Abstract
Background
The risk of heart failure (HF) in underweight diabetes mellitus (DM) patients has rarely been studied. We conducted a cohort study to investigate the association between underweight (BMI < 18.5 kg/m2) and BMI change over time and the risk of HF in patients with type 2 DM.
Methods
We utilized the health screening data from the National Health Insurance Service and the Korean National Health Screening database from 2009 to 2012, with follow‐up until December 2018. Participants with DM were categorized into four groups based on their BMI at 4 years before study inclusion and BMI at the study entry: (1) Always Normal Weight (BMI at 4 years ago/BMI at study entry ≥18.5/≥18.5 kg/m2, reference group); (2) Transitioned to Underweight (≥18.5/<18.5 kg/m2); (3) Transitioned to Normal Weight (<18.5/≥18.5 kg/m2) and (4) Always Underweight (<18.5/<18.5 kg/m2). Participants were followed until the development of HF or at the end of the follow‐up. Initial screening data included participants with DM who had the health screening during the study period (n = 2,746,079). Participants aged <20 years (n = 390), those who did not undergo health examination 4 years prior (n = 1,306,520), and those with missing data (n = 77,410) were excluded. Participants diagnosed with HF before study participation (n = 81,645) and within 1 year of study enrolment (n = 11,731) were excluded. After applying exclusion criteria, 1,268,383 participants were finally included in the analysis. The primary outcome was the development of HF. We employed Cox proportional hazards models, adjusting for various confounding factors, to assess the risk of developing HF.
Results
Median follow‐up duration was 6.88 years and men were 63.16%. The mean ages of each groups were as follows: Always Normal Weight (57.92 ± 11.64 years), Transitioned to Underweight (62 ± 13.5 years), Transitioned to Normal Weight (56.6 ± 15.29 years) and Always Underweight (57.76 ± 15.35 years). In comparison with the Always Normal Weight group (n = 1,245,381, HF = 76,360), Transitioned to Underweight group (≥18.5/<18.5 kg/m2, n = 9304, HF = 880, adjusted Hazard Ratio (aHR)1.389, 95% confidence interval (CI) 1.3–1.485) or Transitioned to Normal Weight (<18.5/≥18.5 kg/m2, n = 6024, HF = 478, aHR 1.385, 95% CI 1.266–1.515) exhibited an increased risk of HF. The highest risk was observed in the Always Underweight group (<18.5/<18.5 kg/m2, n = 7674, HF = 665, aHR 1.612, 95% CI 1.493–1.740).
Conclusions
Underweight was significantly associated with the risk of HF in the DM population. Active surveillance for HF in an underweight DM population is needed.
Keywords: Heart failure, Type 2 diabetes mellitus, Underweight
Introduction
The association between obesity, overweight, and diabetes mellitus (DM) has been extensively studied. 1 However, there is an increasing prevalence of DM in underweight individuals, particularly in Asian and developed countries like the United States. 2 , 3 Due to this increasing prevalence, the development of DM in lean individuals has recently attracted the attention of clinicians and researchers, as relatively little is known about the disease progression and outcome in such patients. 3 , 4
Although the effect of obesity on the development of cardiovascular disease (CVD) has been well investigated, relatively few studies have assessed the association between underweight and the risk of CVD. Recent studies have indicated that underweight individuals may face an elevated risk of CVD, suggesting that underweight should be considered an emerging risk factor for CVD in the general population. 5 , 6 , 7 The association between underweight and an increased risk of cardiovascular disease (CVD) has been proposed to be influenced by several factors, including aging, sarcopenia, inflammation, and poor nutritional status. 5 , 8 Additionally, factors such as immunologic and neurohormonal activation, systemic catabolism, and decreased testosterone levels have also been proposed as contributors to this relationship. 8 , 9 , 10
Kwon et al. previously investigated the incidence of CVD and mortality in underweight individuals within the DM population. However, their findings were inconclusive; they observed an elevated risk of myocardial infarction (MI) only in moderately underweight individuals, while no significant association was found in mildly and severely underweight populations. 6 Data is scarcer when it comes to specific CVD components, especially heart failure (HF). Given the advent of sodium‐glucose cotransporter 2 (SGLT2) inhibitors, researchers are actively exploring the links and risk factors between DM and HF development. 11 The prevalence of HF in individuals with DM is up to 22%, four times higher than that in the general population, with an increasing incidence rate. 12 , 13 , 14 While underweight HF patients have a worse prognosis than that of normal weight HF patients, it is unclear whether being underweight is associated with HF development, especially in the DM population. 15
Therefore, we investigated whether being underweight increases the risk of HF in the DM population. Furthermore, as a change in body weight status can affect CVD risk, we aimed to assess whether transitions from normal weight to underweight or vice versa, or persistent underweight, have different impacts on HF risk within the DM population. 16 To address these clinical questions, we conducted a large‐scale national cohort study.
Methods
Study population
We utilized data from the National Health Insurance Service (NHIS) records and the Korean National Health Screening (KNHS) database. The NHIS is a large‐scale cohort in South Korea representing 97.2% of the population. 17 , 18 , 19 It collects a wide range of data, including patient anthropometric data, demographic information, examination findings, administered treatments, and ICD‐10 codes derived from health check‐ups and clinic visits. 20 In addition, Korean adults aged 20 years and older are mandated to undergo regular health check‐ups conducted by the NHIS in every 1–2 years. During these health examinations, participants complete health screening questionnaires, and undergo anthropometric examinations and laboratory tests. All these data are integrated into the KNHS database. 17 , 20 The study was approved by the Institutional Review Board of the Soongsil University (SSU‐202003‐HR‐201‐01). The requirement for informed consent was waived because we used only anonymized and de‐identified data.
Participant selection, grouping, and definitions of conditions
Participants with type 2 DM who underwent health examinations between 2009–2012 were included in the analysis (n = 2,746,079). Type 2 DM was diagnosed based on fasting blood glucose level of ≥126 mg/dL as recorded in the KNHS database, a documented claims history characterized by the ICD‐10 codes E11–14, and antidiabetic medication prescription records dated before January 2009. 21 , 22 Individuals with type 1 DM were not included in the study. We applied several exclusion criteria. Participants aged less than 20 years (<20 years, n = 390), those who did not undergo health examination 4 years prior (n = 1,306,520), and those with missing data (n = 77,410) were excluded from the analysis. Since the primary objective of this study was to identify incident cases of HF, we excluded participants with a prior diagnosis of HF. HF was defined using the ICD‐10 code I50 and records of hospital admission due to HF. As a result, participants diagnosed with HF before study enrolment (n = 81,645) and those diagnosed within 1 year of study initiation (n = 11,731) were excluded. After the application of these exclusion criteria, 1,268,383 participants were finally included in the analysis (Figure 1).
