All-cause Mortality and Incidence of Cardiovascular Diseases in Lean Patients With Newly Diagnosed Type 2 Diabetes

Abstract

Context

A few studies have evaluated all-cause mortality or risk of cardiovascular diseases according to the body mass index (BMI) level in patients with type 2 diabetes.

Objective

We evaluated all-cause mortality and the incidence of cardiovascular diseases in lean patients with newly diagnosed type 2 diabetes in Korea. Additionally, we aimed to determine a difference in all-cause mortality and the incidence of cardiovascular diseases according to changes in BMI over time among patients with newly diagnosed type 2 diabetes in Korea.

Methods

We analyzed 419 509 patients with newly diagnosed type 2 diabetes who underwent health screening between 2010 and 2014 and followed up until 2019. We conducted a multivariate Cox proportional hazards model to determine the association between BMI and all-cause mortality or risk of cardiovascular diseases.

Results

Lean patients with type 2 diabetes had a higher risk of all-cause mortality [hazard ratio (HR): 2.106, 95% confidence interval (CI): 1.974-2.248], cardiovascular disease (HR: 1.132, 95% CI: 1.078-1.189), coronary heart disease (HR: 1.219, 95% CI: 1.124-1.323), heart failure (HR: 1.405, 95% CI: 1.279-1.543), stroke (HR: 1.155, 95% CI: 1.024-1.302), and ischemic stroke (HR: 1.205, 95% CI: 1.045-1.388) compared to patients with type 2 diabetes and normal BMI. Patients with newly diagnosed type 2 diabetes had the highest all-cause mortality when they remained lean during the follow-up.

Conclusion

Our findings underscore the critical role of maintaining an appropriate weight status to reduce all-cause mortality and incidence of cardiovascular diseases among lean patients with newly diagnosed type 2 diabetes.

The prevalence of diabetes mellitus (DM) is increasing worldwide, and patients with DM have a higher risk of all-cause and cardiovascular mortality and cardiovascular disease (CVD) compared to people without DM (1). Type 2 diabetes mellitus (T2DM) is the most prevalent form of DM, with about 90% of patients with DM classified as having T2DM (2). Because T2DM is a complex disorder, T2DM is considerably heterogeneous among patients (3). Recently, a refined classification of DM has been suggested, with new clustering of patients with adult-onset T2DM being able to identify patients at high risk of diabetic complications at diagnosis and provide information about underlying disease mechanisms using cluster analysis of 6 commonly measured variables (4). Although overweight and obesity are well-recognized risk factors of T2DM, T2DM can also develop in lean subjects, especially among Asian populations, who develop T2DM at a lower body mass index (BMI) than Europeans (5).

Several studies evaluated the features of lean patients with T2DM. These patients had a male predominance, higher prevalence of insulin use, lower levels of triglycerides (TG)/high-density lipoprotein cholesterol (HDL-C), and higher prevalence of alcoholism and pancreatitis compared to patients with T2DM and obesity in the United States (6). The low BMI group (BMI ≤ 19 kg/m²) with DM had lower insulin secretory response and endogenous glucose production compared with patients with T2DM among South Indian men using hyperinsulinemic-euglycemic pancreatic clamp (7). The male sex, older age, lower education, hypertension, and hypertriglyceridemia were associated with a higher risk of nonobese DM in China (8).

Previous studies presented the results of chronic complications of lean patients with T2DM. The prevalence of microvascular complications was higher than macrovascular complications in lean Indian patients with T2DM. The prevalence of microvascular complications among lean Indians with T2DM was related to the duration of DM and glycemic control, and secondary oral hypoglycemic agent failure was observed in 27% of lean Indians with T2DM (9). Being underweight was significantly associated with an increased risk of sudden cardiac death in the 2 602 577 Korean population with DM using the Korean National Health Insurance Service (NHIS). The underweight group showed a 2.4-fold increased risk of sudden cardiac death, and patients with DM and BMI of 25 to 30 kg/m² showed the lowest risk of sudden cardiac death (10).

