Abstract
Objective
To examine the impact of income and eating speed on new-onset diabetes among men.
Design
This was a retrospective cohort study.
Setting
We used the administrative claims and health check-up data for fiscal years 2010–2015 obtained from the Fukuoka branch of the Japan Health Insurance Association.
Participants
Participants were 15 474 non-diabetic male employees, aged between 40 and 74 years. They were categorised based on their eating speeds (ie, fast, normal and non-fast).
Primary and secondary outcome measures
To calculate the OR of the development of diabetes, we created generalised linear regression models with diabetes onset as the dependent variable and eating speed and income as covariates and calculated corresponding 95% CI values. The analyses were performed after adjusting the data for age, obesity and comorbidities.
Results
Of the total participants, 620 developed diabetes during the 5-year study period. A univariate analysis using the generalised linear regression model revealed that eating fast (OR: 1.35, 95% CI 1.17 to 1.55) and having a low income wereincome (OR: 1.47, 95% CI 1.24 to 1.74) were significantly associated with the onset of diabetes. After adjusting for age, obesity and comorbidities, both eating fast (OR: 1.17, 95% CI 1.02 to 1.35) and having a low income (OR: 1.24, 95% CI 1.03 to 1.50) were recognised as independent risk factors for diabetes onset.
Conclusions
The study revealed that eating fast and having a low income were independent risk factors, leading to the development of diabetes. While it is difficult to address income differences, it may be possible to address the factors that contribute to income differences to manage diabetes appropriately and at low healthcare costs. However, eating speed can be controlled. Hence, the provision of education and coaching on dietary habits, including eating speed, may be effective in preventing diabetes onset.
Keywords: health services administration & management, general diabetes, health policy, preventive medicine, social medicine
Strengths and limitations of this study.
We conducted a retrospective cohort study on a sample of 15 474 participants.
We followed up the participants’ diabetes onset for a period of approximately 5 years.
We evaluated diabetes onset based on haemoglobin A1c levels, diagnoses and drug prescriptions according to health insurance claims.
Unlike other similar studies, our study considered eating speed and income together as factors affecting diabetes onset.
This study has some limitations, such as not considering possible changes in eating speed and income during the 5-year observation period.
Introduction
The Global Report on Diabetes (2016) indicated that approximately 422 000 000 individuals worldwide were affected by diabetes in 2014. It also reported that the diabetes incidence rate doubled from 4.7% in 1980 to 8.5% in 2014.1 In Japan, the 2018 National Health and Nutrition Survey reported that 18.7% and 9.3% of the total male and female population, respectively, were suspected to have diabetes.2 Furthermore, it was estimated that approximately 3 289 000 individuals (1 848 000 men and 1 442 000 women) underwent treatment for diabetes in 2017.3
Eating behaviour is a factor that contributes to both obesity and diabetes. Several studies have investigated the impact of eating speed in this context and demonstrated its association with obesity and diabetes.4–13 For instance, a Japanese study conducted on a large cohort of 197 825 participants over a period of 3 years reported that eating fast was the sole predisposing factor among several eating habits that led to new-onset diabetes.5 Studies have found that obesity,8 insulin resistance14 15 and postprandial glucose metabolism,16 which are all affected by eating speed, influence one another and cause diabetes. However, in a survey conducted on 2050, male Japanese employees of a metal product factory, although eating speed was associated with the incidence of diabetes, it was found to be insignificant once adjustments were made to account for body mass index (BMI).9
Furthermore, lifestyle, which encompasses eating habits such as eating speed, is affected by not only personal choices but also socioeconomic factors. The 2018 National Health and Nutrition Survey conducted by the Ministry of Health, Labour and Welfare demonstrated that the risks of occurrence of obesity and stroke are approximately 1.5 times higher in low-income than high-income groups.2 Furthermore, the 2010 Japan Gerontological Evaluation Study demonstrated that the risk of diabetes was negatively correlated with income among individuals aged over 65 years who did not qualify for the Certification of Needed Long-Term Care.17 Collectively, these findings suggest that societal factors can contribute to the onset of diabetes.
Personal lifestyle factors, such as eating and exercising habits, and the societal factors that determine individuals’ personal lifestyle choices contribute to the development of lifestyle diseases, including diabetes. However, there is a lack of evidence on the impact of personal lifestyle and socioeconomic factors on diabetes. A cross-sectional study conducted by Nagamine et al17 evaluated the association between diabetes and socioeconomic factors, such as income, education and employment, and adjusted their data to account for the time that participants spent walking per day and their intake of meat, fish and vegetables. However, to date, no study has simultaneously evaluated the association between eating speed and income on diabetes onset. Simultaneously evaluating these factors makes it possible to investigate whether income affects the association between eating speed and diabetes onset. To bridge this research gap, we conducted a 5-year retrospective cohort study to examine the impact of eating speed and income on diabetes onset in male employees of small and medium-sized enterprises (SMEs).