Figure 1.
Flow diagram of study participants.
We utilized the World Health Organization (WHO) BMI cutoffs for adult Asians, with a BMI of 18.5 kg/m2 as the threshold for classifying individuals as underweight or normal weight. 23 Based on BMI measurements 4 years prior to the study entry, as retrieved from the health examination data in the KNHS database, and BMI values at the time of study entry, participants were categorized into four groups: Always Normal Weight (BMI at 4 years before the study entry/BMI at study entry ≥18.5/≥18.5 kg/m2, reference group), Transitioned to Underweight (≥18.5/<18.5 kg/m2), Transitioned to Normal Weight (from <18.5/≥18.5 kg/m2), and Always Underweight (<18.5/<18.5 kg/m2). Participants were monitored until December 31, 2018, or until new‐onset HF was detected, whichever came first.
Definition of covariates and co‐morbidities
All covariates used for the analysis were collected at the study enrollment. Participants were categorized as low‐income if their income was in the lowest 25 percentile of the population. Participants were categorized as never smokers, ex‐smokers, or current smokers 24 Alcohol consumption was categorized based on daily intake: no drinking, consuming less than 30 grams of alcohol per day (mild drinking), or consuming 30 g or more of alcohol per day (heavy drinking). 25 Physical activity was assessed using a self‐report questionnaire during the health examinations. Physical activity was defined as participants engaging in high‐intensity physical activity for a minimum of 20 minutes on at least 3 days per week or participating in moderate‐intensity physical activity for a minimum of 30 minutes on at least 5 days per week. 26
Medical co‐morbidities were defined as follows. Dyslipidemia was defined as the total cholesterol level ≥240 mg/dL or ICD code E78 with a history of dyslipidemia medication claims. Hypertension was defined as either systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg. Additionally, a history of antihypertensive medication claims under ICD codes 10–15 was used to define hypertension. Chronic kidney disease (CKD) was defined as the estimated glomerular filtration rate of <60 mL/min/1.73 m2, calculated using the Modification of Diet in Renal Disease method. 27 , 28 MI was defined as hospitalization records due to MI, with ICD‐10 codes I21 and I22. Stroke was defined as hospitalization history due to an ischemic stroke with the ICD‐10 codes I63 and I64, along with imaging finding consistent with an ischemic stroke. Data related to the duration of DM, insulin use, and the number of oral hypoglycemic agents (OHA) used were collected from medication claim records (ICD‐10, E11–14) and antidiabetic medication prescription records.
As part of an additional analysis, we explored mortality due to HF across different BMI groups. Mortality due to HF was defined as mortality due to the ICD‐10 code I50.
Statistical analysis
Categorical variables were presented as frequency (%), while continuous variables were expressed as mean ± standard deviation or geometric mean (95% confidence interval, CI) depending on the distribution of the variable. Group differences were assessed using one‐way analysis of variance (ANOVA) and chi‐square test. To investigate the association between BMI and the risk of HF, several models were developed. Model 1 was a crude model. Model 2 was adjusted for age and sex. Model 3 included additional adjustments for income, smoking and drinking habits, physical activity, hypertension, dyslipidemia, CKD, and a history of MI or stroke. Model 4 further adjusted for serum glucose level, duration of DM, insulin use, and the use of three or more oral hypoglycaemic agents (OHA, ≥3 OHA).
The multivariable regression analysis was performed using the Cox proportional hazards model, and the results were reported as hazard ratios (HRs) with corresponding 95% CIs. The assumption of proportionality was checked using Schoenfeld residuals and log–log plots. An interaction analysis was conducted using multiplicative interaction terms. The incidence rate was reported as per 1000 person‐years. Adjusted survival curve was plotted using the Kaplan–Meier method.
Statistical significance was set as P < 0.05. Statistical Analysis Software (SAS) (ver. 9.4; SAS Institute Inc., Cary, North Carolina, USA) was used for all statistical analyses.
Results
Baseline characteristics
The median follow‐up duration was 6.88 (5.69–7.87) years. The participants (n = 1,268,383) were grouped according to BMI changes as follows: Always Normal Weight (n = 1,245,381), Transitioned to Underweight (n = 9304), Transitioned to Normal Weight (n = 6024), and Always Under Weight (n = 7674). Among these groups, the Always Under Weight group represented the highest proportion of current smokers and high‐density lipoprotein (HDL) levels. Conversely, this group exhibited the lowest rates of physical activity, hypertension, dyslipidemia, CKD, history of MI or stroke, DM duration of ≥5 years, insulin use, use of three or more OHA, waist circumference, high systolic and diastolic blood pressure, low total cholesterol, serum low‐density lipoprotein (LDL), and triglyceride (TG) levels. All characteristics were significantly different between the groups (Table 1).
Table 1.
Baseline characteristics of study participants.