However, a few studies have evaluated all-cause mortality and the incidence of cardiovascular events according to changes in weight status in patients with T2DM. Furthermore, only a few studies have assessed all-cause mortality and the incidence of cardiovascular events in lean patients with newly diagnosed T2DM, as well as the characteristics of lean patients with T2DM who develop CVD in Korea. We aimed to analyze all-cause mortality and the incidence of cardiovascular events of lean patients with newly diagnosed T2DM and to evaluate the features of such patients with CVD in Korea. Additionally, we aimed to determine a difference in all-cause mortality and the incidence of cardiovascular events according to changes in weight status over time among lean patients with newly diagnosed T2DM in Korea.

Materials and Methods

Study Design

We used a customized NHIS database (NHIS-2023-2-065). In South Korea, subscribing to health insurance is mandatory, and NHIS data contain information from almost the entire national population (11). These data include information regarding demographics, medical use, diagnosis, medications, health checkups, death, etc. (12). We included patients with newly diagnosed type 2 diabetes between 2010 and 2014. We considered the time when these patients first underwent health screening between 2010 and 2014 as the index date and followed them until 2019.

People with a diagnosis of DM or a history of CVD before the index date were excluded. In this study, patients were defined as having T2DM if any of the 3 conditions were met. First, we defined the T2DM (E11-E14) patients based on the 10th revision of the International Classification of Diseases (ICD-10). Second, we considered patients with fasting blood glucose (FBG) levels ≥126 mg/dL as patients with T2DM. Third, we also considered people who were prescribed oral glucose-lowering medications or insulin as patients with T2DM. Subjects with ICD-10 code E10 (type 1 DM) were excluded. The following exclusion criteria were considered for initially identified 852 496 patients. First, we excluded subjects without information about sex, age, location of residence, and income level (n = 15 612). Second, we excluded subjects without health checkup data (n = 417 375).

In this study, we aimed to evaluate all-cause mortality and the incidence of cardiovascular events in lean patients with newly diagnosed T2DM. Therefore, we divided BMI groups into 4 categories: lean (<18.5 kg/m²), normal (18.5-23 kg/m²), overweight (23-25 kg/m²), and obese (≥25 kg/m²).

Health Outcomes

The main health outcomes were all-cause death and CVD incidence. For CVD events, we used the first hospitalization based on ICD-10. We further categorized CVD into coronary heart disease (CHD; ICD-10: I20-I25), heart failure (HF; ICD-10: I50), stroke (ICD-10: I60-I64), hemorrhagic stroke (HS; ICD-10: I60-I62), and ischemic stroke (IS; ICD-10: I63). All health outcome variables are dichotomous for “non-event” or “event.” In the case of CVD incidence, if the patient died during the follow-up, it was censored.

Covariates

For covariates, we used individual data for the baseline. We extracted age (years), sex (men vs women), income level (low vs high), and location of residence (urban vs rural) from the demographic information NHIS database. The urban areas included 8 major cities (Seoul, Busan, Daegu, Incheon, Gwangju, Daejeon, Ulsan, and Sejong). The rural areas included 9 provinces (Gyeonggi, Gangwon, North Cungcheong, South Cungcheong, North Jeolla, South Jeolla, North Gyeongsang, South Gyeongsang, and Jeju).

We also extracted medication use information from the procedure NHIS database. The medication used information in 3 categories: antihypertensives, oral glucose-lowering medications or insulin, and lipid-lowering agents.

From the health checkup information, smoking status (nonsmoking vs smoking), alcohol consumption (non-heavy drinking vs heavy drinking), physical activity (none vs 1 time/week), systolic blood pressure (SBP) and diastolic blood pressure (DBP; mmHg), FBG (mg/dL), total cholesterol (mg/dL), TG (mg/dL), HDL-C (mg/dL), and low-density lipoprotein cholesterol (LDL-C; mg/dL) values were obtained.

