Abstract
Background
We investigated the association between income parameters and the risk of type 2 diabetes (T2D) influenced by obesity.
Methods
We analyzed the Korean National Health Insurance Service data (2002–2022), including 3,462,002 adults (aged ≥ 30 years) without diabetes who underwent health examinations in 2012. Income status was assessed annually from 2008 to 2012, and sustained income status was defined as maintaining the same income category for all five consecutive years. Hazard ratios (HRs) for incident T2D were estimated according to income parameters, and stratified by obesity status and BMI categories.
Results
During a mean follow-up of 8.73 years, 447,688 T2D cases were identified. Individuals with sustained very low-income (multivariable adjusted HR: 1.599; 95% CI: 1.556–1.642) had a higher hazard of T2D than those who had never been in this condition These associations were stronger in participants without obesity, particularly those with underweight (HR: 2.223; 95% CI: 1.919–2.574 for sustained very low-income; p for interaction < 0.0001). Conversely, a reduced T2D hazard associated with sustained high-income (HR: 0.886; 95% CI: 0.880–0.893) and an increased hazard associated with high income variability (HR: 1.108; 95% CI: 1.098–1.117) were more pronounced in individuals with obesity (p for interaction < 0.0001).
Conclusion
Low-income had a greater impact on incident T2D risk among individuals without obesity, whereas among those with obesity, the lower risk observed in sustained high-income groups and the adverse effect of income variability were more pronounced.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12889-026-26986-1.
Keywords: Adult, Cohort study, Diabetes mellitus , Type 2, Income , Obesity, Socioeconomic factor
Background
Globally, type 2 diabetes (T2D) is one of the most significant causes of mortality and morbidity [1] and is also closely related to increased use of medical care and expenditures [2, 3]. Owing to its high prevalence and projected increase in prevalence [4], it is anticipated that the financial costs associated with T2D would rise further. Therefore, as an important global public health issue, identifying populations at high risk of T2D for whom preventive strategies should be concentrated is crucial.
The risk of T2D is influenced by various parameters, including social and environmental factors, collectively expressed as social determinants of health (SDOH) [5].
Income status is a key aspect of SDOH [5]. In previous studies, a low-income state and income decline were associated with excess T2D risk, whereas sustained high-income and income rise were related to a lower risk of T2D [6, 7].
Income status is closely related to obesity status, including the risk of underweight, overweight, or obesity [8, 9]. Furthermore, a systematic review and meta-analysis revealed significant associations between lower-income and the risk of obesity, as well as between obesity and subsequent income. This indicates a possible bidirectional link between social causation and reverse causality processes [10]. Additionally, the risk of T2D varies unequivocally with obesity status and having overweight or obesity pose significant increased risk [11]. Considering the potentially complicated links between the aforementioned factors, it is likely that the association between income status and T2D risk may be related to the obesity status. Nevertheless, the association between income and T2D risk in relation to obesity status categorized by body mass index (BMI) remains unclear.
Therefore, we aimed to determine the longitudinal associations between low- and high-income status and the risk of incident T2D, assessed both cross-sectionally and cumulatively using serial income measures over 5 years. We also examined whether the association between income parameters and incident T2D risk differed according to obesity status, as defined by BMI categories, among Korean adults without a prior history of T2D.
Methods
Data sources
We analyzed the data from the Korean National Health Insurance Service (KNHIS) from January 2002 to December 2022. The KNHIS, operated by the Korean government, is the sole nationwide insurer. This insurer provides coverage to all residents in Korea through two programs: the National Health Insurance (NHI), which covers approximately 97% of the population, and Medical Aid (MA), which supports the remaining 3% with the lowest-income [12, 13]. The KNHIS constructs a public database by consolidating medical treatments, health-screening records, and eligibility data [14]. This comprehensive database includes demographics, monthly health insurance premiums determined by income status, residential region, diagnoses according to the International Classification of Diseases-10th Revision (ICD-10), prescription details, medical procedures, dates of hospital visits, and admissions for all residents of Korea, as well as dates of death for the deceased. Additionally, the KNHIS administers a nationwide health screening program, advocating standardized health examinations at least every 2 years. These health examinations, conducted exclusively by KNHIS-accredited hospitals, are also integrated into the database. The recorded data includes smoking and alcohol consumption history, physical activity metrics, height, weight, waist circumference, blood pressure, fasting plasma glucose (FPG) level, lipid profile, and estimated glomerular filtration rate (eGFR). A detailed description of this database is available in previous publications [12–14].
The study protocol was approved by the Institutional Review Board (IRB) of the Samsung Medical Center (SMC 2023-11-025). This study utilized anonymized public data that had already been secured for other purposes and only involved reviewing this anonymized database without any additional interventions. Thus, it falls under the category of minimal risk research with no additional risk to the subjects. Furthermore, the researchers received data from the KNHIS with all personal identification information removed, making it impossible to identify the participants included. As such, given that the data provided by the KNHIS was anonymized, the requirement for informed consent was waived by the IRB. After obtaining IRB approval, we accessed the data under the permission of the KNHIS for a limited period of utilization. All methods were performed in accordance with the relevant guidelines and regulations.
Study cohort, outcomes, and follow-up
This was a nationwide population-based longitudinal cohort study. Using systematic stratified random sampling with proportional allocation in each stratum, a sample cohort representing 40% of all adults (aged ≥ 20 years) who underwent at least one health examinations in 2012 was selected. The stratifications were based on age and sex. In this sample population, only individuals aged ≥ 30 years were selected to exclude economically inactive young adults. The timepoint of the health examination in 2012 was considered the baseline. We excluded individuals with any missing variable to assess household income status; those who had FPG ≥ 126 mg/dL or prescriptions for antidiabetes medication or claims under ICD-10 codes E11-14 at or before baseline; those with missing data for at least one other variable; and those who died or developed T2D within a year after baseline data collection (Fig. 1).
Fig. 1.
Flow chart depicting the study population.*At least one variable indicates one or more of the following variables: region, fasting plasma glucose, body mass index (height and weight), waist circumference, lifestyle factors (smoking status, alcohol consumption, regular exercise), blood pressure, lipid profile (total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglycerides), liver enzymes, and serum creatinine
The end point was incident T2D, which was defined as at least one claim per year for the prescription of anti-diabetes medication under ICD-10 codes E11–14 or having a FPG ≥ 126 mg/dL [15–17]. Follow-up data were collected from baseline until the date of outcome development, death, or December 31, 2022, whichever occurred first.
Assessment of income status
Each year, the KNHIS assesses the household income of all enrollees and designates the lowest-income individuals (bottom 3%) for enrollment in the MA. For the remaining 97% covered by the NHI, the monthly health insurance premium in each year is determined by household wage income (for employees insured) or by income and household property value (for the self-employed insured), making it a useful proxy for household income status [6, 18]. Low-income status included MA beneficiary individuals or those having health insurance premiums in the lowest quartile for a given year. Very low-income status was defined as qualifying as an MA recipient. High-income status was defined as having health insurance premiums in the highest quartile, and the very high-income group as those in the top 5% for a given year. for a given year. To determine the prebaseline income status, income was assessed from the baseline year (2012) and preceding 4 years (until 2008) [18]. For income parameters, including the cumulative number of years in very low-, low-, or high-income status; income variability; and baseline income status, were used. Thus, sustained income status was defined as maintaining the same income category for all five consecutive years. Parameter definitions as per previous studies [18] are summarized in Additional file 1: Table S1.
Measurements and definitions
Smoking/alcohol consumption history and physical activity data were collected using questionnaires. Venous samples drawn after overnight fasting, were used for blood tests, including FPG and lipid profiles. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration Eqs. [19, 20]. Definitions of BMI, abdominal obesity, alcohol consumption status, regular exercise, hypertension, dyslipidemia, and chronic kidney disease (CKD) are presented in Additional file 1: Table S2.
Statistical analysis
Statistical analyses were performed using the SAS software (version 9.4; SAS Institute, Cary, NC, USA). Two-sided p-values < 0.05 were considered significant. Baseline characteristics are presented according to income parameter groups. We depicted continuous variables with normal distributions as mean ± standard deviation, those with non-normal distributions as geometric means [95% confidence intervals (CIs)], and categorical variables as frequency (percentage).