Methods
Data source
In this study, we collected specific health check-up and claims data from the Fukuoka branch of the Japanese Health Insurance Association, which insured employees of SMEs. The participants’ health check-up data included details like check-up dates, BMI values, and the results of blood chemical analysis. They included a self-reported questionnaire to obtain the participants’ responses to several questions on medical history, comorbidities and lifestyle factors, like eating habits.18 The claims data included information on the dates of consultations, treatments and participants’ details like gender, age, diagnoses, specific treatments and healthcare expenditure. Furthermore, the health check-up and claims data were linked at the individual subject level for analysis.
Study population
We selected non-diabetic individuals from a group of 81 193 male employees who had undergone a specific health check-up documented at the Fukuoka branch of the Japan Health Insurance Association in 2010. This study only included men since only a small proportion of those who are insured with the Fukuoka branch of the Japan Health Insurance Association are women (36.9%) and since previous studies have reported that women eat slower than men.4 12 Additionally, since the specific health check-up is legally required of those aged 40–74 years, the participants of this study are also aged 40–74 years. Individuals were excluded from the study if (1) they reported a history of cerebrovascular disease, cardiac disease, renal failure and/or dialysis, (2) sufficient data on their eating speed were not available or (3) adequate data on their income were not available. Among the 19 214 men who met the inclusion criteria, 15 474 men, whose information could determine the presence of diabetes according to the 2015 check-up and health insurance claim, were eventually included in the analysis.
Outcome
The onset of diabetes was defined as the primary outcome in this study. Since the Japan Diabetes Society unit for haemoglobin A1c (HbA1c) measurement was used in specific health check-ups until 2011, for this study, we converted the HbA1c value measured in 2011 to the National Glycohemoglobin Standardization Program unit of measurement.19 Participants were considered diabetic if (1) their HbA1c (National Glycohemoglobin Standardization Program) values from the check-up exceeded 6.5% or (2) they were diagnosed with diabetes (International Classification of Diseases 10th revision codes: E10–14) according to the health insurance claim, and they were administered either an oral antidiabetic medication or insulin.
Variables
The exposure variables consisted of: eating speed and income. Based on their responses to the self-reported questionnaire, participants were categorised as having fast (the fast eater group), slow or normal (the non-fast eater group) eating speeds. Furthermore, according to the information on personal standard monthly income, the participants were categorised into those with income US$<2500; income=US$2500 to 3499 and income US$≥3500. This categorisation was made based on the information that the average income in Japan in 2010 was approximately US$3050 (US$1=Kr100).20
The covariates were selected from factors thought to influence the onset of diabetes. Participants were categorised into three age groups (40–49, 50–59 and ≥60 years of age) based on their age as on 1 April 2010. Based on participants’ health check-up records, obesity was defined as having a BMI ≥25 kg/m2.21 Comorbidities were noted if participants reported taking medications for hypertension and/or dyslipidaemia.
Statistical analysis
Initially, we conducted a χ2 test to compare participants’ characteristics, including age, obesity, comorbidity, income, eating speed and the onset of diabetes, over the past 5 years. The χ2 test also helped compare the outcomes in the three income groups. Subsequently, to calculate the OR of the development of diabetes among participants, generalised linear regression models were created with diabetes onset as the dependent variable and eating speed and income as covariates, and 95% CI values were calculated. The analyses were performed after adjusting for age, obesity and comorbidity. Model 1 was adjusted for eating speed and income; model 2 was adjusted for age added to model 1; model 3 was adjusted for obesity added to model 1; model 4 was adjusted for age and obesity added to model 1 and model 5 was adjusted for comorbidities added to model 4. The same analysis was also conducted for each age group to assess the influence of age on eating speed and income level. Model 1 was adjusted for eating speed and income; model 2 was adjusted for obesity added to model 1 and model 3 was adjusted for comorbidities added to model 2. The following were used as reference categories for covariates: non-fast eating speed and a standard monthly income US$≥3500. Furthermore, the log-likelihood ratio test, Akaike’s information criterion and Bayesian information criterion were used as indicators to determine how well the generalised linear regression model fitted the data. Data were analysed using Stata V.15.0 (Stata Corp, College Station, Texas), and statistical significance was set at p<0.05.
Patient and public involvement statement
We used administrative claim data and health check-up data and, hence, patients were not directly involved in this study.
Results
Table 1 summarises the characteristics of participants categorised by eating speed. It was found that 5866 participants (37.5%) were fast eaters, and the remaining 9608 (62.1%) were non-fast eaters. Fast eaters were more likely to be younger, obese (BMI ≥25 kg/m2), have one or more comorbidities and have a higher income (p<0.001) than non-fast eaters. The 5-year risk of developing diabetes was significantly higher in the fast eater group than in the non-fast eater group (p<0.001).
Table 1.