| BMI (kg/m2) | ≥18.5/≥18.5 | ≥18.5/<18.5 | <18.5/≥18.5 | <18.5/<18.5 | P‐value |
|---|---|---|---|---|---|
| n | 1,245,381 | 9304 | 6024 | 7674 | |
| Age (years) | |||||
| <40 | 86,091 (6.91) | 648 (6.96) | 1058 (17.56) | 1216 (15.85) | <0.0001 |
| 40–64 | 778,735 (62.53) | 4248 (45.66) | 2841 (47.16) | 3564 (46.44) | |
| ≥ 65 | 380,555 (30.56) | 4408 (47.38) | 2125 (35.28) | 2894 (37.71) | |
| Sex, men | 787,290 (63.22) | 5346 (57.46) | 3631 (60.28) | 4953 (64.54) | <0.0001 |
| Income, lowest 25% percentile | 219,002 (17.59) | 1862 (20.01) | 1214 (20.15) | 1487 (19.38) | <0.0001 |
| Smoking habit | <0.0001 | ||||
| No | 682,646 (54.81) | 5248 (56.41) | 3253 (54) | 3799 (49.5) | |
| Ex | 259,482 (20.84) | 1267 (13.62) | 994 (16.5) | 1013 (13.2) | |
| Current | 303,253 (24.35) | 2789 (29.98) | 1777 (29.5) | 2862 (37.29) | |
| Drinking habit | <0.0001 | ||||
| No | 692,649 (55.62) | 6064 (65.18) | 3567 (59.21) | 4477 (58.34) | |
| Mild | 434,918 (34.92) | 2465 (26.49) | 1981 (32.89) | 2516 (32.79) | |
| Heavy | 117,814 (9.46) | 775 (8.33) | 476 (7.9) | 681 (8.87) | |
| Physical activity | 283,952 (22.8) | 1623 (17.44) | 1010 (16.77) | 1197 (15.6) | <0.0001 |
| Hypertension | 704,613 (56.58) | 3818 (41.04) | 2291 (38.03) | 2403 (31.31) | <0.0001 |
| Dyslipidemia | 525,578 (42.2) | 2431 (26.13) | 1495 (24.82) | 1356 (17.67) | <0.0001 |
| Chronic kidney disease | 131,336 (10.55) | 1079 (11.6) | 638 (10.59) | 654 (8.52) | <0.0001 |
| MI or stroke | 63,002 (5.06) | 636 (6.84) | 286 (4.75) | 292 (3.81) | <0.0001 |
| MI | 10,960 (0.88) | 90 (0.97) | 39 (0.65) | 54 (0.7) | 0.0639 |
| Stroke | 53,371 (4.29) | 560 (6.02) | 252 (4.18) | 248 (3.23) | <0.0001 |
| DM duration ≥5 years | 413,487 (33.2) | 3575 (38.42) | 1782 (29.58) | 2218 (28.9) | <0.0001 |
| Insulin use | 91,081 (7.31) | 1259 (13.53) | 770 (12.78) | 780 (10.16) | <0.0001 |
| ≥3 OHA | 172,640 (13.86) | 1604 (17.24) | 758 (12.58) | 928 (12.09) | <0.0001 |
| Metformin | 557,892 (44.8) | 4197 (45.11) | 1872 (31.08) | 2307 (30.06) | <0.0001 |
| Sulfonylurea | 508,832 (40.86) | 3654 (39.27) | 1914 (31.77) | 2175 (28.34) | <0.0001 |
| Meglitinides | 19,316 (1.55) | 290 (3.12) | 121 (2.01) | 157 (2.05) | <0.0001 |
| Thiazolidinedione | 78,443 (6.3) | 466 (5.01) | 330 (5.48) | 288 (3.75) | <0.0001 |
| DPP‐4 inhibitor | 95,631 (7.68) | 781 (8.39) | 294 (4.88) | 407 (5.3) | <0.0001 |
| Alpha glucosidase inhibitor | 138,640 (11.13) | 1539 (16.54) | 710 (11.79) | 961 (12.52) | <0.0001 |
| Age (years) | 57.92 ± 11.64 | 62 ± 13.5 | 56.6 ± 15.29 | 57.76 ± 15.35 | <0.0001 |
| BMI (kg/m2) | 25.11 ± 3.13 | 17.7 ± 0.79 | 20.06 ± 1.91 | 17.23 ± 0.94 | <0.0001 |
| Waist circumference (cm) | 85.6 ± 8.17 | 71.07 ± 6.52 | 74.71 ± 6.82 | 68.79 ± 5.72 | <0.0001 |
| Systolic blood pressure (mmHg) | 128.76 ± 15.21 | 122.9 ± 17.03 | 124.3 ± 16.48 | 121.63 ± 16.69 | <0.0001 |
| Diastolic blood pressure (mmHg) | 78.89 ± 9.97 | 75.22 ± 10.59 | 75.93 ± 10.28 | 74.93 ± 10.35 | <0.0001 |
| Glucose (mg/dL) | 142 ± 43.31 | 149.43 ± 63.69 | 140.56 ± 43.49 | 144.31 ± 50.47 | <0.0001 |
| Total cholesterol (mg/dL) | 195.24 ± 41.38 | 183.93 ± 42.06 | 188.39 ± 39.65 | 182.2 ± 37.39 | <0.0001 |
| HDL (mg/dL) | 51.45 ± 20.11 | 58.9 ± 26.78 | 57.39 ± 24.92 | 61.06 ± 27.16 | <0.0001 |
| LDL (mg/dL) | 110.73 ± 40 | 102.64 ± 41.79 | 106.25 ± 37.62 | 100.14 ± 38.18 | <0.0001 |
| TG (mg/dL) | 144.61 (144.47–144.75) | 101.05 (99.93–102.18) | 109.87 (108.38–111.39) | 95.2 (94.08–96.34) | <0.0001 |
Data are expressed as mean ± standard deviation, geometric mean (95% CI), and frequency (%). ≥18.5/≥18.5: Participant whose BMI was ≥18.5 kg/m2 at 4 years ago and ≥18.5 kg/m2 at baseline (study entry). ≥18.5/<18.5: Participant whose BMI was ≥18.5 kg/m2 at 4 years ago and <18.5 kg/m2 at baseline. <18.5/≥18.5: Participant whose BMI was <18.5 kg/m2 at 4 years ago and ≥18.5 kg/m2 at baseline. <18.5/<18.5: Participant whose BMI was <18.5 kg/m2 at 4 years ago and <18.5 kg/m2 at baseline.
BMI, body mass index; DM, diabetes mellitus; DPP‐4, dipeptidyl peptidase‐4; HDL, high‐density lipoprotein cholesterol; LDL, low‐density lipoprotein cholesterol; MI, myocardial infarction; OHA, oral hypoglycemic agent; TG, triglyceride.