The incidence rates of T2D were determined by the number of incident cases divided by the total follow-up duration (person-years). Multivariable Cox regression analysis was used to calculate the hazard ratios (HRs) and 95% CIs for the incidence of T2D in the groups of income parameters. Model 1 was unadjusted (crude model); Model 2 was adjusted for age, sex, and residential area (metropolitan, urban, and rural areas); and Model 3, in addition to the variables included in Model 2, was adjusted for BMI, smoking history, alcohol consumption, regular exercise, FPG, presence of hypertension, dyslipidemia, and CKD. To examine the potential effect modification by the presence of obesity (defined as a BMI ≥ 25 kg/m2) [21] and categories of BMI in detail, the HRs (95% CIs) of T2D incidence according to the groups of income parameters were estimated after stratifying the study population by the BMI categories, and p for interaction was calculated.
Results
Baseline characteristics
A total of 3,462,002 individuals (aged ≥ 30 years) without diabetes mellitus at baseline were included in this study (Fig. 1). Among them, during the 5 years (2008–2012), 2,309,458 (66.7%) had never been in a low-income status, while 234,248 (6.8%) had been consistently in the low-income status. Baseline characteristics are presented according to the cumulative number of years of very low-, low-, high, or very high-income status (Table 1). Compared to individuals who had never been in very low- or low-income status, those who had ever been in very low- or low-income states were more likely to be female, having higher eGFR levels, and were less likely to have CKD. As the cumulative number of years in the very low- or low-income status advanced, while an increasing trend toward the proportions of those who were underweight (BMI < 18.5 kg/m2) [22], never smokers, and nondrinkers was noted, a trend of decrease was noted in the proportion of alcohol consumers (heavy drinkers and mild to moderate drinkers). Individuals being in very low-income status for longer duration were more likely to have class II or III obesity (BMI ≥ 30.0 kg/m2) [22] and abdominal obesity, and were less likely to exercise regularly. Furthermore, as the cumulative duration of high and very-high-income status increased, trends toward an increase were observed in the proportion of men, those categorized in class I obesity (BMI 25–29.9 kg/m2) [22], ex-smokers, mild to moderate drinkers, those exercising regularly, and those with CKD. Conversely, individuals with a longer duration of high-income status were less likely to be current smokers, underweight, or have class II or III obesity.
Table 1.
Baseline characteristics by cumulative number of years being in the very low-, low-, and high-income groups
| Total | Cumulative number of years being in the very low-income status | Cumulative number of years being in the low-income status | Cumulative number of years being in the very high-income status | Cumulative number of years being in the high-income status | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 1–4 | 5 | p-value | 0 | 1–4 | 5 | p-value | 0 | 1–4 | 5 | p-value | 0 | 1–4 | 5 | p-value | ||
| No. | 3,462,002 | 3,403,061 | 32,500 | 26,441 | 2,309,458 | 918,296 | 234,248 | 3,089,254 | 276,437 | 96,311 | 1,540,014 | 968,903 | 953,085 | ||||
| Categorical variables, n(%) | |||||||||||||||||
| Age, ≥ 65 years | 394,940 (11.41) | 388,107 (11.40) | 5,564 (17.12) | 1,269 (4.80) | < 0.0001 | 260,852 (11.29) | 94,637 (10.31) | 39,451 (16.84) | < 0.0001 | 332,959 (10.78) | 43,583 (15.77) | 18,398 (19.1) | < 0.0001 | 155,312 (10.09) | 109,118 (11.26) | 130,510 (13.69) | < 0.0001 |
| Sex | < 0.0001 | < 0.0001 | < 0.0001 | < 0.0001 | |||||||||||||
| Male | 1,862,636 (53.80) | 1,841,746 (54.12) | 11,083 (34.10) | 9,807 (37.09) | 1,397,221 (60.5) | 379,543 (41.33) | 85,872 (36.66) | 1,646,785 (53.31) | 155,839 (56.37) | 60,012 (62.31) | 754,737 (49.01) | 508,505 (52.48) | 599,394 (62.89) | ||||
| Female | 1,599,366 (46.20) | 1,561,315 (45.88) | 21,417 (65.90) | 16,634 (62.91) | 912,237 (39.5) | 538,753 (58.67) | 148,376 (63.34) | 1,442,469 (46.69) | 120,598 (43.63) | 36,299 (37.69) | 785,277 (50.99) | 460,398 (47.52) | 353,691 (37.11) | ||||
| Health insurance type | < 0.0001 | < 0.0001 | < 0.0001 | < 0.0001 | |||||||||||||
| Self-employed | 751,236 (21.70) | 740,954 (21.77) | 10,282 (31.64) | - | 532,502 (23.06) | 177,845 (19.37) | 40,889 (17.46) | 644,984 (20.88) | 78,271 (28.31) | 27,981 (29.05) | 306,585 (19.91) | 231,147 (23.86) | 213,504 (22.4) | ||||
| Employee | 2,672,535 (77.20) | 2,662,107 (78.23) | 10,428 (32.09) | - | 1,776,956 (76.94) | 732,848 (79.81) | 162,731 (69.47) | 2,406,188 (77.89) | 198,017 (71.63) | 68,330 (70.95) | 1,196,478 (77.69) | 736,476 (76.01) | 739,581 (77.6) | ||||
| MAB | 38,231 (1.10) | - | 11,790 (36.28) | 26,441 (100) | - | 7,603 (0.83) | 30,628 (13.08) | 38,082 (1.23) | 149 (0.05) | - | 36,951 (2.4) | 1,280 (0.13) | - | ||||
| Baseline income | < 0.0001 | < 0.0001 | < 0.0001 | < 0.0001 | |||||||||||||
| MAB | 382,31 (1.10) | - | 11,790 (36.28) | 26,441 (100) | - | 7,603 (0.83) | 30,628 (13.08) | 38,082 (1.23) | 149 (0.05) | - | 36,951 (2.4) | 1,280 (0.13) | - | ||||
| Q1 | 578,605 (16.71) | 568,788 (16.71) | 9,817 (30.21) | - | - | 374,985 (40.83) | 203,620 (86.92) | 560,883 (18.16) | 17,722 (6.41) | - | 434,436 (28.21) | 144,169 (14.88) | - | ||||