Statistical Analysis

We utilized the Cox proportional hazard model. We calculated the time from the baseline (2010-2014) to the end of the follow-up (December 31, 2019). The event variable was all-cause death or CVD incidence. We considered both continuous and categorical variables for BMI to evaluate the association of all-cause death or CVD incidence with weight status. Then we constructed 2 models: an unadjusted model and a model adjusted for age, sex, location of residence, income level, smoking status, alcohol consumption, physical activity, use of medication, SBP, DBP, FBG, total cholesterol, TG, HDL-C, and LDL-C. Hazard ratios (HRs) and 95% CIs for BMI groups are presented with normal reference levels (18.5-23 kg/m²).

A stratified analysis was conducted to evaluate sex, age group (<65 vs ≥65), and use of medication differences in the effect of BMI on all-cause mortality or CVD incidence. All data analyses were conducted with SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) and R version 4.2.1. A P < .05 indicated statistical significance.

Sensitivity Analysis

We conducted 2 sensitivity analyses. First, to evaluate all-cause mortality and CVD incidence according to changes in weight status by BMI, we defined BMI changes divided into 4 categories from baseline to the endpoint as (1) lean to lean, (2) lean to non-lean, (3) non-lean to lean, and (4) non-lean to non-lean (reference level). We set the reference level for the BMI of non-lean patients at baseline and the endpoint. Second, we additionally excluded patients with chronic conditions, such as cancer (ICD-10: C00-C97) or chronic obstructive pulmonary disease (ICD-10: J44), within 1 year before the baseline (n = 19 098) and then analyzed the association between BMI group and all-cause mortality or CVD incidence.

Statement of Ethics

This study was exempted from review by the Institutional Review Board of the Ewha Womans University Seoul Hospital, South Korea (IRB no. SEUMC 2023-03-022).

Results

A total of 419 509 patients with newly diagnosed T2DM were included (Supplementary Fig. S1) (13). Of all patients, 54.5% were men. The mean age of included patients was 48.5 years (SD = 11.7) (Table 1). Among all patients, 14 518 (3.5%) were lean, 149 110 (35.5%) had normal BMI, 101 521 (24.2%) were overweight, and 154 360 (36.8%) were obese. A total of 5.4% of patients used medications, 39.9% of patients were current smokers, 18.5% of patients were heavy drinkers, and 55.7% of patients reported no physical activity. As shown in Table 1, the proportion of lean patients with newly diagnosed T2DM was higher in women (62.8%) than in men (37.2%). Lean patients with newly diagnosed T2DM had a lower mean age (46.7 years) than other BMI groups. In men, the mean age (53.1 years) of lean patients with T2DM was the highest compared to other BMI groups, while in women, the mean age (42.9 years) of lean patients with newly diagnosed T2DM was the lowest compared to other BMI groups (Supplementary Table S1) (13). Lean patients with newly diagnosed T2DM exhibited the lowest levels of total cholesterol, TG, FBG, SBP, and DBP at baseline and the highest level of HDL-C (Table 1).

In our study, we observed patients for 2 867 486 person-years with a median follow-up of 6.8 years for all-cause deaths and 2 751 283 person-years with a median follow-up of 6.6 years for CVD incidence. During the follow-up, 11 272 patients (2.7%) with T2DM died, and 51 388 CVD cases (12.2%) occurred. Among CVD subtypes, CHD (n = 17 364) was the most common, while the incidence of HS (n = 1885) was the lowest (Table 2). Supplementary Tables S2 to S8 (13) present the baseline characteristics according to lean or non-lean patients for each health outcome (all-cause deaths or CVD). For all-cause deaths, baseline characteristics of lean patients with T2DM included a higher proportion of men (76.6%), older adults (>50 years), and greater use of medication (event: 23.4% vs non-event: 4.8%) compared to the non-event group.