Risk of heart failure development according to body mass index
In Model 1 (unadjusted model), the Transitioned to Underweight group showed the highest risk of HF (HR 1.827, 95% CI 1.709–1.952), whereas the Transitioned to Normal Weight group (HR 1.407, 95% CI 1.286–1.539) and Always Underweight group (HR 1.629, 95% CI 1.509–1.758) showed relatively weaker associations. However, after adjusting for age and sex, the Transitioned to Underweight (aHR 1.396, 95% CI 1.306–1.492) and the Transitioned to Normal Weight (aHR 1.372, 95% CI 1.254–1.501) groups showed similar associations with the risk of HF. The Always Underweight group showed the highest risk of HF (aHR 1.503, 95% CI 1.392–1.622). This association trend remained the same in Models 3 and 4. In Model 4, after complete adjustment, the Transitioned to Underweight group (aHR 1.389, 95% CI 1.3–1.485) showed a similar association as that of the Transitioned to Normal Weight group (aHR 1.385, 95% CI 1.266–1.515). The Always Underweight group showed the strongest association with the risk of HF (aHR 1.612, 95% CI 1.493–1.740) (Table 2). The survival curve showed an overall increased risk of HF in the other three groups compared to the Always Normal Weight (reference); (Figure 2). In addition, the mortality due to HF showed a similar association. Transitioned to Underweight group showed elevated risk of mortality due to HF. The Always Underweight group showed the strongest association with the risk of mortality due to HF (Table S1).
Table 2.
Risk of HF according to BMI at 4 years ago and current BMI.
| BMI (kg/m2) | N | HF | Duration | IR (per 1000) | Model 1 | Model 2 | Model 3 | Model 4 |
|---|---|---|---|---|---|---|---|---|
| HR (95% CI) | HR (95% CI) | HR (95% CI) | HR (95% CI) | |||||
| ≥18.5/≥18.5 | 1,245,381 | 76,360 | 8,302,486.3 | 9.1972 | 1 (ref.) | 1 (ref.) | 1 (ref.) | 1 (ref.) |
| ≥18.5/<18.5 | 9304 | 880 | 55,075.23 | 15.9781 | 1.827 (1.709, 1.952) | 1.396 (1.306, 1.492) | 1.458 (1.364, 1.558) | 1.389 (1.3, 1.485) |
| <18.5/≥18.5 | 6024 | 478 | 37,844.2 | 12.6307 | 1.407 (1.286, 1.539) | 1.372 (1.254, 1.501) | 1.424 (1.301, 1.558) | 1.385 (1.266, 1.515) |
| <18.5/<18.5 | 7674 | 665 | 46,268.32 | 14.3727 | 1.629 (1.509, 1.758) | 1.503 (1.392, 1.622) | 1.619 (1.499, 1.748) | 1.612 (1.493, 1.74) |
Model 1 was unadjusted. Model 2 was adjusted for age and sex. Model 3 was adjusted for age, sex, income, smoking and drinking habits, physical activity, hypertension, dyslipidemia, chronic kidney disease, history of myocardial infarctionor stroke. Model 4 was adjusted for age, sex, income, smoking and drinking habits, physical activity, hypertension, dyslipidemia, chronic kidney disease, history of myocardial infarction or stroke, glucose, DM duration, insulin use, use of ≥3 types of oral hypoglycemic agents. ≥18.5/≥18.5: Participant whose BMI was ≥18.5 kg/m2 at 4 years ago and ≥18.5 kg/m2 at baseline (study entry). ≥18.5/<18.5: Participant whose BMI was ≥18.5 kg/m2 at 4 years ago and <18.5 kg/m2 at baseline. <18.5/≥18.5: Participant whose BMI was <18.5 kg/m2 at 4 years ago and ≥18.5 kg/m2 at baseline. <18.5/<18.5: Participant whose BMI was <18.5 kg/m2 at 4 years ago and <18.5 kg/m2 at baseline.
BMI, body mass index; CI, confidence interval; HF, heart failure; HR, hazard ratio; IR, incidence rate; ref., reference.
Figure 2.
Kaplan–Meier survival curve of development of heart failure according to BMI group. X axis: Time (years); Y axis: Incidence probability. Always Normal weight: BMI was ≥18.5 kg/m2 4 years prior to the study entry and BMI was ≥18.5 kg/m2 at the time of study entry. Transitioned to underweight: BMI was ≥18.5 kg/m2 4 years prior to the study entry and BMI was <18.5 kg/m2 at the time of study entry. Transitioned to Normal weight: BMI was <18.5 kg/m2 4 years prior to the study entry and BMI was ≥18.5 kg/m2 at the time of study entry. Always underweight: BMI was <18.5 kg/m2 4 years prior to the study entry and BMI was <18.5 kg/m2 at the time of study entry.
Subgroup analysis
Among the variables, only sex showed a significant interaction (P < 0.0001). Men showed a higher risk of HF than that of women in the Transitioned to Underweight (aHR 1.538, 95% CI 1.413–1.675), Transitioned to Normal Weight (aHR 1.494, 95% CI 1.333–1.675), and Always Underweight (aHR 1.74, 95% CI 1.587–1.908) groups. Variables such as age, CKD, history of MI or stroke, insulin use, number of OHA, and duration of DM did not display significant interactions (Table 3).
Table 3.