| Q2 | 668,388 (19.31) | 661,578 (19.44) | 6,810 (20.95) | - | 361,650 (15.66) | 306,738 (33.4) | - | 650,449 (21.06) | 17,939 (6.49) | - | 529,291 (34.37) | 139,097 (14.36) | - | ||||
| Q3 | 937,374 (27.08) | 934,290 (27.45) | 3,084 (9.49) | - | 780,598 (33.8) | 156,776 (17.07) | - | 914,407 (29.6) | 22,967 (8.31) | - | 539,336 (35.02) | 361,122 (37.27) | 36,916 (3.87) | ||||
| Q4 | 1,239,404 (35.80) | 1,238,405 (36.39) | 999 (3.07) | - | 116,7210 (50.54) | 72,194 (7.86) | - | 925,433 (29.96) | 217,660 (78.74) | 96,311 (100) | - | 323,235 (33.36) | 916,169 (96.13) | ||||
| Residential area, Urban | 1,585,058 (45.78) | 1,560,285 (45.85) | 13,625 (41.92) | 11,148 (42.16) | < 0.0001 | 106,2750 (46.02) | 41,5363 (45.23) | 10,6945 (45.65) | < 0.0001 | 1,396,673 (45.21) | 135,376 (48.97) | 53,009 (55.04) | < 0.0001 | 689,672 (44.78) | 445,356 (45.96) | 450,030 (47.22) | < 0.0001 |
| BMI level, kg/m2 | < 0.0001 | < 0.0001 | < 0.0001 | < 0.0001 | |||||||||||||
| < 18.5 | 1,071,54 (3.10) | 104,298 (3.06) | 1,353 (4.16) | 1,503 (5.68) | 67,294 (2.91) | 31,276 (3.41) | 8,584 (3.66) | 98,264 (3.18) | 6,898 (2.5) | 1,992 (2.07) | 55,331 (3.59) | 29,189 (3.01) | 22,634 (2.37) | ||||
| 18.5–22.9 | 1,355,081 (39.14) | 1,331,751 (39.13) | 13,044 (40.14) | 10,286 (38.9) | 877,570 (38) | 382,121 (41.61) | 95,390 (40.72) | 1,216,699 (39.38) | 103,800 (37.55) | 34,582 (35.91) | 626,836 (40.7) | 379,247 (39.14) | 348,998 (36.62) | ||||
| 23.0-24.9 | 874,382 (25.26) | 861,151 (25.31) | 7,433 (22.87) | 5,798 (21.93) | 595,263 (25.78) | 222,065 (24.18) | 57,054 (24.36) | 771,637 (24.98) | 75,023 (27.14) | 27,722 (28.78) | 370,627 (24.07) | 243,702 (25.15) | 260,053 (27.29) | ||||
| 25.0-29.9 | 1,006,626 (29.08) | 990,025 (29.09) | 9,179 (28.24) | 7,422 (28.07) | 692,347 (29.98) | 249,534 (27.17) | 64,745 (27.64) | 893,745 (28.93) | 83,073 (30.05) | 29,808 (30.95) | 428,670 (27.84) | 282,949 (29.2) | 295,007 (30.95) | ||||
| ≥ 30.0 | 118,759 (3.43) | 115,836 (3.4) | 1,491 (4.59) | 1,432 (5.42) | 76,984 (3.33) | 33,300 (3.63) | 8,475 (3.62) | 108,909 (3.53) | 7643 (2.76) | 2,207 (2.29) | 58,550 (3.8) | 33,816 (3.49) | 26,393 (2.77) | ||||
| Abdominal obesity | 646,269 (18.67) | 633,129 (18.6) | 7,174 (22.07) | 5,966 (22.56) | < 0.0001 | 436,324 (18.89) | 163,764 (17.83) | 4,6181 (19.71) | < 0.0001 | 571,270 (18.49) | 54,904 (19.86) | 20,095 (20.86) | < 0.0001 | 279,350 (18.14) | 183,004 (18.89) | 183,915 (19.3) | < 0.0001 |
| Smoking status | < 0.0001 | < 0.0001 | < 0.0001 | < 0.0001 | |||||||||||||
| Never | 2,061,184 (59.54) | 2,020,579 (59.38) | 22,242 (68.44) | 18,363 (69.45) | 128,3435 (55.57) | 612,013 (66.65) | 165,736 (70.75) | 1,837,609 (59.48) | 166,970 (60.4) | 56,605 (58.77) | 940,052 (61.04) | 586,365 (60.52) | 534,767 (56.11) | ||||
| Former | 555,307 (16.04) | 550,153 (16.17) | 2,998 (9.22) | 2,156 (8.15) | 426,350 (18.46) | 104,092 (11.34) | 24,865 (10.61) | 475,530 (15.39) | 56,199 (20.33) | 23,578 (24.48) | 198,619 (12.9) | 149,019 (15.38) | 207,669 (21.79) | ||||
| Current | 845,511 (24.42) | 832,329 (24.46) | 7,260 (22.34) | 5,922 (22.4) | 599,673 (25.97) | 202,191 (22.02) | 43,647 (18.63) | 776,115 (25.12) | 53,268 (19.27) | 16,128 (16.75) | 401,343 (26.06) | 233,519 (24.1) | 210,649 (22.1) | ||||
| Alcohol consumption | < 0.0001 | < 0.0001 | < 0.0001 | < 0.0001 | |||||||||||||
| Non | 1,780,778 (51.44) | 1,739,653 (51.12) | 21,708 (66.79) | 19,417 (73.44) | 1,111,256 (48.12) | 518,643 (56.48) | 150,879 (64.41) | 1,590,194 (51.48) | 141,919 (51.34) | 48,665 (50.53) | 821,531 (53.35) | 501,096 (51.72) | 458,151 (48.07) | ||||
| Mild to moderate | 1,432,936 (41.39) | 1,418,342 (41.68) | 8,908 (27.41) | 5,686 (21.5) | 1,017,472 (44.06) | 343,704 (37.43) | 71,760 (30.63) | 1,276,664 (41.33) | 115,401 (41.75) | 40,871 (42.44) | 610,364 (39.63) | 399,583 (41.24) | 422,989 (44.38) | ||||
| Heavy | 248,288 (7.17) | 245,066 (7.2) | 1,884 (5.8) | 1,338 (5.06) | 180,730 (7.83) | 559,49 (6.09) | 116,09 (4.96) | 222,396 (7.2) | 19,117 (6.92) | 6,775 (7.03) | 108,119 (7.02) | 68,224 (7.04) | 71,945 (7.55) | ||||
| Regular exercise | 669,455 (19.34) | 660,762 (19.42) | 4,886 (15.03) | 3,807 (14.4) | < 0.0001 | 460,338 (19.93) | 164,218 (17.88) | 44,899 (19.17) | < 0.0001 | 581,995 (18.84) | 63,508 (22.97) | 23,952 (24.87) | < 0.0001 | 272,456 (17.69) | 181,165 (18.7) | 215,834 (22.65) | < 0.0001 |
| Hypertension | 843,663 (24.37) | 825,168 (24.25) | 10,693 (32.9) | 7,802 (29.51) | < 0.0001 | 552,130 (23.91) | 219,605 (23.91) | 71,928 (30.71) | < 0.0001 | 742,422 (24.03) | 72,551 (26.25) | 28,394 (29.48) | < 0.0001 | 373,725 (24.27) | 226,461 (23.37) | 243,477 (25.55) | < 0.0001 |
| Dyslipidemia | 631,748 (18.25) | 618,115 (18.16) | 7,492 (23.05) | 6,141 (23.23) | < 0.0001 | 416,449 (18.03) | 164,482 (17.91) | 50,817 (21.69) | < 0.0001 | 551,485 (17.85) | 57,669 (20.86) | 22,568 (23.43) | < 0.0001 | 272,857 (17.72) | 173,778 (17.94) | 185,113 (19.42) | < 0.0001 |
| CKD | 108,971 (3.15) | 105,895 (3.11) | 1,833 (5.64) | 1,243 (4.70) | < 0.0001 | 69,713 (3.02) | 28,873 (3.14) | 10,385 (4.43) | < 0.0001 | 94,236 (3.05) | 10,462 (3.78) | 4,273 (4.44) | < 0.0001 | 45,889 (2.98) | 29,963 (3.09) | 33,119 (3.47) | < 0.0001 |
| Continuous variables, mean ± SD or geometric means (95% CI) | |||||||||||||||||