Table 3 and Supplementary Figs. S2 to S8 (13) show HRs for all-cause mortality and CVD incidence among patients with T2DM according to BMI. Overall, lean patients with newly diagnosed T2DM had a higher risk of all-cause mortality, CVD, CHD, HF, stroke, and IS compared to patients with T2DM and normal BMI (all-cause mortality: 2.106, 95% CI: 1.974-2.248; CVD: 1.132, 95% CI: 1.078-1.189; CHD: 1.219, 95% CI: 1.124-1.323; HF: 1.405, 95% CI: 1.279-1.543; stroke: 1.155, 95% CI: 1.024-1.302; IS: 1.205, 95% CI: 1.045-1.388). However, there was no statistically significant difference in the HS incidence (HR: 1.085, 95% CI: 0.865-1.362) between lean patients with newly diagnosed T2DM and those with normal BMI.

Supplementary Table S9 (13) demonstrates the results of stratified analysis by sex. We found a variation in effect between weight status and all-cause mortality or CVD risk according to sex. Among both men and women, lean patients with newly diagnosed T2DM had a higher risk of all-cause mortality compared to patients with T2DM and normal BMI. The HRs for women (HR: 2.298, 95% CI: 2.009-2.629) were slightly higher than those for men (HR: 2.032, 95% CI: 1.886-2.188), whereas the HR for CVD (men: 1.222, 95% CI: 1.144-1.304; women: 1.123, 95% CI: 1.063-1.187), CHD (men: 1.320, 95% CI: 1.192-1.461; women: 1.073, 95% CI: 0.938-1.229), HF (men: 1.460, 95% CI: 1.300-1.641; women: 1.311, 95% CI: 1.119-1.536), stroke (men: 1.194, 95% CI: 1.031-1.382; women: 1.081, 95% CI: 0.876-1.334), and IS incidence (men: 1.251, 95% CI: 1.057-1.481; women: 1.104, 95% CI: 0.850-1.435) was higher among lean patients with T2DM in men compared to women (Supplementary Table S9) (13).

We found no difference in effectiveness between age-specific body weight status and all-cause mortality or CVD risk, except for IS particularly (Supplementary Table S10) (13). Lean patients with T2DM in younger age groups (<65) had a higher risk of all-cause mortality and CVD incidence compared to normal BMI, except for HS incidence.

Supplementary Table S11 (13) demonstrates the results of stratified analysis by use of medication (no vs yes). We found a variation in effect between weight status and all-cause mortality (No: 2.227, 95% CI: 2.068-2.399; Yes: 1.768, 95% CI: 1.545-2.023). However, no differences were found regarding cardiovascular incidence.

Table 4 and Supplementary Tables S12 to S18 (13) show the association of the risk of all-cause death and CVD events according to changes in weight status among patients with T2DM. The HR was higher for patients who remained lean (or changed from non-lean to lean) during the follow-up compared to non-lean patients with T2DM. The HRs and 95% CIs for all-cause mortality and cardiovascular events among patients who remained underweight compared to the reference level were as follows: 4.825 (95% CI: 4.320-5.389) for all-cause mortality, 1.464 (95% CI: 1.353-1.584) for total CVD, 1.770 (95% CI: 1.563-2.004) for CHD, 2.150 (95% CI: 1.846-2.504) for HF, 1.849 (95% CI: 1.522-2.247) for stroke, 1.784 (95% CI: 1.202-2.647) for HS, and 1.984 (95% CI: 1.581-2.490) for IS.

The sensitivity analysis excluding patients with a history of cancer and chronic obstructive pulmonary disease within 1 year before the baseline showed results similar to the main findings (Supplementary Table S19) (13).

Discussion

Main Results

In our study, the incidence of lean T2DM was 3.5% among patients with newly diagnosed T2DM. Lean patients with newly diagnosed T2DM were associated with increased all-cause mortality and a higher incidence of fatal CVD, CHD, HF, and stroke. Especially, the incidence of fatal CHD and stroke was higher in lean patients with newly diagnosed T2DM compared to patients with normal BMI among men with T2DM but not among women with T2DM. The mortality rate was the highest when the lean state was maintained among patients with newly diagnosed T2DM during the follow-up.