Number and proportion of participants by eating speed
| Speed of eating category | Total | P value | ||||
| Fast | Non-fast | |||||
| Total | n=5866 | (37.5%) | n=9608 | (62.1%) | n=15 474 | |
| Age, year | ||||||
| 40–49 | 3345 | (57.0%) | 4852 | (50.5%) | 8197 | <0.001 |
| 50–59 | 1976 | (33.7%) | 3609 | (37.6%) | 5585 | |
| ≥60 | 545 | (9.3%) | 1147 | (11.9%) | 1692 | |
| Obese | ||||||
| BMI <25 | 3641 | (62.1%) | 7271 | (75.7%) | 10 912 | <0.001 |
| BMI≥25 | 2225 | (37.9%) | 2337 | (24.3%) | 4562 | |
| Hypertension with medication | ||||||
| Yes | 1815 | (30.9%) | 2673 | (27.8%) | 4488 | <0.001 |
| No | 4051 | (69.1%) | 6935 | (72.2%) | 10 986 | |
| Dyslipidaemia with medication | ||||||
| Yes | 2682 | (45.7%) | 3804 | (39.6%) | 6486 | <0.001 |
| No | 3183 | (54.3%) | 5804 | (60.4%) | 8987 | |
| Standard monthly remuneration category | ||||||
| <2500 USD | 952 | (16.2%) | 1979 | (20.6%) | 2931 | <0.001 |
| 2500–3499 USD | 1719 | (29.3%) | 3128 | (32.6%) | 4847 | |
| ≥3500 USD | 3195 | (54.5%) | 4501 | (46.8%) | 7696 | |
| Diabetes onset | ||||||
| Yes | 278 | (4.7%) | 342 | (3.6%) | 620 | <0.001 |
| No | 5588 | (95.3%) | 9266 | (96.4%) | 14 854 | |
P values were calculated using χ2 test.
1 USD=100 Japanese Yen.
BMI, body mass index.
Table 2 summarises the characteristics of participants categorised by income. According to the table, 2931 participants (18.9%) had a monthly income US$<2500; 4847 (31.3%) had a monthly income between US$2500 and US$3499; and the remaining 7696 participants (49.7%) had a monthly income US$>3500. The low-income group was more likely to be aged over 60 years. Additionally, they had BMI <25 kg/m2, hypertension and were more likely to develop diabetes than the middle-income and high-income groups (p<0.001).
Table 2.
Number and proportion of participants by standard monthly remuneration category
| Standard monthly remuneration category | P value | ||||||||
| US$<2500 | US$2500–3499 | US$≥3500 | Total | ||||||
| Total | n=2931 | (18.9%) | n=4847 | (31.3%) | n=7696 | (49.7%) | n=15 474 | (100.0%) | |
| Age, year | |||||||||
| 40–49 | 777 | (26.5%) | 2885 | (59.5%) | 4535 | (58.9%) | 8197 | (51.8%) | <0.001 |
| 50–59 | 1151 | (39.3%) | 1610 | (33.2%) | 2824 | (36.7%) | 5585 | (35.3%) | |
| ≥60 | 1003 | (34.2%) | 352 | (7.3%) | 337 | (4.4%) | 1692 | (10.7%) | |
| Obese | |||||||||
| BMI <25 | 2143 | (73.1%) | 3458 | (71.3%) | 5311 | (69.0%) | 10 912 | (69.0%) | <0.001 |
| BMI≥25 | 788 | (26.9%) | 1389 | (28.7%) | 2385 | (31.0%) | 4562 | (28.8%) | |
| Hypertension with medication | |||||||||
| Yes | 1086 | (37.1%) | 1348 | (27.8%) | 2054 | (26.7%) | 4488 | (28.4%) | <0.001 |
| No | 1845 | (62.9%) | 3499 | (72.2%) | 5642 | (73.3%) | 10 986 | (69.5%) | |
| Dyslipidaemia with medication | |||||||||
| Yes | 1208 | (41.2%) | 2013 | (41.5%) | 3265 | (42.4%) | 6486 | (41.0%) | <0.001 |
| No | 1723 | (58.8%) | 2834 | (58.5%) | 4430 | (57.6%) | 8987 | (56.8%) | |
| Category of eating speed | |||||||||
| Fast | 952 | (32.5%) | 1719 | (35.5%) | 3195 | (41.5%) | 5866 | (37.1%) | <0.001 |
| Non-fast | 1979 | (67.5%) | 3128 | (64.5%) | 4501 | (58.5%) | 9608 | (60.7%) | |
| Diabetes onset | |||||||||
| Yes | 157 | (5.4%) | 182 | (3.8%) | 281 | (3.7%) | 864 | (5.5%) | <0.001 |
| No | 2774 | (94.6%) | 4665 | 96.2 | 7415 | (96.3%) | 14 954 | (94.5%) | |
P values were calculated using χ2 test.
BMI, body mass index.
Table 3 depicts the risk of diabetes (OR and 95% CI data) as determined by the generalised linear regression model. Univariate analysis demonstrated that eating fast (OR: 1.35, 95% CI 1.17 to 1.55) and having a low income (OR: 1.47, 95% CI 1.24 to 1.74) were significantly associated with the onset of diabetes. Additionally, multivariate analysis revealed that the OR of the low-income group decreased by 0.30 from 1.51 (95% CI 1.27 to 1.80) to 1.21 (95% CI 1.01 to 1.46) once adjustment was made to account for age (model 2). Furthermore, it was noted that after adjusting for obesity (model 3), the OR of eating fast decreased by 0.21 from 1.41 (95% CI 1.23 to 1.62) to 1.20 (95% CI 1.04 to 1.39). In each model, after adjusting for age, obesity and comorbidities, both eating fast and having a low-income were recognised as independent risk factors for diabetes onset.