Subgroup analysis
| N | HF | Duration | IR (per 1000) | Model 4 | P for interaction | ||
|---|---|---|---|---|---|---|---|
| HR (95% CI) | |||||||
| <40 | ≥18.5/≥18.5 | 86,091 | 974 | 591,922.44 | 1.6455 | 1 (ref.) | 0.3572 |
| ≥18.5/<8.5 | 648 | 8 | 4404.44 | 1.8164 | 1.308 (0.652, 2.623) | ||
| <18.5/≥18.5 | 1,058 | 7 | 7218.84 | 0.9697 | 0.803 (0.382, 1.687) | ||
| <18.5/<18.5 | 1,216 | 15 | 8290.93 | 1.8092 | 1.477 (0.887, 2.46) | ||
| 40–64 | ≥18.5/≥18.5 | 778,735 | 30,261 | 5,299,453.26 | 5.7102 | 1 (ref.) | |
| ≥18.5/<18.5 | 4,248 | 242 | 27,141.82 | 8.9161 | 1.51 (1.33, 1.714) | ||
| <18.5/≥18.5 | 2,841 | 153 | 18,643.29 | 8.2067 | 1.463 (1.248, 1.715) | ||
| <18.5/<18.5 | 3,564 | 216 | 22,717.38 | 9.5081 | 1.735 (1.518, 1.984) | ||
| ≥65 | ≥18.5/≥18.5 | 380,555 | 45,125 | 2,411,110.61 | 18.7154 | 1 (ref.) | |
| ≥18.5/<18.5 | 4,408 | 630 | 23,528.98 | 26.7755 | 1.347 (1.245, 1.457) | ||
| <18.5/≥18.5 | 2,125 | 318 | 11,982.07 | 26.5396 | 1.365 (1.223, 1.525) | ||
| <18.5/<18.5 | 2,894 | 434 | 15,260 | 28.4404 | 1.556 (1.416, 1.711) | ||
| Men | ≥18.5/≥18.5 | 787,290 | 44,883 | 5,235,949.78 | 8.5721 | 1 (ref.) | <.0001 |
| ≥18.5/<18.5 | 5,346 | 542 | 30,731.82 | 17.6364 | 1.538 (1.413, 1.675) | ||
| <18.5/≥18.5 | 3,631 | 297 | 22,346.95 | 13.2904 | 1.494 (1.333, 1.675) | ||
| <18.5/<18.5 | 4,953 | 461 | 29,018.96 | 15.8862 | 1.74 (1.587, 1.908) | ||
| Women | ≥18.5/≥18.5 | 458,091 | 31,477 | 3,066,536.52 | 10.2647 | 1 (ref.) | |
| ≥18.5/<18.5 | 3,958 | 338 | 24,343.41 | 13.8847 | 1.203 (1.08, 1.339) | ||
| <18.5/≥18.5 | 2,393 | 181 | 15,497.25 | 11.6795 | 1.237 (1.069, 1.432) | ||
| <18.5/<18.5 | 2,721 | 204 | 17,249.36 | 11.8265 | 1.383 (1.205, 1.588) | ||
| CKD (−) | ≥18.5/≥18.5 | 1,114,045 | 60,359 | 7,453,413.3 | 8.0982 | 1 (ref.) | 0.3643 |
| ≥18.5/<18.5 | 8,225 | 708 | 49,370.2 | 14.3406 | 1.393 (1.293, 1.5) | ||
| <18.5/≥18.5 | 5,386 | 389 | 34,161.19 | 11.3872 | 1.404 (1.271, 1.551) | ||
| <18.5/<18.5 | 7,020 | 573 | 42,626.39 | 13.4424 | 1.659 (1.527, 1.801) | ||
| CKD (+) | ≥18.5/≥18.5 | 131,336 | 16,001 | 849,073 | 18.8453 | 1 (ref.) | |
| ≥18.5/<18.5 | 1,079 | 172 | 5705.04 | 30.1488 | 1.385 (1.192, 1.609) | ||
| <18.5/≥18.5 | 638 | 89 | 3683.01 | 24.165 | 1.308 (1.062, 1.611) | ||
| <18.5/<18.5 | 654 | 92 | 3641.93 | 25.2614 | 1.372 (1.118, 1.685) | ||
| MI or stroke (−) | ≥18.5/≥18.5 | 1,182,379 | 67,897 | 7,907,531.17 | 8.5864 | 1 (ref.) | 0.2338 |
| ≥18.5/<18.5 | 8,668 | 779 | 51,886.03 | 15.0137 | 1.413 (1.316, 1.516) | ||
| <18.5/≥18.5 | 5,738 | 429 | 36,257.26 | 11.8321 | 1.399 (1.272, 1.538) | ||
| <18.5/<18.5 | 7,382 | 624 | 44,814.16 | 13.9242 | 1.642 (1.517, 1.777) | ||
| MI or stroke (+) | ≥18.5/≥18.5 | 63,002 | 8,463 | 394,955.13 | 21.4278 | 1 (ref.) | |
| ≥18.5/<18.5 | 636 | 101 | 3189.2 | 31.6694 | 1.244 (1.023, 1.513) | ||
| <18.5/≥18.5 | 286 | 49 | 1586.94 | 30.8771 | 1.28 (0.967, 1.694) | ||
| <18.5/<18.5 | 292 | 41 | 1454.15 | 28.1951 | 1.269 (0.934, 1.725) | ||
| Insulin (−) | ≥18.5/≥18.5 | 1,154,300 | 64,480 | 7,726,261.42 | 8.3456 | 1 (ref.) | 0.6524 |
| ≥18.5/<18.5 | 8,045 | 695 | 48,364.86 | 14.3699 | 1.398 (1.298, 1.507) | ||
| <18.5/≥18.5 | 5,254 | 367 | 33,414.36 | 10.9833 | 1.366 (1.233, 1.514) | ||
| <18.5/<18.5 | 6,894 | 543 | 42,012.78 | 12.9246 | 1.582 (1.454, 1.722) | ||
| Insulin (+) | ≥18.5/≥18.5 | 91,081 | 11,880 | 576,224.88 | 20.617 | 1 (ref.) | |
| ≥18.5/<18.5 | 1,259 | 185 | 6710.37 | 27.5693 | 1.359 (1.176, 1.572) | ||
| <18.5/≥18.5 | 770 | 111 | 4429.84 | 25.0574 | 1.453 (1.205, 1.752) | ||
| <18.5/<18.5 | 780 | 122 | 4255.54 | 28.6685 | 1.766 (1.477, 2.11) | ||
| OHA < 3 | ≥18.5/≥18.5 | 1,072,741 | 60,784 | 7,157,616.71 | 8.4922 | 1 (ref.) | 0.5055 |
| ≥18.5/<18.5 | 7,700 | 681 | 45,686.76 | 14.9059 | 1.418 (1.314, 1.529) | ||
| <18.5/≥18.5 | 5,266 | 381 | 33,251.91 | 11.458 | 1.405 (1.27, 1.554) | ||
| <18.5/<18.5 | 6,746 | 554 | 40,907.07 | 13.5429 | 1.64 (1.508, 1.783) | ||
| OHA ≥ 3 | ≥18.5/≥18.5 | 172,640 | 15,576 | 1,144,869.59 | 13.605 | 1 (ref.) | |
| ≥18.5/<18.5 | 1,604 | 199 | 9388.47 | 21.1962 | 1.301 (1.131, 1.496) | ||
| <18.5/≥18.5 | 758 | 97 | 4592.29 | 21.1224 | 1.311 (1.074, 1.601) | ||
| <18.5/<18.5 | 928 | 111 | 5361.25 | 20.7041 | 1.486 (1.233, 1.791) | ||
| DM duration <5 years | ≥18.5/≥18.5 | 831,894 | 38,801 | 5,587,273.93 | 6.9445 | 1 (ref.) | 0.8297 |
| ≥18.5/<18.5 | 5,729 | 445 | 34,543.82 | 12.8822 | 1.429 (1.302, 1.57) | ||
| <18.5/≥18.5 | 4,242 | 262 | 27,205.04 | 9.6306 | 1.409 (1.247, 1.591) | ||
| <18.5/<18.5 | 5,456 | 391 | 33,569.07 | 11.6476 | 1.622 (1.468, 1.793) | ||
| DM duration ≥5 years | ≥18.5/≥18.5 | 413,487 | 37,559 | 2,715,212.38 | 13.8328 | 1 (ref.) | |
| ≥18.5/<18.5 | 3,575 | 435 | 20,531.41 | 21.187 | 1.351 (1.229, 1.485) | ||
| <18.5/≥18.5 | 1782 | 216 | 10,639.16 | 20.3023 | 1.357 (1.187, 1.552) | ||
| <18.5/<18.5 | 2,218 | 274 | 12,699.24 | 21.5761 | 1.598 (1.418, 1.799) |
Model 4 was adjusted for age, sex, income, smoking and drinking habits, physical activity, hypertension, dyslipidemia, chronic kidney disease, history of myocardial infarction or stroke, glucose, DM duration, insulin use, and use of ≥3 types of oral hypoglycemic agents. ≥18.5/≥18.5: Participant whose BMI was ≥18.5 kg/m2 at 4 years ago and ≥18.5 kg/m2 at baseline (study entry). ≥18.5/<18.5: Participant whose BMI was ≥18.5 kg/m2 at 4 years ago and <18.5 kg/m2 at baseline. <18.5/≥18.5: Participant whose BMI was <18.5 kg/m2 at 4 years ago and ≥18.5 kg/m2 at baseline. <18.5/<18.5: Participant whose BMI was <8.5 kg/m2 at 4 years ago and <18.5 kg/m2 at baseline.