Previous Studies

T2DM prevalence based on BMI varies depending on ethnicity. From the representative samples of 4906 Asian Indians in South Asia and 2868 Whites in the United States, Asian Indians exhibited a higher prevalence of T2DM at lower levels of BMI than Whites. T2DM prevalence was 5.8% in underweight, 38.5% in normal weight, 38.2% in overweight, and 17.6% in obese Asian Indians. T2DM prevalence was 1.2% in underweight, 30.4% in normal weight, 34.0% in overweight, and 34.3% in obese Whites (14). Nonobese people with DM accounted for 60.8% and 62.0% of all the subjects with DM in Japanese men and women, respectively, from a voluntary health checkup program between 1998 and 2006 (15). In our study, the incidence of lean T2DM was 3.5% among patients with newly diagnosed T2DM from 2010 to 2014.

Several previous studies have investigated the association between weight status and mortality in patients with DM. Patients with normal weight at the time of incident diabetes had higher total and cardiovascular mortality than those who were overweight or obese (16). In another study, there was a J-shape association between BMI and all-cause mortality among participants with incident diabetes and no evidence of lower mortality among patients with incident diabetes who were overweight or obese at diagnosis compared to those with normal weight after excluding patients who were underweight (BMI <18.5 kg/m²) (17). Lean patients with diabetes in Germany were characterized by male predominance, smoking, and alcohol consumption and were associated with an increased risk of death (18). The risk for mortality was the highest among lean patients with newly diagnosed T2DM in South Korea (19). Furthermore, patients with a BMI <22 kg/m² and a BMI of 22 to 25 kg/m² had higher all-cause mortality compared to patients with a BMI of 30 to 35 kg/m² at baseline among patients from the PROspective pioglitazone Clinical Trial In macroVascular Events (PROactive) trial population with T2DM and pre-existing CVD (20). Consistently, in our study, lean patients with T2DM had an increased risk of all-cause mortality.

Sex Differences

In our study, there were sex differences in the incidence of cardiovascular events among lean patients with newly diagnosed T2DM. Lean men with newly diagnosed T2DM had a higher incidence of fatal CVD, CHD, HF, and stroke compared to men with newly diagnosed T2DM and normal BMI. On the other hand, lean women with newly diagnosed T2DM had a higher incidence of fatal HF compared to women with newly diagnosed T2DM and normal BMI, but the incidence of fatal CHD or stroke did not differ between lean women with newly diagnosed T2DM and those with normal BMI. Lean men with newly diagnosed T2DM were older than men with newly diagnosed T2DM and normal BMI. However, lean women with T2DM were younger than women with T2DM and normal BMI. The large number of elderly patients among lean men with T2DM might have caused the high CVD incidence. Lean patients with T2DM who died were older and mainly male. Some studies investigated sex differences in the association between diabetes and CVD risk; however, a significant heterogeneity existed between these studies (21). Female patients with diabetes had a greater mortality risk than male patients with diabetes in Canada (22). Since results may differ depending on study design or participation criteria, additional research on the effect of sex on CVD incidence or mortality in lean patients with T2DM is needed.

All-cause Mortality and Incidence of CVDs and BMI Changes in Patients With T2DM

Several studies investigated the association between weight change and all-cause mortality among patients with DM. In patients with T2DM from the PROactive trial population and pre-existing CVD, weight loss was associated with increased total and cardiovascular mortality and the composite of myocardial infarction and stroke (20). In patients with newly diagnosed T2DM who underwent health examination using the Korean NHIS, weight loss ≥−10% during the first 2 years after diabetes diagnosis had the highest HR for all-cause mortality (HR: 1.86, 95% CI: 1.61-2.14), while the HR for weight gain ≥10% was 1.61 (95% CI: 1.37-1.89). A weight gain ≥10% during the first 2 years after diabetes diagnosis was associated with an increased risk of stroke in patients with newly diagnosed T2DM (19). The severity of being underweight was associated with a higher risk of stroke, myocardial infarction, and all-cause mortality in the general population using data from the Korean NHIS (23). Diabetic patients who maintained an underweight status exhibited the highest risk of HF using the data from the NHIS and the Korean National Health Screening database from 2009 to 2012 (24). In our study, when patients were lean at diabetes diagnosis but later changed to a non-lean state, the all-cause mortality rate did not increase compared to patients who maintained a non-lean state during their disease. Those who were lean at diabetes diagnosis but remained lean thereafter had the highest mortality rate. Various factors, such as sarcopenia and poor nutritional status, were suggested as mechanisms of the increased CVD risk in the underweight group (25). In addition, BMI was an inverse predictor of N-terminal pro–B-type natriuretic peptide in a large individual heart failure patient dataset (26). The responsible mechanism is unclear, but the decreased muscle mass or latent undiagnosed disease might influence the increased mortality in lean patients with T2DM.