Table 3.
ORs and 95% CIs for the onset of diabetes
| Univariate | Multivariate | |||||||||||||||||
| Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | ||||||||||||||
| OR | 95% CI | P value | OR | 95% CI | P value | OR | 95% CI | P value | OR | 95% CI | P value | OR | 95% CI | P value | OR | 95% CI | P value | |
| Category of eating speed | ||||||||||||||||||
| Fast | 1.35 | 1.17 to 1.55 | <0.001 | 1.38 | 1.20 to 1.59 | <0.001 | 1.41 | 1.23 to 1.62 | <0.001 | 1.17 | 1.02 to 1.35 | 0.027 | 1.20 | 1.04 to 1.39 | 0.010 | 1.17 | 1.02 to 1.35 | 0.028 |
| Non-fast | (Reference) | (Reference) | (Reference) | (Reference) | (Reference) | (Reference) | ||||||||||||
| Standard monthly remuneration category | ||||||||||||||||||
| <US$2500 | 1.47 | 1.24 to 1.74 | <0.001 | 1.51 | 1.27 to 1.80 | <0.001 | 1.21 | 1.01 to 1.46 | 0.042 | 1.58 | 1.33 to 1.88 | <0.001 | 1.26 | 1.04 to 1.52 | 0.017 | 1.24 | 1.03 to 1.50 | 0.025 |
| US$2500–3499 | 1.07 | 0.91 to 1.26 | 0.398 | 1.09 | 0.93 to 1.28 | 0.277 | 1.08 | 0.92 to 1.27 | 0.330 | 1.11 | 0.95 to 1.31 | 0.197 | 1.10 | 0.94 to 1.30 | 0.233 | 1.09 | 0.93 to 1.29 | 0.285 |
| ≥US$3500 | (Reference) | (Reference) | (Reference) | (Reference) | (Reference) | (Reference) | ||||||||||||
| Log-likelihood | −3332.2 | −3311.3 | −3219.3 | −3195.5 | −3168.7 | |||||||||||||
| AIC | 6672.4 | 6632.5 | 6448.6 | 6403.1 | 6353.4 | |||||||||||||
| BIC | 6703.1 | 6670.9 | 6486.9 | 6449.1 | 6414.8 | |||||||||||||
Model 1 was adjusted for category of eating speed and standard monthly remuneration category.
Model 2 was adjusted for category of eating speed, standard monthly remuneration category and age.
Model 3 was adjusted for category of eating speed, standard monthly remuneration category and obese.
Model 4 was adjusted for category of eating speed, standard monthly remuneration category, age and obese.
Model 5 was adjusted for category of eating speed, standard monthly remuneration category, age, obese, hypertension with medication and dyslipidaemia with medication.
AIC, Akaike’s information criterion; BIC, Bayesian information criterion.
Table 4 shows the results of the same analysis performed on age, after adjusting for obesity and comorbidities. After adjusting for obesity and comorbidities (model 3), eating fast and having low income were recognised as significant risk factors for diabetes onset for participants only in their 40s, but none was found for those in their 50s and 60s or older.
Table 4.
ORs and 95% CIs for the onset of diabetes by age
| Univariate | Multivariate | |||||||||||
| Model 1 | Model 2 | Model 3 | ||||||||||
| OR | 95% CI | P value | OR | 95% CI | P value | OR | 95% CI | P value | OR | 95% CI | P value | |
| Age: 40–49 year | ||||||||||||
| Category of eating speed | ||||||||||||
| Fast | 1.62 | 1.27 to 2.06 | <0.001 | 1.65 | 1.29 to 2.10 | <0.001 | 1.34 | 1.05 to 1.71 | 0.020 | 1.31 | 1.03 to 1.68 | 0.030 |
| Non-fast | (Reference) | (Reference) | (Reference) | (Reference) | ||||||||
| Standard monthly remuneration category | ||||||||||||
| <US$2500 | 1.46 | 0.99 to 2.14 | 0.054 | 1.52 | 1.04 to 2.23 | 0.033 | 1.51 | 1.03 to 2.23 | 0.036 | 1.49 | 1.01 to 2.20 | 0.045 |
| US$2500–3499 | 1.18 | 0.91 to 1.53 | 0.212 | 1.21 | 0.93 to 1.57 | 0.151 | 1.23 | 0.95 to 1.60 | 0.122 | 1.21 | 0.93 to 1.58 | 0.149 |
| ≥US$3500 | (Reference) | (Reference) | (Reference) | (Reference) | ||||||||
| Log-likelihood | −1193.6 | −1133.4 | −1116.9 | |||||||||
| AIC | 2395.2 | 2276.7 | 2247.8 | |||||||||
| BIC | 2423.3 | 2311.8 | 2296.9 | |||||||||
| Age: 50–59 year | ||||||||||||
| Category of eating speed | ||||||||||||
| Fast | 1.40 | 1.08 to 1.82 | 0.011 | 1.40 | 1.08 to 1.82 | 0.011 | 1.19 | 0.91 to 1.55 | 0.196 | 1.15 | 0.88 to 1.50 | 0.301 |
| Non-fast | (Reference) | (Reference) | (Reference) | (Reference) | ||||||||
| Standard monthly remuneration category | ||||||||||||
| <US$2500 | 1.28 | 0.94 to 1.75 | 0.118 | 1.32 | 0.96 to 1.80 | 0.085 | 1.36 | 0.99 to 1.86 | 0.057 | 1.33 | 0.97 to 1.83 | 0.075 |
| US$2500–3499 | 0.84 | 0.61 to 1.16 | 0.289 | 0.87 | 0.63 to 1.19 | 0.384 | 0.90 | 0.65 to 1.23 | 0.503 | 0.89 | 0.65 to 1.23 | 0.479 |
