiles as low, mid-low, mid-high, and high. Economic activity was classified as employed, unemployed, or economically inactive, and household composition was classified as living alone or with a spouse and/or other relatives. BMI, calculated using measurements taken by trained personnel, was classified per World Health Organization Asia/Pacific obesity criteria into underweight (< 18.5 kg/m²), normal (18.5–22.9 kg/m²), overweight (23–24.9 kg/m²), and obese (≥ 25 kg/m²) [29]. Owing to the low number of underweight individuals, underweight and normal individuals were combined for analysis into underweight/normal, overweight, and obese categories. For physical activity, the metabolic equivalent task-hours per week (METs-h/week) were calculated, with weights assigned based on the intensity of each exercise [30]. Smoking status was divided into nonsmokers, former smokers, and current smokers. Alcohol consumption was quantified by multiplying the frequency and quantity per occasion over the past year to determine the daily intake (servings/day), classifying the participants as non-drinkers or drinkers. Diabetes treatments were categorized as no treatment, oral hypoglycemic agents and/or insulin therapy, diet and/or exercise therapy alone, or in combination with medication. Duration of diabetes was calculated in years from the survey time and physician diagnosis, categorizing participants into < 5 years, 5–9 years, and ≥ 10 years, based on their distribution.

The dietary survey was conducted through face-to-face interviews with trained investigators who visited the participants’ homes. Information on the food consumed on the previous day was collected using the 24-hour recall method [28]. The study analyzed the estimated intake levels of nutrients, including energy, carbohydrates, proteins, fats, vitamins A, B₁, B₂, C, niacin, calcium, phosphorus, iron, sodium, potassium and fiber. The acceptable macronutrient distribution ranges (AMDR) was based on the 2020 Korean Dietary Reference Intakes [31]. Each nutrient’s proportion of total energy was classified into three categories: less than, acceptable (carbohydrates: 55%–65%, protein: 7%–20%, fat: 15%–30%), and more than the recommended range.

In this study, the participants were defined as patients with T2DM who were aware of their condition. They were identified based on 1) self-reported health questionnaires in which they responded to having been diagnosed by a doctor, 2) currently suffering from T2DM, and 3) undergoing treatment with oral hypoglycemic agents, insulin therapy, or diet/exercise regimens.

To elucidate the association between comorbid chronic diseases and HRQoL in patients with T2DM, this study explicitly defined several chronic diseases as comorbidities. Hypertension, dyslipidemia, stroke, myocardial infarction/angina, renal failure, and cancer were included as comorbidities [32]. Cancer was approached more specifically and included patients diagnosed with any of the following: gastric, liver, colorectal, breast, uterine, lung, or bronchial cancers. Each comorbid disease included in the survey was defined based on 1) having been diagnosed by a doctor, 2) suffering from the disease in the past year or currently, or 3) currently receiving treatment. Based on the number of accompanying diseases, patients were categorized as having 0, 1, 2, or more than three comorbidities.

In this study, HRQoL in older patients with T2DM was analyzed based on the level of comorbidities and dietary patterns using the EQ-5D, an HRQoL measurement tool provided by KNHANES [28]. The EQ-5D is an index that subjectively evaluates HRQoL and consists of five dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression [33]. Each question can be answered at three levels (1: no problem, 2: somewhat problematic, and 3: serious problem), and through this, a total of 3⁵ = 243 unique health states can be expressed [33]. The maximum score for the EQ-5D is 1, indicating the best possible quality of life, while the minimum score is –0.171, which indicates the worst possible quality of life.

The composite EQ-5D index was calculated using quality-of-life weights provided by the KDCA [34]. For ease of interpretation, responses for each item of EQ-5D in this study were reclassified into the following categories: 0 for 'extreme problems', 0.5 for 'some problems', and 1 for 'no problems', and the analysis was conducted accordingly. In other words, the higher the EQ-5D score, the better the quality of life, while a lower score indicates a poorer quality of life.

Statistical analyses were conducted considering the weights, strata, and primary sampling units of the complex sample design of the KNHANES. To analyze the general characteristics and lifestyle habits of older patients with T2DM, categorical variables were presented as frequencies and percentages, and significance was tested using the chi-squared test. Analysis of the mean EQ-5D scores and calorie nutrient intake based on the number of comorbidities was conducted using multivariate linear regression, presenting the means and standard errors. To compare the adjusted means between groups, analysis of covariance and Tukey post-hoc test were conducted. Confounding variables were selected based on a review of previous studies and preliminary analyses [35-38]. The included variables were age, sex, education level, monthly household income, smoking status, alcohol consumption, physical activity level, BMI, and duration of diabetes. For the analysis of nutrient intake by comorbidity level, the adjusted mean was calculated considering age, sex, and total energy intake. P for trends were determined using linear regression analysis with the median values of each variable.

All statistical processing in this study was conducted using SAS 9.4 version (statistical analysis system; SAS Institute), and statistical significance was tested at the α = 0.05 level.

The general characteristics and lifestyle factors of the participants were compared and analyzed according to the number of comorbidities (Table 1). As the number of comorbidities increased, the proportion of participants engaged in economic activities decreased (P < 0.001), and the proportion of those who were overweight tended to increase (P < 0.001). Additionally, as the number of comorbidities increased, physical activity levels, measured in METs, significantly decreased (P < 0.001), and a significantly lower proportion of participants reported not currently receiving treatment for diabetes (P = 0.002).