| Age, years | 49.12 ± 11.95 | 49.07 ± 11.96 | 53.26 ± 11.79 | 51.09 ± 8.23 | < 0.0001 | 48.73 ± 12 | 49.01 ± 11.81 | 53.45 ± 11.13 | < 0.0001 | 48.66 ± 11.91 | 52.13 ± 11.75 | 55.21 ± 10.55 | < 0.0001 | 48.28 ± 12.02 | 48.52 ± 12.29 | 51.09 ± 11.22 | < 0.0001 |
| BMI, kg/m2 | 23.74 ± 3.15 | 23.74 ± 3.14 | 23.73 ± 3.45 | 23.71 ± 3.66 | 0.343 | 23.81 ± 3.11 | 23.61 ± 3.22 | 23.63 ± 3.23 | < 0.0001 | 23.73 ± 3.17 | 23.81 ± 2.96 | 23.87 ± 2.82 | < 0.0001 | 23.65 ± 3.26 | 23.76 ± 3.15 | 23.87 ± 2.95 | < 0.0001 |
| WC, cm | 80.14 ± 8.89 | 80.15 ± 8.88 | 79.63 ± 9.24 | 79.72 ± 9.58 | < 0.0001 | 80.66 ± 8.78 | 79.05 ± 9.02 | 79.35 ± 9.03 | < 0.0001 | 80.03 ± 8.92 | 80.91 ± 8.65 | 81.69 ± 8.33 | < 0.0001 | 79.53 ± 9.04 | 80.12 ± 8.93 | 81.14 ± 8.5 | < 0.0001 |
| SBP, mmHg | 121.75 ± 14.54 | 121.76 ± 14.53 | 122.03 ± 15.57 | 120.6 ± 15.34 | < 0.0001 | 121.84 ± 14.39 | 121.35 ± 14.73 | 122.47 ± 15.22 | < 0.0001 | 121.79 ± 14.55 | 121.39 ± 14.5 | 121.34 ± 14.28 | < 0.0001 | 121.9 ± 14.71 | 121.46 ± 14.49 | 121.8 ± 14.31 | < 0.0001 |
| DBP, mmHg | 76.19 ± 9.95 | 76.19 ± 9.94 | 76.18 ± 10.25 | 75.81 ± 10.28 | < 0.0001 | 76.31 ± 9.91 | 75.85 ± 9.99 | 76.3 ± 10.13 | < 0.0001 | 76.22 ± 9.95 | 75.85 ± 9.91 | 75.94 ± 9.79 | < 0.0001 | 76.26 ± 10 | 75.93 ± 9.93 | 76.31 ± 9.87 | < 0.0001 |
| Fasting glucose, mg/dL | 93.46 ± 10.97 | 93.45 ± 10.96 | 93.64 ± 11.36 | 93.49 ± 11.54 | 0.0105 | 93.58 ± 10.89 | 93.07 ± 11.07 | 93.73 ± 11.3 | < 0.0001 | 93.38 ± 11.01 | 93.91 ± 10.68 | 94.45 ± 10.55 | < 0.0001 | 93.28 ± 11.19 | 93.21 ± 10.88 | 93.98 ± 10.69 | < 0.0001 |
| Total cholesterol, mg/dL | 197.55 ± 35.77 | 197.58 ± 35.72 | 196.84 ± 38.1 | 194.5 ± 38.79 | < 0.0001 | 197.68 ± 35.49 | 196.98 ± 36.14 | 198.51 ± 36.99 | < 0.0001 | 197.33 ± 35.82 | 199.27 ± 35.37 | 199.88 ± 34.86 | < 0.0001 | 196.92 ± 36.2 | 197.41 ± 35.72 | 198.72 ± 35.08 | < 0.0001 |
| HDL-C, mg/dL | 55.63 ± 17.46 | 55.62 ± 17.44 | 56.37 ± 19.58 | 55.29 ± 18.22 | < 0.0001 | 55 ± 17.15 | 56.93 ± 18.08 | 56.64 ± 17.69 | < 0.0001 | 55.72 ± 17.62 | 54.93 ± 16.51 | 54.49 ± 14.64 | < 0.0001 | 56.4 ± 17.98 | 55.62 ± 17.4 | 54.38 ± 16.59 | < 0.0001 |
| LDL-C, mg/dL | 116.70 ± 33.45 | 116.74 ± 33.42 | 115.21 ± 34.89 | 113.82 ± 35.14 | < 0.0001 | 117 ± 33.3 | 115.84 ± 33.62 | 117.15 ± 34.18 | < 0.0001 | 116.35 ± 33.53 | 119.36 ± 32.82 | 120.23 ± 32.21 | < 0.0001 | 115.52 ± 33.83 | 116.66 ± 33.32 | 118.66 ± 32.87 | < 0.0001 |
| eGFR, mL/min/1.73m2 | 91.32 ± 34.39 | 91.32 ± 34.43 | 91.26 ± 31.27 | 92.5 ± 33.83 | < 0.0001 | 91.05 ± 35.67 | 92.21 ± 31.72 | 90.58 ± 31.43 | < 0.0001 | 91.61 ± 33.82 | 89.34 ± 36.36 | 87.69 ± 44.76 | < 0.0001 | 92.49 ± 32.61 | 91.59 ± 33.38 | 89.17 ± 37.97 | < 0.0001 |
| aTriglyceride, mg/dL | 109.76 (109.69-109.82) | 109.75 (109.69-109.82) | 109.83 (109.18-110.48) | 110.07 (109.35–110.8) | 0.6893 | 112.09 (112.01-112.17) | 104.82 (104.71-104.94) | 106.87 (106.64-107.11) | < 0.0001 | 109.69 (109.63-109.76) | 110.02 (109.8-110.24) | 111.05 (110.68-111.42) | < 0.0001 | 108.26 (108.16-108.35) | 109.08 (108.96–109.2) | 112.94 (112.81-113.06) | < 0.0001 |
aThis variable exhibited a non-normal distribution and is therefore presented as geometric means (95% confidence intervals)
Abbreviations: BMI body mass index, CI confidence interval, CKD chronic kidney disease, DBP diastolic blood pressure, eGFR estimated glomerular filtration rate HDL-C high-density lipoprotein cholesterol, IQR interquartile range, LDL-C low-density lipoprotein cholesterol, MAB Medical Aid beneficiaries indicating the very low-income status, Q1 health insurance premium quartile 1 (the lowest quartile), Q2 health insurance premium quartile 2, Q3 health insurance premium quartile 3, Q4 health insurance premium quartile 4 (the highest quartile), SBP systolic blood pressure, SD standard deviation, WC waist circumference
Groups of income parameters and T2D incidence
During a mean follow-up period of 8.73 years (30,213,293 person-years), the cohort had 447,688 cases of T2D (12.9% of the study participants). A significant dose–response relationship was observed across baseline income percentiles, with a monotonic increase in the hazard of type 2 diabetes toward the lowest income categories (p for trend < 0.001, Table S3 and Fig. S1,). The accumulated number of years of very low- or low-income status was significantly associated with an increased hazard of T2D; the increase being dose-responsive (Table 2). Compared to individuals who had never been in the very low- or low-income status (reference), the highest hazards were demonstrated in those with sustained very low- or low-income status for 5 years [Model 3, HR (95% CI): 1.599 (1.556–1.642) in those with sustained very low-income and 1.148 (1.136–1.160) in those with sustained low-income status]. In contrast, individuals with consecutive high-income status for 5 years had an 11.3% lower hazard of T2D incidence (Model 3, 95% CI 0.880–0.893). Similar findings were observed among those in the very-high-income status. With respect to baseline status, when individuals in the high-income status were used as a reference, those in lower-income status were seen to be associated with increased hazards of T2D [HR, 1.678 (95% CI, 1.640–1.717) among MA beneficiaries]. Likewise, higher income variability was also associated with increased hazards of T2D [Model 3, Q1 vs. Q4 for SD: HR 1.108 (95% CI 1.098–1.117)] (Table S4).
Table 2.
Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters
| Income parameters | N | Events (n) | Duration | Incidence rate, per 1000 person-years | Hazard ratios (95% CIs) | ||
|---|---|---|---|---|---|---|---|
| Person-years | Model 1 | Model 2 | Model 3 | ||||
| Cumulative number of years being in the very low-income status | |||||||
| 0 | 3,403,061 | 435,754 | 29,728,852 | 14.66 | 1 (Ref.) | 1 (Ref.) | 1 (Ref.) |
| 1 | 9,805 | 1,890 | 81,101 | 23.30 | 1.600 (1.529, 1.674) | 1.409 (1.347, 1.474) | 1.345 (1.285, 1.407) |
| 2 | 9,570 | 2,135 | 76,664 | 27.85 | 1.919 (1.839, 2.002) | 1.509 (1.446, 1.575) | 1.355 (1.299, 1.414) |
| 3 | 6,266 | 1,187 | 52,478 | 22.62 | 1.551 (1.465, 1.641) | 1.624 (1.534, 1.719) | 1.504 (1.420, 1.592) |
| 4 | 6,859 | 1,359 | 56,362 | 24.11 | 1.657 (1.571, 1.748) | 1.499 (1.421, 1.581) | 1.409 (1.336, 1.486) |
| 5 | 26,441 | 5,363 | 217,837 | 24.62 | 1.689 (1.644, 1.735) | 1.763 (1.716, 1.811) | 1.599 (1.556, 1.642) |
| p for trend | < 0.0001 | < 0.0001 | < 0.0001 | ||||
| Cumulative number of years being in the low-income status | |||||||
| 0 | 2,309,458 | 289,849 | 20,212,831 | 14.34 | 1 (Ref.) | 1 (Ref.) | 1 (Ref.) |
| 1 | 360,725 | 44,740 | 3,160,362 | 14.16 | 0.987 (0.977, 0.997) | 1.075 (1.064, 1.085) | 1.072 (1.061, 1.083) |
| 2 | 238,446 | 30,215 | 2,083,986 | 14.5 | 1.011 (0.999, 1.023) | 1.079 (1.066, 1.092) | 1.082 (1.069, 1.095) |
| 3 | 175,687 | 23,580 | 1,529,117 | 15.42 | 1.076 (1.062, 1.091) | 1.110 (1.096, 1.125) | 1.111 (1.096, 1.126) |
| 4 | 143,438 | 20,544 | 1,239,908 | 16.57 | 1.158 (1.141, 1.174) | 1.113 (1.097, 1.129) | 1.115 (1.099, 1.131) |
| 5 | 234,248 | 38,760 | 1,987,089 | 19.51 | 1.367 (1.353, 1.382) | 1.175 (1.162, 1.187) | 1.148 (1.136, 1.160) |
| p for trend | < 0.0001 | < 0.0001 | < 0.0001 | ||||
| Cumulative number of years being in the very high-income status | |||||||
| 0 | 3,089,254 | 395,248 | 26,979,230 | 14.65 | 1 (Ref.) | 1 (Ref.) | 1 (Ref.) |
| 1 | 114,951 | 15,242 | 1,001,646 | 15.22 | 1.039 (1.022, 1.056) | 0.908 (0.894, 0.923) | 0.934 (0.919, 0.949) |
| 2 | 65,310 | 9,019 | 567,033 | 15.91 | 1.086 (1.063, 1.109) | 0.908 (0.889, 0.927) | 0.935 (0.916, 0.955) |
| 3 | 50,008 | 7,109 | 434,142 | 16.37 | 1.118 (1.092, 1.144) | 0.899 (0.878, 0.921) | 0.933 (0.911, 0.955) |
| 4 | 46,168 | 6,689 | 399,969 | 16.72 | 1.142 (1.115, 1.170) | 0.890 (0.869, 0.912) | 0.918 (0.896, 0.941) |
| 5 | 96,311 | 14,381 | 831,273 | 17.3 | 1.182 (1.163, 1.202) | 0.851 (0.837, 0.866) | 0.888 (0.873, 0.903) |
| p for trend | < 0.0001 | < 0.0001 | < 0.0001 | ||||
| Cumulative number of years being in the high-income status | |||||||
| 0 | 1,540,014 | 200,863 | 13,413,730 | 14.97 | 1 (Ref.) | 1 (Ref.) | 1 (Ref.) |
| 1 | 311,780 | 38,322 | 2,732,846 | 14.02 | 0.935 (0.925, 0.946) | 0.943 (0.933, 0.954) | 0.954 (0.943, 0.964) |
| 2 | 228,445 | 28,321 | 2,001,681 | 14.15 | 0.944 (0.932, 0.956) | 0.929 (0.917, 0.941) | 0.948 (0.936, 0.960) |
| 3 | 214,657 | 26,963 | 1,879,507 | 14.35 | 0.957 (0.945, 0.969) | 0.919 (0.907, 0.931) | 0.943 (0.931, 0.955) |
| 4 | 214,021 | 27,601 | 1,869,582 | 14.76 | 0.985 (0.972, 0.997) | 0.909 (0.897, 0.920) | 0.930 (0.919, 0.942) |
| 5 | 953,085 | 125,618 | 8,315,948 | 15.11 | 1.008 (1.001, 1.015) | 0.854 (0.848, 0.860) | 0.886 (0.880, 0.893) |
| p for trend | 0.0113 | < 0.0001 | < 0.0001 | ||||
| Baseline income status | |||||||
| MAB | 38,231 | 7,728 | 315,354 | 24.51 | 1.656 (1.618, 1.694) | 1.881 (1.838, 1.924) | 1.678 (1.640, 1.717) |
| Q1 | 578,605 | 83,444 | 4,995,409 | 16.70 | 1.124 (1.115, 1.133) | 1.134 (1.124, 1.143) | 1.116 (1.107, 1.126) |
| Q2 | 668,388 | 82,783 | 5,848,218 | 14.16 | 0.951 (0.943, 0.959) | 1.120 (1.111, 1.130) | 1.096 (1.086, 1.105) |
| Q3 | 937,374 | 112,505 | 8,229,330 | 13.67 | 0.918 (0.911, 0.925) | 1.094 (1.086, 1.102) | 1.057 (1.049, 1.065) |
| Q4 | 1,239,404 | 161,228 | 10,824,983 | 14.89 | 1 (Ref.) | 1 (Ref.) | 1 (Ref.) |
| p for trend | < 0.0001 | < 0.0001 | < 0.0001 | ||||
Baseline income status was categorized as MAB (very low-income status) and health insurance premium quartiles 1 (the lowest quartile), 2, 3, and 4 (the highest quartile), constructing a total of five groups
Model 1: Unadjusted
Model 2: Adjusted for age, sex, and residential area (metropolitan, urban, and rural)
Model 3: Adjusted for the variables included in Model 2 and body mass index, smoking history, drinking, regular exercise, fasting plasma glucose, hypertension, dyslipidemia, and chronic kidney disease
Abbreviations: CI confidence interval, MAB Medical Aid beneficiaries indicating a very low-income status, Q1 quartile 1 (the lowest quartile), Q2 quartile 2, Q3 quartile 3, Q4 quartile 4
Similar results were observed in individuals aged < 65 and ≥ 65 years (Table S5 and S6).
Stratified analyses according to obesity status and BMI categories
We further investigated the association between various income parameters and the hazards of T2D after stratifying the study population according to the presence of obesity (Fig. 2, Table S7). Regardless of the presence of obesity, increased hazard of T2D in individuals with higher cumulative durations of very-low-income status was consistently observed. Thus, individuals with obesity and sustained very-low-income status demonstrated the highest hazard [0 years without obesity vs. 5 years with obesity: HR 2.715 (95% CI 2.612–2.823)] in the fully adjusted model (Model 3). However, the associations between cumulative duration in the very-low-income group and the hazard of T2D were more prominent in individuals without obesity (p for interaction < 0.0001). While a sustained very low-income status for 5 years was associated with a 51.8% increased hazard in the subpopulation with obesity [Model 3, 0 vs. 5 years: (95% CI 1.460–1.578)], it was associated with a 69.8% higher hazard in the subpopulation without obesity [0 vs. 5 years: (95% CI 1.636–1.763)]. The analyses replacing the income parameter with the cumulative number of years in the low-income group yielded similar patterns of interaction by the presence of obesity (p for interaction < 0.0001). In addition, the increased hazard of T2D associated with baseline very-low-income status was more pronounced in participants without obesity (p for interaction < 0.0001).
Fig. 2.
Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameter groups, stratified by the presence of obesity (BMI ≥ 25 kg/m2), Reference groups: individuals within each obesity category (non-obesity and obesity) who had never been in the very low-, low-, high-, or very high–income statuses, or those who were in baseline health insurance premium quartile 4 (the highest income group) or the lowest quartile of income variability. Baseline income status was categorized as MAB (very low-income status) and health insurance premium quartiles 1 (the lowest quartile), 2, 3, and 4 (the highest quartile), constructing a total of five groups Adjusted for age, sex, residential area (metropolitan, urban, and rural areas), body mass index, smoking history, drinking, regular exercise, fasting plasma glucose, hypertension, dyslipidemia, and chronic kidney disease
In contrast, the lower hazard of T2D associated with advanced cumulative duration in the high-income status was more prominent in the group with obesity than in the group without obesity (p for interaction < 0.0001). The increased hazard of T2D associated with higher income variability was more pronounced in individuals with obesity (p for interaction < 0.0001) (Table S8). In analyses stratified by age group (Tables S9 and S10), the results among adults aged < 65 years—who are largely unaffected by retirement—were consistent with our initial findings.