| ≥US$3500 | (Reference) | (Reference) | (Reference) | (Reference) | ||||||||
| Log-likelihood | −996.7 | −967.3 | −958.5 | |||||||||
| AIC | 2001.4 | 1944.6 | 1930.9 | |||||||||
| BIC | 2027.9 | 1977.7 | 1977.3 | |||||||||
| Age: over 60 year | ||||||||||||
| Category of eating speed | ||||||||||||
| Fast | 0.84 | 0.54 to 1.31 | 0.438 | 0.84 | 0.54 to 1.31 | 0.435 | 0.80 | 0.51 to 1.25 | 0.320 | 0.78 | 0.50 to 1.22 | 0.282 |
| Non-fast | (Reference) | (Reference) | (Reference) | (Reference) | ||||||||
| Standard monthly remuneration category | ||||||||||||
| <US$2500 | 0.96 | 0.57 to 1.60 | 0.868 | 0.95 | 0.57 to 1.59 | 0.843 | 0.99 | 0.59 to 1.67 | 0.983 | 0.99 | 0.59 to 1.65 | 0.955 |
| US$2500–3499 | 0.91 | 0.48 to 1.70 | 0.761 | 0.90 | 0.48 to 1.70 | 0.750 | 0.91 | 0.48 to 1.71 | 0.767 | 0.90 | 0.48 to 1.69 | 0.738 |
| ≥US$3500 | (Reference) | (Reference) | (Reference) | (Reference) | ||||||||
| Log-likelihood | −382.2 | −378.6 | −375.9 | |||||||||
| AIC | 772.5 | 767.2 | 765.9 | |||||||||
| BIC | 794.2 | 794.4 | 803.9 | |||||||||
Model 1 was adjusted for category of eating speed and standard monthly remuneration category.
Model 2 was adjusted for category of eating speed, standard monthly remuneration category and obese.
Model 3 was adjusted for category of eating speed, standard monthly remuneration category, obese, hypertension with medication and dyslipidaemia with medication.
AIC, Akaike’s information criterion; BIC, Bayesian information criterion.
Discussion
We conducted a 5-year retrospective cohort study on male employees of SMEs to examine how eating speed and income affected diabetes onset. The results indicated that eating fast and having a low income were independent risk factors for the development of diabetes.
The study investigated three mechanisms in the context of the association between eating speed and diabetes, namely, obesity, insulin resistance and postprandial glucose metabolism.7 9–11 14–16 Sakurai et al9 reported that the associations between eating speed and diabetes incidence were not significant in models once they were additionally adjusted for BMI. Furthermore, they mentioned that eating speed might be associated with diabetes since it affects body weight gain in a causal pathway and any BMI adjustment could be an overadjustment. In the current study, the OR of diabetes onset was reduced by 21% after BMI adjustment, and the results were similar to the findings of an earlier study that suggested that obesity affects the relationship between eating speed and diabetes onset.8 Additionally, eating fast was found to be an independent risk factor for the development of diabetes even after adjustments were made for other factors like BMI and income. Kudo et al5 reported that fast eaters have a higher risk of developing diabetes, irrespective of whether they undergo an increase in body weight. This can be due to the following reasons: first, a high eating speed reduces the consumption of energy after meals.22 Second, a decrease in the number of chews significantly increases postchewing blood glucose and insulin levels.23 In this study, insulin resistance and postprandial glucose metabolism have not been investigated. However, it is noted that these mechanisms affect each other and cause diabetes; hence, further studies should be conducted on these mechanisms.
Furthermore, like many previous studies, the study identified low-income as a risk factor for new-onset diabetes.17 24 25 In a Swedish study by Agardh et al,26 it was found that BMI explained the low-income group’s excessive risk for subjectively diagnosed type 2 diabetes. Nonetheless, some other studies that investigated the association between diabetes and socioeconomic status found that BMI did not explain the associations between income and diabetes.17 25 Similarly, in the current study, BMI did not explain the associations between income and diabetes; low income was an independent risk factor for diabetes even after adjustment was made for BMI.