Strengths

This is the first study to analyze the all-cause mortality and incidence of CVD in lean Korean patients with newly diagnosed T2DM with a large sample size, allowing for the generalization of study results to lean patients with newly diagnosed T2DM in Korea. We adjusted multiple covariates that can influence the development of CVD, such as age, sex, location of residence, income level, smoking status, alcohol consumption, physical activity, use of medication, SBP, DBP, FBG, total cholesterol, TG, HDL-C, and LDL-C. Our findings demonstrated the robustness in 2 sensitivity analyses (BMI change and excluding comorbidities).

Limitations

There are several limitations in our study. First, lean individuals may develop T2DM through a different mechanism compared to non-lean individuals. We cannot evaluate the underlying mechanisms of different all-cause mortality and CVD risks between lean and non-lean patients with T2DM because our study had a retrospective design. Second, we did not evaluate waist circumference or body fat distribution, which could affect the development of CVD in patients with T2DM. Third, we did not evaluate dietary patterns, which could influence the differences in T2DM manifestation. Fourth, we did not consider pancreatic beta cell dysfunction and insulin resistance, although it is the key mechanism in T2DM development (27). Underweight patients with T2DM had lower fasting insulin and poorer insulin secretion compared to overweight or obese people with T2DM among Asian Indians and Whites (13). Fifth, we did not evaluate the proportion of insulin use in lean patients with T2DM. Additionally, therapeutic responses in lean patients with T2DM were not considered. Sixth, we did not evaluate the severity of T2DM, such as hemoglobin A1c, or the degree of weight change. Although we adjusted various covariates, there is a possibility of undetected confounders. Seventh, in this study, we defined CVD using ICD-10 codes. However, inaccuracies in the diagnostic codes might exist due to omissions, coding errors, or misclassification in the process of defining the outcome variables using insurance claim data. This could have introduced some level of imprecision into the study results. Nevertheless, despite the possibility of some inaccuracies (11), we anticipate that they would not significantly influence study findings. Lastly, we did not analyze the specific cause of death. Complications of DM, including CVD, renal diseases, retinopathy, and neurological complications, might be associated with mortality (28). Therefore, future research should include an analysis of the cause of death according to BMI groups among patients with T2DM.

Conclusions

This study aimed to analyze the association between weight status and all-cause mortality, as well as the risk of CVD events, in patients with newly diagnosed T2DM. Patients with T2DM and low BMI, in comparison to individuals with normal BMI, exhibited elevated risks of all-cause mortality and incidence of CVD, CHD, HF, stroke, and IS. These findings emphasize the importance of considering BMI as a significant factor in assessing mortality and cardiovascular risk in patients with T2DM. Particularly in older men with DM, avoiding underweight status can reduce all-cause mortality and the incidence of CVD. The mortality rate was the highest when the lean state was maintained among patients with newly diagnosed T2DM during the follow-up. Maintaining an appropriate weight status is necessary to reduce all-cause mortality and incidence of CVD among lean patients with newly diagnosed T2DM.

Acknowledgments

This study used customized National Health Insurance Data made by the NHIS. The authors declare no conflict of interest with NHIS. This study was supported by the project titled “Institute of Ewha-SCL for Environmental Health (IESEH)” and the Research of Environmental Examination Model for Children and Women (no. 1-2022-0205-001-2).

Funding

None.

Disclosures

The authors declare that they have no known competing or personal financial interests.

Data Availability

All datasets generated during and/or analyzed the current study are available from the corresponding author on reasonable request.

References

Author notes