BMI, body mass index; CI, confidence interval; CKD, chronic kidney disease; DM, diabetes mellitus; HF, heart failure; HF, heart failure; HR, hazard ratio; IR, incidence rate; MI, myocardial infarction; OHA, oral hypoglycemic agent; ref., reference.
Discussion
Our study findings indicate that individuals with DM who were underweight either 4 years prior to the study, or at the time of study entry have an increased risk of HF. Notably, those who maintained an underweight status at both of these time points exhibited the highest risk of HF. Importantly, this association was more pronounced in men compared to women. To our knowledge, this is the first study to investigate the relationship between BMI changes over time, specifically 4 years prior to the study entry and at the time of study entry, and HF risk in individuals with DM who are underweight.
Being underweight is an acknowledged risk factor for metabolic shifts that may contribute to the development and progression of severe pathologies. 29 While recent research has shed light on the health risks and related diseases associated with being underweight, it has received considerably less attention compared to the studies focused on overweight or obesity‐related issues. 29
Multiple studies have investigated the association between being underweight and CVD. Park et al. conducted a cross‐sectional analysis involving 491,773 adults in the United States and showed that underweight population had a 19.7% greater risk of CVD than that of normal‐weight population. CVD, in this study, was defined by the episodes of heart attack, MI, angina, or coronary artery disease, and did not include HF. 5 In addition, due to the cross‐sectional nature of the study, reverse causality could not be ruled out. 5 Since individuals may transition between weight categories (underweight, normal weight, and obese) over time, cross‐sectional studies may not fully capture the dynamic nature of weight status.
Funada et al. conducted a prospective cohort study of 43,916 Japanese individuals to assess the association between being underweight and the risk of CVD mortality. The study included a large number of participants (n = 1627) with BMI < 18.5 kg/m2. The results showed that being underweight at baseline was associated with increased CVD and ischaemic heart disease mortality. However, the study needs cautious interpretation as the number of events in the underweight group was relatively small (n = 60 for CVD mortality), and because the study objective was to assess CVD mortality, no information about the risk of CVD development was provided. 30 Compared with this study, our study incorporated a larger number of participants and outcomes, which enhances the reliability of our study results.
Arafa et al. conducted a prospective cohort study involving 4746 Japanese adults to investigate the relationship between weight status (normal, underweight, and overweight) at the age of 20 years and at study entry, and their subsequent risk of cardiovascular disease (CVD) mortality. Their findings revealed that individuals who were of normal weight at age 20 but underweight at study entry exhibited an elevated risk of CVD mortality. In contrast, those who were either of normal weight at age 20 and overweight at study entry or overweight at age 20 and normal weight at study entry did not show a significant association with CVD mortality. Arafa et al.'s study and our study had different primary outcomes (CVD mortality vs. HF risk) and study populations (general population vs. DM population). Furthermore, Arafa et al.'s study relied on self‐reported weight at age 20, which introduces the potential for self‐reporting bias. The study participants had a mean age of 50s and 60s. Such a large interval between weight measurements (age 20 years and study entry) could lead to inaccurate assessment of metabolic burden according to weight change, considering the possible long duration of weight fluctuations. 16 On the other hand, our study has the strength of using nationwide health examination records, which eliminates the risk of recall bias.
Our study yielded several interesting findings. First, the risk of HF increased in both the Transitioned to Underweight and Transitioned to Normal Weight groups, and the strength of association was similar. This result suggests that weight loss and becoming underweight over time, and weight gain in the underweight DM population and becoming normal or overweight are associated with an increased risk of HF. Weight gain increases the risk of HF by predisposing individuals to atherosclerosis, whereas weight loss can increase the risk of HF by inducing muscle and fat wasting. 31 Our results align with those of previous studies, suggesting a U‐shaped association between weight loss, weight gain, and an increased risk of HF. 31 , 32 Notably, our study extends this association to the DM population, providing valuable insights beyond what previous studies in the general population have shown. The mortality risk due to HF exhibited a similar trend of association across all BMI groups with the risk of developing HF.
Second, those who maintained underweight showed the highest risk of HF. Several factors such as aging, sarcopenia, and poor nutritional status have been suggested as potential mechanisms for increased CVD risk in the underweight population. 5 In addition, “metabolically obese underweight” has been suggested as attributed to the elevated CVD risk in the underweight population. 5 A low BMI has been associated with multiple cardiac structural changes, including an increased risk of decrease in dimensions and size of left ventricular chamber, abnormal mitral valve motion, and decreased cardiac index associated with systolic dysfunction. 30 , 33 Our result suggests that the cumulative metabolic effect of the underweight might be more strongly associated with the risk of HF than the impact of weight change.