As reported by Nishi et al,27 individuals with low income tend to avoid regular check-ups and, consequently, have poorer control over their blood sugar levels than those with high income. Individuals with low income have limited access to materials and services, such as a balanced diet and healthcare facilities required to prevent diabetes onset. Such limitations can cause social isolation, which can act as a source of psychological stress. Stress not only negatively impacts health behaviour but also has a direct effect on the stress hormone levels that regulate blood sugar levels and insulin resistance.28 In Japan, households with low-income consume more staple foods but less vegetable, fish and fruit compared with those with high income.29 Furthermore, it has been reported that people with low-income are less likely to eat a balanced diet and less likely to realise its importance.2 30 Moreover, they are more likely to consume foods with a high glycaemic index and high calories, such as fast food, sweet bread and soft drinks that can easily satisfy hunger.31–33 This can potentially lead to high levels of obesity and postprandial glucose metabolism and make people with low income more prone to diabetes than those with high income.
The OR of eating speed and low income for diabetes onset was increased significantly among participants in their 40s. Compared with the whole, the OR of eating fast in participants in their 40s increased by 14%, and the OR of low-income participants increased by 25%. It has been reported that people who eat fast tend to be young and have a high BMI.5 9 34 Furthermore, the proportion of eating fast was higher in the younger generation, which may indicate that the influence of BMI was greater in the younger generation. Conversely, low income was not affected by BMI or comorbidities. This suggests that the impact of low income on diabetes onset in the younger generation is more likely due to restricted access to health services rather obesity. Additionally, no significant OR for the onset of diabetes was observed for both eating fast and low income in patients those aged 60 years and older. Furthermore, although the proportion of participants aged 60 and over with a low income was large, the proportion of those who eat fast was small, and it is considered that they may have been less affected by low income, such as guaranteed access to medical care compared with younger people.
In this study, people who eat fast were found to be at increased risk of diabetes, despite the small number of low-income participants. Contrarily, low-income participants were at high risk of diabetes, although fewer people reported eating fast. Thus, the risk factors for diabetes, fast eating and low income are thought to be inversely related to each other. However, since each factor was an independent risk factor even after adjusting the covariates, it was suggested that eating fast is not affected by low-income on diabetes onset, and the mechanism by which eating fast and low income affect diabetes onset is different.
This study has several limitations. First, it did not consider any objective measure of eating speed and used only a self-reported questionnaire to determine each participant’s eating speed. However, self-reported eating speed has been shown to correlate well with those reported by friends or objectively measured ones.35 36 Furthermore, the questionnaire is clinically validated37 and has been used in the Japanese Specific Health Check and Guidance System that aims to prevent obesity.35 Second, we were unable to assess changes in eating speed during the study period. We evaluated the effect of eating speed on new-onset diabetes based on the self-reported data on eating speed collected in 2010. However, participants could have changed their eating speed over the 5-year period, and we were not able to determine whether such a change could have affected diabetes onset. Nevertheless, studies do suggest that individuals who are fast eaters at a young age will be fast eaters even as adults.7 Furthermore, the study’s participants were healthy working men who were unlikely to undergo any interventions that would alter their eating habits, including eating speed, during the 5-year study period. Third, we did not consider the changes in income with time since we considered only the data obtained in 2010 to categorise participants into income groups. However, this probably introduced little to no bias since changing jobs are uncommon in the Japanese population.38 Fourth, we did not consider the effects of factors such as family history and energy intake that may affect the progression of diabetes. The check-up results and claim data used in this study did not provide this information. Future research should consider including the information to obtain more precise results. Additionally, this study did not consider other socioeconomic factors other than industry and income, such as occupational class and educational background, which are associated with the development of obesity and diabetes. As the insurance data did not include these, they could not be considered in this study. However, given the information obtained from income, some occupational classes could be inferred. Finally, a selection bias might have occurred since the study was not a randomised controlled study. The participants of the study were people who regularly underwent medical check-ups. Therefore, they were more aware and conscious about their health compared with those who do not undergo medical check-ups on a regular basis.
Conclusions
We found that low income and eating speed were two risk factors for diabetes. Hence, lifestyle factors, such as dietary changes and instilling exercise habits, are crucial to avoid the onset of diabetes. The social factors that define lifestyle factors influence the onset of non-communicable diseases, like diabetes. While it is difficult to address income differences, our findings suggest that it is possible to address social factors such as differences in income and educational background to appropriately manage diabetes with low healthcare costs. However, eating speed is a risk factor that can be effectively controlled. Therefore, the provision of education and coaching on dietary habits, including eating speed, may be effective in preventing diabetes onset.
Supplementary Material
Footnotes
Contributors: RI led the study design, conducted literature search, extracted and analysed data, and wrote the manuscript. AB contributed to study design, data analysis and manuscript revision. NL and RY contributed to the study design and analysis. RI is the paper’s guarantor.
Funding: This work was supported by JSPS KAKENHI Grant Number 15K21568.