Table 2 presents the distribution of responses across each level within the EQ-5D dimensions according to the number of comorbidities. In all dimensions, the highest proportion of participants reported ‘no problems,’ followed by ‘some problems’ and ‘extreme problems.’ As the number of comorbidities increased, there was a tendency for the proportion of participants reporting ‘no problems’ in all dimensions to decrease, while the proportion reporting ‘some problems’ and ‘extreme problems’ tended to increase (all, P <0.05).

Table 3 presents the mean scores of the EQ-5D according to the level of comorbidities among the participants. After adjusting for age, sex, education level, monthly household income, smoking status, alcohol consumption, physical activity level, BMI, and duration of diabetes, a clear trend was observed; As the number of comorbidities increased, scores across all dimensions of the EQ-5D, which are rated on a scale from 1 (no problems) to 0 (extreme problems), tended to decrease. Significant differences were observed in mobility (P = 0.009), with participants having ≥ 3 comorbidities showing lower mobility scores compared to those with 1 comorbidity. No significant differences were found in self-care across different levels of comorbidities (P = 0.047). For usual activity, significant differences were observed (P < 0.001), with lower scores for participants with ≥ 3 comorbidities compared to those with fewer comorbidities. Pain/discomfort levels also showed significant differences (P < 0.001), with participants having 2 or ≥ 3 comorbidities reporting lower scores (indicating more problems) compared to those with 1 comorbidity. Anxiety/depression scores were significantly lower (indicating more problems) in participants with ≥ 3 comorbidities compared to those with 0 comorbidities (P = 0.004). Finally, the EQ-5D index, where a score of 1 indicates no problems and 0 indicates extreme problems in any dimension, showed significant differences (P < 0.001), with lower overall scores in participants with more comorbidities.

The table also includes P for trend values, which indicate a consistent decline in EQ-5D levels as the number of comorbidities increases. Mobility (P for trend = 0.003), usual activity (P for trend < 0.001), pain/discomfort (P for trend < 0.001), anxiety/depression (P for trend < 0.001), and the overall EQ-5D index (P for trend < 0.001) all showed significant decreasing trends. Self-care, while not significantly different between specific groups, also demonstrated a significant overall trend (P for trend = 0.020), indicating a general decline in self-care ability with more comorbidities.

Table 4 presents the average nutrient intakes levels of participants, according to the number of comorbidities. Significant differences were observed in several nutrient intakes. Participants with three or more comorbidities had a significantly lower carbohydrate intake (P = 0.009) compared to those with zero or one comorbidity, while fat intake was significantly higher in participants with three or more comorbidities (P < 0.001) compared to those with no comorbidities. Participants with two comorbidities had significantly higher fiber intake compared to those with zero or one comorbidity (P < 0.001), while the fiber intake of participants with three or more comorbidities was intermediate. Additionally, participants with two or more comorbidities had significantly higher vitamin B₂ intake (P < 0.001), and those with three or more comorbidities had significantly lower sodium intake compared to those with no comorbidities (P = 0.008), based on Tukey post-hoc test results.

Table 5 presents the distribution of participants consuming less than, within, or more than the AMDR for carbohydrates, proteins, and fats, according to the number of comorbidities. In all groups, the majority of participants consumed carbohydrates at levels exceeding the AMDR (> 65%). Protein intake was mostly within the AMDR (7%–20%) across all groups. For fat, the highest percentage of participants consumed less than the AMDR (< 15%), but fat intake distribution differed significantly based on the number of comorbidities (P < 0.001), with a higher proportion of participants with more comorbidities consuming fat within or above the AMDR.

Nevertheless, our study is not without limitations. First, while confounding factors were progressively adjusted based on a review of previous literature and a preliminary analysis comparing the quality of life among older patients with T2DM with different numbers of comorbidities, there remains the possibility of residual confounding due to unmeasured or unknown potential confounders. For example, the severity of T2DM, treatment type, and management efficacy were not adjusted for in the present study. This oversight might lead to residual confounding as these factors can affect the HRQoL of patients. Additionally, while we considered analyzing the type of comorbidities, the small sample size in subgroups with multiple overlapping conditions made it difficult to conduct robust statistical analyses. As a result, we focused on the number of comorbidities rather than specific types, which could limit the depth of our analysis regarding how different comorbidity types impact HRQoL. Second, as the study analyzed the KNHANES data from to 2008–2020, its cross-sectional nature limited its ability to establish causal relationships between causes and outcomes. For instance, it is challenging to prove a direct causal relationship between higher number of comorbidities in older patients with T2DM and lower HRQoL scores observed in the study. Various intermediary factors between comorbidity levels and low HRQoL may be involved; however, these factors may not completely account for the observed relationship. Therefore, the study results provide preliminary insights into the association between comorbidities and HRQoL. Finally, the study relied on self-reported data, which may have been subject to subjective biases. In particular, there could be errors in information such as dietary records, physical activity levels, and medical history.

Our study demonstrates a multifaceted relationship between comorbidities, diet, and HRQoL in Korean patients over 65 years of age with T2DM. We observed that an increase in the number of comorbidities correlated with a decline in HRQoL. Notably, different patterns in nutrient intake, such as reduced carbohydrate consumption and increased fat intake, were associated with varying comorbidity levels. These findings underscore the need for national health policies and support systems that focus on both medical treatments and nutritional care. Further large-scale prospective cohort studies and clinical trials are essential to deepen our understanding of these relationships and develop comprehensive management strategies to improve the quality of life for older diabetic patients in Korea.