After stratifying the study population by BMI category, the HRs (95% CIs) of T2D according to income parameter groups were further evaluated in detail (Fig. 3 and Additional file 1: Table S11). In terms of the income parameters of the cumulative number of years being in the very low- or low-income status and baseline very low-income status, the associations with the hazard of T2D were stronger in individuals with lower BMI categories; the most prominent association being demonstrated in participants with underweight (p for interaction < 0.0001). The sustained very low-income status for 5 years exhibited 2.223-fold (95% CI 1.919–2.574), 1.803-fold (95% CI 1.712–1.899), 1.629-fold (95% CI 1.537–1.727), 1.535-fold (95% CI 1.470–1.604), and 1.239-fold (95% CI 1.138–1.350) increased risk of T2D compared to those who never experienced the very low-income status in individuals with BMI < 18.5, 18.5–22.9, 23.0–24.9, 25.0–29.9, and ≥ 30.0 kg/m2, respectively (All Model 3). With respect to the cumulative number of years in the high-income status and income variability groups, the association was stronger in individuals with a higher BMI (all p for interaction < 0.0001). Compared to individuals who never experienced high-income status, those with sustained high-income status for 5 years had a 16.9% lower risk of T2D (Model 3, 95% CI 0.810–0.853) in those with a BMI of ≥ 30 kg/m2, and a 6.5% lower risk (Model 3, 95% CI 0.878–0.995) in participants with underweight. Similar results were observed in the sustained very–high–income group.
Fig. 3.
Adjusted hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income status, stratified analyses by the BMI categories. Comparison between (A) cumulative number of years being in the very low-income status: 5 years versus 0 years (ref.); (B) cumulative number of years being in the low-income status: 5 years versus 0 years (ref.); (C) cumulative number of years being in the very high-income status: 5 years versus 0 years (ref.); (D) cumulative number of years being in the high-income status: 5 years versus 0 years (ref.); (E) baseline income: very low-income status (medical aid beneficiaries) versus health insurance premium quartile 4 (the highest quartile, ref.); (F) income variability: quartile 4 (the highest quartile) versus quartile 1 (the lowest quartile, ref.)
Sensitivity analyses
Sensitivity analyses excluding participants with myocardial infarction, stroke, and/or cancer (Tables S12 and S13), sub-distribution HRs accounting for all-cause deaths as competing events using the Fine and Gray method [23] (Tables S14 and S15), sensitivity analyses excluding those who died or developed T2D within two years after baseline data collection (Table S16 and S17) showed results consistent with those from the main analyses.
Discussion
This nationwide study, which included 3,462,002 Korean adults aged ≥ 30 years, including those 65 years and above, demonstrated an increased risk of T2D associated with cumulative exposure to very low- or low-income status, baseline low-income status, and higher income variability. Conversely, a reduced risk of T2D was observed with an increased cumulative duration in high-income status. Furthermore, stratified analyses by BMI category revealed that participants in lower BMI categories had a more pronounced increased risk of T2D associated with cumulative exposure to very low- or low-income status and very low-income status at baseline. Among participants with obesity and higher BMI categories, the reduced risk of T2D associated with cumulative duration in high-income status was more prominent, as was the increased risk associated with higher income variability.
Consistent with our findings, a previous study that included a Korean population demonstrated an elevated T2D risk in individuals who experienced sustained low-income status or income decline and a reduced T2D risk in those with consecutive high-income status or income rise [6]. The authors also found a more prominent impact of sustained low-income in individuals without obesity than those with obesity. However, unlike in the present study, their analysis did not include individuals aged ≥ 65 years. Furthermore, their study did not stratify obesity status in as much detail as ours and considered only the presence of obesity (BMI ≥ 25 kg/m2). To the best of our knowledge, our study is the first to comprehensively examine how the relationship between various income parameters and T2D risk differs according to detailed obesity categories.
With respect to the obesity status, compared to those with prominent obesity, the impact of sustained very low- and low-income status was more pronounced in the population without obesity and not at risk of T2D. In our analysis, among individuals with underweight, sustained exposure to a very low-income status was associated with a 2.184-fold higher hazard of T2D when compared to nonexposure. This value is significantly higher than the HR of 1.253 associated with sustained exposure to very low-income status in individuals with a BMI ≥ 30.0 kg/m2. Moreover, the value was comparable to the HR of T2D (1.843) observed in overweight individuals (BMI 23.0–24.9 kg/m2) who had never experienced a very low-income status when compared to their underweight counterparts. The key to preventing the increasing global burden of T2D is identifying high-risk individuals who could benefit from diabetes screening and prevention, similar to considering BMI as a significant factor. Our findings indicate that individuals experiencing sustained very low-income status for 5 years face a risk of T2D similar to those who are overweight (BMI ≥ 23 kg/m2), despite not being at risk of diabetes due to obesity.
Several factors may explain the strong relationship between low-income status and excess T2D risk in individuals without obesity. First, unhealthy dietary and physical activity patterns may have contributed [24, 25]. Earlier studies have hypothesized that poor diet quality and food insecurity, despite a lower overall caloric intake, play significant role [26, 27]. Individuals with lower-incomes are more likely to choose cheaper, calorie-rich foods over a balanced diet with protein and fiber. Additionally, the financial and time constraints necessary for maintaining employment may hinder regular exercise, potentially leading to reduced muscle mass [28]. Stress might force them into harmful habits such as physical inactivity [29, 30]. Thus, individuals consistently exposed to lower-income conditions may be more prone to developing unhealthy body composition than their nonexposed counterparts with similar BMI. The impact of differences in body composition may be more pronounced in individuals with a lower BMI than in those already living with obesity and at risk of T2D. In addition, C-reactive protein, an inflammatory marker, has also been reported to be associated with low socioeconomic status (SES), including low-income, resulting in unhealthy outcomes including T2D [31, 32]. The association of this elevated inflammation with low-income may be partly mediated by the development of obesity [32]. Further, regardless of exposure to low-income status, as obesity itself is closely related to systemic inflammation [33], the impact of low-income conditions may have been even more pronounced in participants without obesity.
The finding that sustained high-income status attenuated the risk of T2D, particularly in those with a BMI ≥ 30 kg/m2, is notable. A previous Korean study also reported a similar significant interaction, indicating that the reduction in T2D risk associated with sustained exposure to high-income status, was more pronounced in individuals with obesity compared to nonexposure [6]. However, as the authors had classified obesity status solely by the presence of obesity (BMI ≥ 25 kg/m²), only a minimal difference in HR (0.87 in individuals without obesity vs. 0.83 in those with obesity) was observed. Notably, through detailed stratification of obesity status, we found that 5 years of sustained high-income status reduced the hazard of T2D by twice as much in individuals with a BMI ≥ 30.0 kg/m2 (17.1% reduction) than those with a BMI < 18.5 kg/m2 (8.0% reduction). This suggests that the protective effects of regular exercise and high-quality, nutrient-balanced food beyond total calorie intake [34, 35] related to high-income conditions may be more pronounced in individuals with obesity and already at high risk for T2D. Additionally, individuals with obesity may generally be more susceptible to stigma and discrimination owing to negative stereotypes, which can lead to stress, reduced psychological resources, and mental health problems, potentially increasing the risk of adverse outcomes, including T2D [10]. However, those with a high SES who maintain a sustained high-income despite having obesity may experience much less social discrimination, which could provide an additional protective effect.
We also found that high income variability was associated with an increased T2D risk. This association was stronger in individuals in the advanced BMI category, with a 15.1% higher risk in those with a BMI ≥ 30 kg/m2. However, no such association was noted in the those with underweight. As in prior studies that reported the link between income variability and various adverse outcomes [36–41], those with high income variability might experience unpredictable and episodic low-income, which may lack the protective benefit of social resources and is associated with psychological depression [37]. According to previous systematic reviews and meta-analyses [10], obesity may be associated with subsequent income, with underlying mechanisms potentially including negative stereotypes about obesity and weight penalties in the labor market, leading to higher job insecurity. The high income variability in the population with obesity is likely to stem from these stigma effects more than in the population without obesity. Therefore, higher income variability may be closely linked to increased experiences of discrimination, maladaptive stress, and the loss of psychological resources, particularly in individuals with obesity. This could explain the more pronounced association between income variability and T2D risk observed in populations with obesity.