Competing interests: None declared.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review: Not commissioned; externally peer reviewed.
Data availability statement
No data are available. No additional data are available.
Ethics statements
Patient consent for publication
Not applicable.
Ethics approval
The study was approved by the Institutional Review Board of Kyushu University (Clinical Bioethics Committee of the Graduate School of Healthcare Sciences, Kyushu University) (approval number 20209).
References
- 1.WHO . Global report on diabetes, 2016. Available: https://www.who.int/publications/i/item/9789241565257
- 2.Ministry of Health LaW . National health and nutrition survey, 2019. Available: http://www.mhlw.go.jp/bunya/kenkou/kenkou_eiyou_chousa.html
- 3.Ministry of Health LaW. . Patient survey, 2017. Available: https://www.mhlw.go.jp/toukei/saikin/hw/kanja/17/index.html
- 4.Hurst Y, Fukuda H. Effects of changes in eating speed on obesity in patients with diabetes: a secondary analysis of longitudinal health check-up data. BMJ Open 2018;8:e019589. 10.1136/bmjopen-2017-019589 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kudo A, Asahi K, Satoh H, et al. Fast eating is a strong risk factor for new-onset diabetes among the Japanese general population. Sci Rep 2019;9:8210. 10.1038/s41598-019-44477-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Ohkuma T, Fujii H, Iwase M, et al. Impact of eating rate on obesity and cardiovascular risk factors according to glucose tolerance status: the Fukuoka diabetes registry and the Hisayama study. Diabetologia 2013;56:70–7. 10.1007/s00125-012-2746-3 [DOI] [PubMed] [Google Scholar]
- 7.Otsuka R, Tamakoshi K, Yatsuya H, et al. Eating fast leads to obesity: findings based on self-administered questionnaires among middle-aged Japanese men and women. J Epidemiol 2006;16:117–24. 10.2188/jea.16.117 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Radzevicienė L, Ostrauskas R. Fast eating and the risk of type 2 diabetes mellitus: a case-control study. Clin Nutr 2013;32:232–5. 10.1016/j.clnu.2012.06.013 [DOI] [PubMed] [Google Scholar]
- 9.Sakurai M, Nakamura K, Miura K, et al. Self-Reported speed of eating and 7-year risk of type 2 diabetes mellitus in middle-aged Japanese men. Metabolism 2012;61:1566–71. 10.1016/j.metabol.2012.04.005 [DOI] [PubMed] [Google Scholar]
- 10.Takayama S, Akamine Y, Okabe T, et al. Rate of eating and body weight in patients with type 2 diabetes or hyperlipidaemia. J Int Med Res 2002;30:442–4. 10.1177/147323000203000413 [DOI] [PubMed] [Google Scholar]
- 11.Tanihara S, Imatoh T, Miyazaki M, et al. Retrospective longitudinal study on the relationship between 8-year weight change and current eating speed. Appetite 2011;57:179–83. 10.1016/j.appet.2011.04.017 [DOI] [PubMed] [Google Scholar]
- 12.Zhu B, Haruyama Y, Muto T, et al. Association between eating speed and metabolic syndrome in a three-year population-based cohort study. J Epidemiol 2015;25:332–6. 10.2188/jea.JE20140131 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gudi SK. Eating speed and the risk of type 2 diabetes: explorations based on real-world evidence. Ann Pediatr Endocrinol Metab 2020;25:80–3. 10.6065/apem.2040028.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Boyko EJ, Leonetti DL, Bergstrom RW, et al. Low insulin secretion and high fasting insulin and C-peptide levels predict increased visceral adiposity. 5-year follow-up among initially nondiabetic Japanese-American men. Diabetes 1996;45:1010–5. 10.2337/diab.45.8.1010 [DOI] [PubMed] [Google Scholar]
- 15.Otsuka R, Tamakoshi K, Yatsuya H, et al. Eating fast leads to insulin resistance: findings in middle-aged Japanese men and women. Prev Med 2008;46:154–9. 10.1016/j.ypmed.2007.07.031 [DOI] [PubMed] [Google Scholar]
- 16.Jenkins DJ, Wolever TM, Ocana AM, et al. Metabolic effects of reducing rate of glucose ingestion by single bolus versus continuous sipping. Diabetes 1990;39:775–81. 10.2337/diab.39.7.775 [DOI] [PubMed] [Google Scholar]
- 17.Nagamine Y, Kondo N, Yokobayashi K, et al. Socioeconomic disparity in the prevalence of objectively evaluated diabetes among older Japanese adults: JAGES cross-sectional data in 2010. J Epidemiol 2019;29:295–301. 10.2188/jea.JE20170206 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Health Service Bureau tMoH, Labour and Welfare . Standard health checkup and counseling guidance program (revised version), 2013. Available: http://www.mhlw.go.jp/stf/seisakunitsuite/bunya/kenkou_iryou/kenkou/seikatsu/
- 19., et alCommittee of the Japan Diabetes Society on the Diagnostic Criteria of Diabetes Mellitus . Report of the Committee on the classification and diagnostic criteria of diabetes mellitus. J Diabetes Investig 2010;1:2–20. 10.1007/s13340-010-0006-7 [DOI] [Google Scholar]