Third, interaction analysis revealed a stronger association between underweight status and the risk of HF in men than in women. The lifetime risk of HF was similar in men and women. 34 However, sex differences exist in the pathology of cardiac diseases. Men showed a greater cardiac mass and a higher incidence of subclinical coronary artery disease and HF with a reduced ejection fraction (HFrEF) than that of women. 35 In contrast, women tend to show less severe atherosclerosis, left ventricular hypertrophy, and cardiomyocyte apoptosis, and had a higher incidence of HF with preserved ejection fraction (HFpEF) than that of men. 35 A previous study showed an abrupt increase in CVD mortality in non‐smoking men than in non‐smoking women, with an increase in BMI. 36 In addition, estrogen might have played a role in decreasing the risk of HF in underweight women. Estrogen exhibits a protective effect against CVD by reducing fibrosis, stimulating angiogenesis and vasodilation, improving mitochondrial function, and reducing oxidative stress. 37 Overall, it is possible that the metabolic effect of weight change and being underweight pose a greater burden on men than on women.
Our study has several strengths. First, our study included a large number of participants with an adequate number of events to assess the association. Second, our investigation was a nationwide study. While the previous study was limited to incorporating only the urban population, our study represents the general Korean DM population. 16 Third, our dataset has high reliability as we used the KNHS data. Finally, we used a standard statistical method to assess the relationship, with extensive adjustments for confounding factors.
Limitations
Despite its strengths, our study has several limitations. Firstly, it is important to note that our analysis did not capture information regarding BMI changes between the two time points we evaluated: 4 years prior to participants' enrolment in the study and at the beginning of the study. Instead, we assessed the participants' BMI at these specific time points. This limitation means that we may not have a complete picture of weight fluctuations over time, which could have an impact on the observed associations between weight patterns and the risk of HF. Secondly, our study was conducted only in South Korean population. A subsequent prospective cohort study is required to determine whether these results can be applied to other races or countries. Thirdly, our data did not indicate whether weight loss or gain was intentional or unintentional. Fourthly, our study exclusively enrolled patients with type 2 DM. It is important to acknowledge that the findings we observed may differ in individuals with type 1 DM. Lastly, it is important to note that we defined HF using ICD‐10 codes and hospital admission records, which did not allow us to differentiate between HFrEF and HFpEF. Given that sex differences are known to exist in HF, the observed variations between men and women in our study may be attributed to the possibility that men developed HFrEF while women developed HFpEF. 34 Therefore, our data require cautious interpretation.
Conclusions
In conclusion, our study showed that DM patients, whether underweight at any point (Transitioned to Normal Weight, Transitioned to Underweight group), had an increased risk of HF. Additionally, DM patients in the Always Underweight group, who were underweight at 4 years prior to participants' enrollment in the study and at the beginning of the study, had the strongest association with the risk of HF. These results suggest that underweight DM patients require more active screening for the development of HF than normal‐weight DM patients.
Conflict of interest
All authors have nothing to disclose.
Supporting information
Table S1. Risk of mortality due to heart failure according to BMI at four years ago and current BMI.
Acknowledgements
None.
Yoo TK, Han K‐D, Rhee E‐J, Lee W‐Y. Association between underweight and risk of heart failure in diabetes patients. Journal of Cachexia, Sarcopenia and Muscle 2024; 10.1002/jcsm.13417
Tae Kyung Yoo and Kyung‐Do Han contributed equally to this work as first authors.
Won‐Young Lee and Eun‐Jung Rhee contributed equally to this work as corresponding authors.
Contributor Information
Eun‐Jung Rhee, Email: drlwy@hanmail.net, Email: hongsiri@hanmail.net.
Won‐Young Lee, Email: drlwy@hanmail.net.
References
- 1. Piché M‐E, Tchernof A, Després J‐P. Obesity phenotypes, diabetes, and cardiovascular diseases. Circ Res 2020;126:1477–1500. [DOI] [PubMed] [Google Scholar]
- 2. Yang W, Lu J, Weng J, Jia W, Ji L, Xiao J, et al. Prevalence of diabetes among men and women in China. N Engl J Med 2010;362:1090–1101. [DOI] [PubMed] [Google Scholar]
- 3. George AM, Jacob AG, Fogelfeld L. Lean diabetes mellitus: An emerging entity in the era of obesity. World J Diabetes 2015;6:613–620. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Gujral UP, Mohan V, Pradeepa R, Deepa M, Anjana RM, Narayan KM. Ethnic differences in the prevalence of diabetes in underweight and normal weight individuals: The CARRS and NHANES studies. Diabetes Res Clin Pract 2018;146:34–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Park D, Lee JH, Han S. Underweight: another risk factor for cardiovascular disease?: A cross‐sectional 2013 Behavioral Risk Factor Surveillance System (BRFSS) study of 491,773 individuals in the USA. Medicine (Baltimore) 2017;96:e8769. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Kwon H, Yun JM, Park JH, Cho BL, Han K, Joh HK, et al. Incidence of cardiovascular disease and mortality in underweight individuals. J Cachexia Sarcopenia Muscle 2021;12:331–338. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Suastika K, Dwipayana P, Saraswati MR, Gotera W, Budhiarta AA, Sutanegara ND, et al. Underweight is an important risk factor for coronary heart disease in the population of Ceningan Island, Bali. Diab Vasc Dis Res 2012;9:75–77. [DOI] [PubMed] [Google Scholar]
- 8. Seko Y, Kato T, Morimoto T, Yaku H, Inuzuka Y, Tamaki Y, et al. Association between body mass index and prognosis of patients hospitalized with heart failure. Sci Rep 2020;10:16663. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Curcio F, Testa G, Liguori I, Papillo M, Flocco V, Panicara V, et al. Sarcopenia and heart failure. Nutrients 2020;12:211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Takiguchi M, Yoshihisa A, Miura S, Shimizu T, Nakamura Y, Yamauchi H, et al. Impact of body mass index on mortality in heart failure patients. Eur J Clin Invest 2014;44:1197–1205. [DOI] [PubMed] [Google Scholar]