The limitations of this study should be acknowledged. First, clarifying the causal relationships was not possible due to the observational design. Second, as our study population included only Korean adults, extrapolation of the findings to other ethnicities or populations with different policies and/or healthcare systems need to be performed with caution. Third, although we demonstrated robust associations between income variability and T2D risk using multiple variability metrics (SD, CV, ARV, and VIM), these measures do not capture the directional nature of income changes (i.e., increasing vs. decreasing patterns), which could have distinct implications for T2D risk. Future studies incorporating trajectory analyses may better clarify the temporal and directional aspects of income dynamics in relation to diabetes risk. Finally, other social determinants of health, including education, occupation, and housing in addition to income, could not be captured due to the inherent structure of the KNHIS dataset. Because such information is unavailable in this administrative database, income level was the only feasible indicator that could be used as a proxy for socioeconomic status. Nevertheless, our study has several strengths. This study was based on a large sample size of > 3.4 million people from a representative nationwide cohort database run by the Korean government and included diverse variables of demographics, lifestyle, comorbidities, and laboratory findings. This permitted adjustment for these potential confounders. Furthermore, we examined the cumulative income status in terms of repeated serial measures, thus demonstrating a dose-response relationship. Additionally, through detailed stratified analyses according to obesity status, the findings emphasized the implications of various income parameters as potential risk factors for the incidence of T2D across different obesity statuses.
Conclusions
In conclusion, this real-world, nationwide, population-based study of Korean adults demonstrated that various income parameters significantly affect the risk of T2D. Additionally, the data reveal that the effects of these factors vary by obesity status, with individuals without obesity being more affected by low-income exposure. Among those with obesity, sustained high income appears to mitigate the risk more prominently, while they remain more vulnerable to income variability. These findings underscore the importance of considering both socioeconomic and obesity status in diabetes prevention strategies.
Supplementary Information
Additional file 1: Figure S1. The dose-response relationship between baseline income percentiles and the risk of incident type 2 diabetes Table S1. Definitions of income parameters. Table S2. Definitions of covariates. Table S3. The dose-response relationship between baseline income percentiles and the risk of incident type 2 diabetes. Table S4. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income variability. Table S5. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters in individuals aged <65 years. Table S6. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters in individuals aged 65 years or older. Table S7. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥25 kg/m2). Table S8. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income variability, stratified by the presence of obesity (BMI ≥25 kg/m2). Table S9. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥ 25 kg/m2) in individuals aged <65 years. Table S10. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥25 kg/m2) in individuals aged 65 years or older. Table S11. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameter groups, stratified analyses by the BMI categories. Table S12. Sensitivity analysis after excluding participants with myocardial infarction, stroke, and/or cancer: Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters. Table S13. Sensitivity analysis after excluding participants with myocardial infarction, stroke, and/or cancer: Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥25 kg/m2). Table S14. Sensitivity analysis accounting for all-cause mortality as a competing event: Sub-distribution hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters. Table S15. Sensitivity analysis accounting for all-cause mortality as a competing event: Sub-distribution hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥25 kg/m2). Table S16. Sensitivity analysis after excluding individuals who died or developed T2D within two years after baseline data collection: Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters. Table S17. Sensitivity analysis after excluding individuals who died or developed T2D within two years after baseline data collection: Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥25 kg/m2)
Acknowledgements
This study was conducted in collaboration with the Korean National Health Insurance Service (KNHIS). The National Health Information Database constructed by the KNHIS was used and the study results do not necessarily represent the opinions of the KNHIS.
Abbreviations
- T2D
Type 2 diabetes
- SDOH
Social determinants of health
- BMI
Body mass index
- KNHIS
Korean National Health Insurance Service
- NHI
National Health Insurance
- MA
Medical Aid
- ICD-10
International Classification of Diseases-10th Revision
- FPG
Fasting plasma glucose
- eGFR
Estimated glomerular filtration rate
- IRB
Institutional Review Board
- CKD
Chronic kidney disease
- Cis
Confidence intervals
- HRs
Hazard ratios
- SES
Socioeconomic status
Authors' contributions
J.H.Y. and Y-B.L. drafted the manuscript. K.H., and K. L. contributed to study design and participated in data analysis planning and statistical analysis. Y-B.L., J.H.Y. and J.E.J. performed literature searches and contributed to conception of the hypothesis. J.H.Y. and Y-B.L. critically edited the manuscript. All authors contributed important intellectual content during manuscript drafting or revision and approved the final version of the manuscript.
Funding
This research was supported by the Chung-Ang University Research Grants in 2024. The funders had no role in the study design, data collection and analysis, decision to publish, or manuscript preparation.
Data availability
The data that support the findings of this study are available from the Korean National Health Insurance Service (KNHIS) but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the corresponding authors upon reasonable request and with permission of the KNHIS.
Declerations
Ethics approval and consent to participate
The study protocol was approved by the Institutional Review Board (IRB) of the Samsung Medical Center (SMC 2023-11-025). This study utilized anonymized public data that had already been secured for other purposes and only involved reviewing this anonymized database without any additional interventions. Thus, it falls under the category of minimal risk research with no additional risk to the subjects. Furthermore, the researchers received data from the KNHIS with all personal identification information removed, making it impossible to identify the participants included. As such, given that the data provided by the KNHIS was anonymized, the requirement for informed consent was waived by the IRB. After obtaining IRB approval, we accessed the data under the permission of the KNHIS for a limited period of utilization. All methods were performed in accordance with the relevant guidelines and regulations. The study was conducted in accordance with the Declaration of Helsinki.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
J.H.Y and Y.-B.L contributed equally to this work as cofirst authors.
Contributor Information
You-Bin Lee, Email: yb.snowyday1224@gmail.com.
Kyungdo Han, Email: hkd917@naver.com.
References
- 1.Hossain MJ, Al-Mamun M, Islam MR. Diabetes mellitus, the fastest growing global public health concern: early detection should be focused. Health Sci Rep. 2024;7(3):e2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.American diabetes association. Economic costs of diabetes in the U.S. in 2017. Diabetes Care. 2018;41(5):917–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Dall TM, Yang W, Gillespie K, Mocarski M, Byrne E, Cintina I, et al. The economic burden of elevated blood glucose levels in 2017: diagnosed and undiagnosed diabetes, gestational diabetes mellitus, and prediabetes. Diabetes Care. 2019;42(9):1661–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Reed J, Bain S, Kanamarlapudi V. A review of current trends with type 2 diabetes epidemiology, aetiology, pathogenesis, treatments and future perspectives. Diabetes Metab Syndr Obes. 2021;14:3567–602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hill-Briggs F, Adler NE, Berkowitz SA, Chin MH, Gary-Webb TL, Navas-Acien A, et al. Social determinants of health and diabetes: a scientific review. Diabetes Care. 2020;44(1):258–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Park JC, Nam GE, Yu J, McWhorter KL, Liu J, Lee HS, et al. Association of sustained low or high income and income changes with risk of incident type 2 diabetes among individuals aged 30 to 64 years. JAMA Netw Open. 2023;6(8):e2330024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Lysy Z, Booth GL, Shah BR, Austin PC, Luo J, Lipscombe LL. The impact of income on the incidence of diabetes: a population-based study. Diabetes Res Clin Pract. 2013;99(3):372–9. [DOI] [PubMed] [Google Scholar]