- 20.Center NS . e-Stat Kouseinenkin hoken Gyotaibetsu/Kibobetsu tekiyoujokyoshirabe (Survey on the Employees’ Pention Insurance by types of operation and scale), 2020. Available: https://www.e-stat.go.jp/stat-search/files?page=1&layout=datalist&toukei=00450464&tstat=000001064003&cycle=0&tclass1=000001069717&cycle_facet=cycle&second=1&second2=1
- 21.Ogawa W, Miyazaki S. Diagnosis criteria for obesity and obesity disease. Health Evaluation and Promotion 2015;42:301–6. 10.7143/jhep.42.301 [DOI] [Google Scholar]
- 22.Hamada Y, Kashima H, Hayashi N. The number of chews and meal duration affect diet-induced thermogenesis and splanchnic circulation. Obesity 2014;22:E62–9. 10.1002/oby.20715 [DOI] [PubMed] [Google Scholar]
- 23.Mattes RD, Considine RV. Oral processing effort, appetite and acute energy intake in lean and obese adults. Physiol Behav 2013;120:173–81. 10.1016/j.physbeh.2013.08.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Krishnan S, Cozier YC, Rosenberg L, et al. Socioeconomic status and incidence of type 2 diabetes: results from the black women's health study. Am J Epidemiol 2010;171:564–70. 10.1093/aje/kwp443 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Tang M, Chen Y, Krewski D. Gender-Related differences in the association between socioeconomic status and self-reported diabetes. Int J Epidemiol 2003;32:381–5. 10.1093/ije/dyg075 [DOI] [PubMed] [Google Scholar]
- 26.Agardh EE, Ahlbom A, Andersson T, et al. Explanations of socioeconomic differences in excess risk of type 2 diabetes in Swedish men and women. Diabetes Care 2004;27:716–21. 10.2337/diacare.27.3.716 [DOI] [PubMed] [Google Scholar]
- 27.Nishi T, Babazono A, Maeda T. Association between income levels and irregular physician visits after a health checkup, and its consequent effect on glycemic control among employees: a retrospective propensity score-matched cohort study. J Diabetes Investig 2019;10:1372–81. 10.1111/jdi.13025 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Berkman LF KI, Glymour MM. Social epidemiology. USA: Oxford University Press, 2014. [Google Scholar]
- 29.Nishi N, Horikawa C, Murayama N. Characteristics of food group intake by household income in the National health and nutrition survey, Japan. Asia Pac J Clin Nutr 2017;26:156–9. 10.6133/apjcn.102015.15 [DOI] [PubMed] [Google Scholar]
- 30.Fukuda Y, Hiyoshi A. High quality nutrient intake is associated with higher household expenditures by Japanese adults. Biosci Trends 2012;6:176–82. 10.5582/bst.2012.v6.4.176 [DOI] [PubMed] [Google Scholar]
- 31.Hulshof KFAM, Brussaard JH, Kruizinga AG, et al. Socio-Economic status, dietary intake and 10 Y trends: the Dutch national food consumption survey. Eur J Clin Nutr 2003;57:128–37. 10.1038/sj.ejcn.1601503 [DOI] [PubMed] [Google Scholar]
- 32.Mark S, Lambert M, O'Loughlin J, et al. Household income, food insecurity and nutrition in Canadian youth. Can J Public Health 2012;103:94–9. 10.1007/BF03404210 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Sausenthaler S, Standl M, Buyken A, et al. Regional and socio-economic differences in food, nutrient and supplement intake in school-age children in Germany: results from the GINIplus and the LISAplus studies. Public Health Nutr 2011;14:1724–35. 10.1017/S1368980010003575 [DOI] [PubMed] [Google Scholar]
- 34.Takahashi F, Hashimoto Y, Kawano R, et al. Eating fast is associated with nonalcoholic fatty liver disease in men but not in women with type 2 diabetes: a cross-sectional study. Nutrients 2020;12:2174. 10.3390/nu12082174 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Nagahama S, Kurotani K, Pham NM, et al. Self-Reported eating rate and metabolic syndrome in Japanese people: cross-sectional study. BMJ Open 2014;4:e005241. 10.1136/bmjopen-2014-005241 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Sasaki S, Katagiri A, Tsuji T, et al. Self-Reported rate of eating correlates with body mass index in 18-y-old Japanese women. Int J Obes Relat Metab Disord 2003;27:1405–10. 10.1038/sj.ijo.0802425 [DOI] [PubMed] [Google Scholar]
- 37.Tayama J, Ogawa S, Takeoka A, et al. Item response theory-based validation of a short form of the eating behavior scale for Japanese adults. Medicine 2017;96:e8334. 10.1097/MD.0000000000008334 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Ministry of Health LaW . Survey on workers who changed jobs, 2015. Available: https://www.mhlw.go.jp/toukei/list/6-18.html
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
No data are available. No additional data are available.