- 11. Rhee EJ. Extra‐glycemic effects of anti‐diabetic medications: Two birds with one stone? Endocrinol Metab (Seoul) 2022;37:415–429. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Dunlay SM, Givertz MM, Aguilar D, Allen LA, Chan M, Desai AS, et al. Type 2 diabetes mellitus and heart failure: A scientific statement from the American Heart Association and the Heart Failure Society of America: This statement does not represent an update of the 2017 ACC/AHA/HFSA heart failure guideline update. Circulation 2019;140:e294–e324. [DOI] [PubMed] [Google Scholar]
- 13. Pop‐Busui R, Januzzi JL, Bruemmer D, Butalia S, Green JB, Horton WB, et al. Heart failure: an underappreciated complication of diabetes. A consensus report of the American Diabetes Association. Diabetes Care 2022;45:1670–1690. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Kim GS, Park JH, Won JC. The role of glucagon‐like peptide 1 receptor agonists and sodium‐glucose cotransporter 2 inhibitors in reducing cardiovascular events in patients with type 2 diabetes. Endocrinol Metab (Seoul) 2019;34:106–116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Curtis JP, Selter JG, Wang Y, Rathore SS, Jovin IS, Jadbabaie F, et al. The obesity paradox: Body mass index and outcomes in patients with heart failure. Arch Intern Med 2005;165:55–61. [DOI] [PubMed] [Google Scholar]
- 16. Arafa A, Kokubo Y, Sheerah HA, Sakai Y, Watanabe E, Li J, et al. Weight change since age 20 and the risk of cardiovascular disease mortality: A prospective cohort study. J Atheroscler Thromb 2022;29:1511–1521. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Rhee EJ, Kwon H, Park SE, Han KD, Park YG, Kim YH, et al. Associations among obesity degree, glycemic status, and risk of heart failure in 9,720,220 Korean adults. Diabetes Metab J 2020;44:592–601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Cho JH, Kwon HM, Park SE, Jung JH, Han KD, Park YG, et al. Protective effect of smoking cessation on subsequent myocardial infarction and ischemic stroke independent of weight gain: A nationwide cohort study. PLoS ONE 2020;15:e0235276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Kyoung D‐S, Kim H‐S. Understanding and utilizing claim data from the Korean National Health Insurance Service (NHIS) and Health Insurance Review & Assessment (HIRA) Database for Research. J Lipid Atheroscler 2022;11:103–110. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Rhee EJ, Cho JH, Kwon H, Park SE, Jung JH, Han KD, et al. Relation between baseline height and new diabetes development: A nationwide population‐based study. Diabetes Metab J 2019;43:794–803. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Committee ADAPP . 2. Classification and diagnosis of diabetes: Standards of medical care in diabetes—2022. Diabetes Care 2021;45:S17–S38. [DOI] [PubMed] [Google Scholar]
- 22. Bae JH, Han KD, Ko SH, Yang YS, Choi JH, Choi KM, et al. Diabetes fact sheet in Korea 2021. Diabetes Metab J 2022;46:417–426. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Appropriate body‐mass index for Asian populations and its implications for policy and intervention strategies. Lancet 2004;363:157–163. [DOI] [PubMed] [Google Scholar]
- 24. Ryan H, Trosclair A, Gfroerer J. Adult current smoking: differences in definitions and prevalence estimates—NHIS and NSDUH, 2008. J Environ Public Health 2012;2012:918368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Hong JW, Noh JH, Kim D‐J. The prevalence of and factors associated with high‐risk alcohol consumption in Korean adults: The 2009–2011 Korea National Health and Nutrition Examination Survey. PLoS ONE 2017;12:e0175299. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Lobelo F, Young DR, Sallis R, Garber MD, Billinger SA, Duperly J, et al. Routine assessment and promotion of physical activity in healthcare settings: A scientific statement from the American Heart Association. Circulation 2018;137:e495–e522. [DOI] [PubMed] [Google Scholar]
- 27. de Boer IH, Khunti K, Sadusky T, Tuttle KR, Neumiller JJ, Rhee CM, et al. Diabetes management in chronic kidney disease: A consensus report by the American Diabetes Association (ADA) and Kidney Disease: Improving Global Outcomes (KDIGO). Diabetes Care 2022;45:3075–3090. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999;130:461–470. [DOI] [PubMed] [Google Scholar]
- 29. Golubnitschaja O, Liskova A, Koklesova L, Samec M, Biringer K, Büsselberg D, et al. Caution, “normal” BMI: Health risks associated with potentially masked individual underweight‐EPMA Position Paper 2021. Epma j 2021;12:243–264. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Funada S, Shimazu T, Kakizaki M, Kuriyama S, Sato Y, Matsuda‐Ohmori K, et al. Body mass index and cardiovascular disease mortality in Japan: The Ohsaki Study. Prev Med 2008;47:66–70. [DOI] [PubMed] [Google Scholar]
- 31. Kim Y‐H, Han K‐d, Jung J‐H, Yoo SJ, Lee S‐S, Lee W‐Y, et al. Weight change and the incidence of heart failure in the Korean population: Data from the National Health Insurance Health checkup 2005–2015. Eur J Prev Cardiol 2020;28:1767–1773. [DOI] [PubMed] [Google Scholar]
- 32. Chaudhry SI, Wang Y, Concato J, Gill TM, Krumholz HM. Patterns of weight change preceding hospitalization for heart failure. Circulation 2007;116:1549–1554. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. de Simone G, Scalfi L, Galderisi M, Celentano A, di Biase G, Tammaro P, et al. Cardiac abnormalities in young women with anorexia nervosa. Br Heart J 1994;71:287–292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Lam CSP, Arnott C, Beale AL, Chandramouli C, Hilfiker‐Kleiner D, Kaye DM, et al. Sex differences in heart failure. Eur Heart J 2019;40:3859–68c. [DOI] [PubMed] [Google Scholar]
- 35. Cediel G, Codina P, Spitaleri G, Domingo M, Santiago‐Vacas E, Lupón J, et al. Gender‐related differences in heart failure biomarkers. Front Cardiovasc Med 2021;7:7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Jee SH, Sull JW, Park J, Lee SY, Ohrr H, Guallar E, et al. Body‐mass index and mortality in Korean men and women. N Engl J Med 2006;355:779–787. [DOI] [PubMed] [Google Scholar]
- 37. Iorga A, Cunningham CM, Moazeni S, Ruffenach G, Umar S, Eghbali M. The protective role of estrogen and estrogen receptors in cardiovascular disease and the controversial use of estrogen therapy. Biol Sex Differ 2017;8:33. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Risk of mortality due to heart failure according to BMI at four years ago and current BMI.