- 8.Reyes Matos U, Mesenburg MA, Victora CG. Socioeconomic inequalities in the prevalence of underweight, overweight, and obesity among women aged 20–49 in low- and middle-income countries. Int J Obes (Lond). 2020;44(3):609–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Ueda P, Kondo N, Fujiwara T. The global economic crisis, household income and pre-adolescent overweight and underweight: a nationwide birth cohort study in Japan. Int J Obes (Lond). 2015;39(9):1414–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kim TJ, von dem Knesebeck O. Income and obesity: what is the direction of the relationship? a systematic review and meta-analysis. BMJ Open. 2018;8(1):e019862. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Moon JH, Jang Y, Oh TJ, Jung SY. The risk of type 2 diabetes mellitus according to changes in obesity status in late middle-aged adults: a nationwide cohort study of Korea. Diabetes Metab J. 2023;47(4):514–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Lee YH, Han K, Ko SH, Ko KS, Lee KU, Taskforce team of diabetes fact sheet of the korean diabetes association. Data analytic process of a nationwide population-based study using national health information database established by national health insurance service. Diabetes Metab J. 2016;40(1):79–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kim MK, Han K, Lee SH. Current trends of big data research using the korean national health information database. Diabetes Metab J. 2022;46(4):552–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Cheol Seong S, Kim YY, Khang YH, Heon Park J, Kang HJ, Lee H, et al. Data resource profile: the national health information database of the national health insurance service in South Korea. Int J Epidemiol. 2017;46(3):799–800. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Baek JH, Park YM, Han KD, Moon MK, Choi JH, Ko SH. Comparison of operational definition of type 2 diabetes mellitus based on data from korean national health insurance service and korea national health and nutrition examination survey. Diabetes Metab J. 2023;47(2):201–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Noh J, Han KD, Ko SH, Ko KS, Park CY. Trends in the pervasiveness of type 2 diabetes, impaired fasting glucose and co-morbidities during an 8-year-follow-up of nationwide Korean population. Sci Rep. 2017;7:46656. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Lee YB, Park SH, Lee KN, Kim B, Kwon SY, Park J, et al. Low household income status and death from pneumonia in people with type 2 diabetes mellitus: a nationwide study. Diabetes Metab J. 2023;47(5):682–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Park YM, Baek JH, Lee HS, Elfassy T, Brown CC, Schootman M, et al. Income variability and incident cardiovascular disease in diabetes: a population-based cohort study. Eur Heart J. 2024;45(21):1920–33. [DOI] [PubMed] [Google Scholar]
- 19.Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9):604–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.KDIGO. Chapter 1: Definition and classification of CKD. Kidney Int Suppl. (2011). 2013;3(1):19–62. [DOI] [PMC free article] [PubMed]
- 21.Kim MK, Lee WY, Kang JH, Kang JH, Kim BT, Kim SM, et al. 2014 clinical practice guidelines for overweight and obesity in Korea. Endocrinol Metab (Seoul). 2014;29(4):405–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Haam JH, Kim BT, Kim EM, Kwon H, Kang JH, Park JH, et al. Diagnosis of obesity: 2022 update of clinical practice guidelines for obesity by the korean society for the study of obesity. J Obes Metab Syndr. 2023;32(2):121–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94(446):496–509. [Google Scholar]
- 24.Jaacks LM, Vandevijvere S, Pan A, McGowan CJ, Wallace C, Imamura F, et al. The obesity transition: stages of the global epidemic. Lancet Diabetes Endocrinol. 2019;7(3):231–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Patja K, Jousilahti P, Hu G, Valle T, Qiao Q, Tuomilehto J. Effects of smoking, obesity and physical activity on the risk of type 2 diabetes in middle-aged Finnish men and women. J Intern Med. 2005;258(4):356–62. [DOI] [PubMed] [Google Scholar]
- 26.Sarlio-Lahteenkorva S, Lahelma E. Food insecurity is associated with past and present economic disadvantage and body mass index. J Nutr. 2001;131(11):2880–4. [DOI] [PubMed] [Google Scholar]
- 27.Te Vazquez J, Feng SN, Orr CJ, Berkowitz SA. Food insecurity and cardiometabolic conditions: a review of recent research. Curr Nutr Rep. 2021;10(4):243–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Rana JS, Li TY, Manson JE, Hu FB. Adiposity compared with physical inactivity and risk of type 2 diabetes in women. Diabetes Care. 2007;30(1):53–8. [DOI] [PubMed] [Google Scholar]
- 29.Algren MH, Ekholm O, Nielsen L, Ersboll AK, Bak CK, Andersen PT. Associations between perceived stress, socioeconomic status, and health-risk behaviour in deprived neighbourhoods in Denmark: a cross-sectional study. BMC Public Health. 2018;18(1):250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Norberg M, Stenlund H, Lindahl B, Andersson C, Eriksson JW, Weinehall L. Work stress and low emotional support is associated with increased risk of future type 2 diabetes in women. Diabetes Res Clin Pract. 2007;76(3):368–77. [DOI] [PubMed] [Google Scholar]
- 31.Rathmann W, Haastert B, Giani G, Koenig W, Imhof A, Herder C, et al. Is inflammation a causal chain between low socioeconomic status and type 2 diabetes? Results from the KORA Survey 2000. Eur J Epidemiol. 2006;21(1):55–60. [DOI] [PubMed] [Google Scholar]
- 32.Liu RS, Aiello AE, Mensah FK, Gasser CE, Rueb K, Cordell B, et al. Socioeconomic status in childhood and C reactive protein in adulthood: a systematic review and meta-analysis. J Epidemiol Community Health. 2017;71(8):817–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Visser M, Bouter LM, McQuillan GM, Wener MH, Harris TB. Elevated C-reactive protein levels in overweight and obese adults. JAMA. 1999;282(22):2131–5. [DOI] [PubMed] [Google Scholar]
- 34.Haramshahi M, TME AE, Daabo HMA, Altinkaynak Y, Hjazi A, Saxena A, et al. Nutrient patterns and risk of diabetes mellitus type 2: a case-control study. BMC Endocr Disord. 2024;24(1):10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Rietz M, Lehr A, Mino E, Lang A, Szczerba E, Schiemann T, et al. Physical activity and risk of major diabetes-related complications in individuals with diabetes: a systematic review and meta-analysis of observational studies. Diabetes Care. 2022;45(12):3101–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Kim S, Subramanian SV. Income volatility and depressive symptoms among elderly Koreans. Int J Environ Res Public Health. 2019;16(19):3580. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Prause J, Dooley D, Huh J. Income volatility and psychological depression. Am J Community Psychol. 2009;43(1–2):57–70. [DOI] [PubMed] [Google Scholar]
- 38.Grasset L, Glymour MM, Elfassy T, Swift SL, Yaffe K, Singh-Manoux A, et al. Relation between 20-year income volatility and brain health in midlife: the CARDIA study. Neurology. 2019;93(20):e1890–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Sosu EM, Schmidt P. Changes in cognitive outcomes in early childhood: the role of family income and volatility. Front Psychol. 2022;13:758082. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Lim G. Income volatility and increased risk of CVD. Nat Rev Cardiol. 2019;16(3):133. [DOI] [PubMed] [Google Scholar]
- 41.Elfassy T, Swift SL, Glymour MM, Calonico S, Jacobs DR Jr., Mayeda ER, et al. Associations of income volatility with incident cardiovascular disease and all-cause mortality in a US cohort. Circulation. 2019;139(7):850–9. [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
Additional file 1: Figure S1. The dose-response relationship between baseline income percentiles and the risk of incident type 2 diabetes Table S1. Definitions of income parameters. Table S2. Definitions of covariates. Table S3. The dose-response relationship between baseline income percentiles and the risk of incident type 2 diabetes. Table S4. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income variability. Table S5. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters in individuals aged <65 years. Table S6. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters in individuals aged 65 years or older. Table S7. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥25 kg/m2). Table S8. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income variability, stratified by the presence of obesity (BMI ≥25 kg/m2). Table S9. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥ 25 kg/m2) in individuals aged <65 years. Table S10. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥25 kg/m2) in individuals aged 65 years or older. Table S11. Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameter groups, stratified analyses by the BMI categories. Table S12. Sensitivity analysis after excluding participants with myocardial infarction, stroke, and/or cancer: Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters. Table S13. Sensitivity analysis after excluding participants with myocardial infarction, stroke, and/or cancer: Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥25 kg/m2). Table S14. Sensitivity analysis accounting for all-cause mortality as a competing event: Sub-distribution hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters. Table S15. Sensitivity analysis accounting for all-cause mortality as a competing event: Sub-distribution hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥25 kg/m2). Table S16. Sensitivity analysis after excluding individuals who died or developed T2D within two years after baseline data collection: Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters. Table S17. Sensitivity analysis after excluding individuals who died or developed T2D within two years after baseline data collection: Hazard ratios and 95% confidence intervals for incident type 2 diabetes according to income parameters, stratified by the presence of obesity (BMI ≥25 kg/m2)
Data Availability Statement
The data that support the findings of this study are available from the Korean National Health Insurance Service (KNHIS) but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the corresponding authors upon reasonable request and with permission of the KNHIS.