Original Article | Published:

Contributors to the obesity and hyperglycemia epidemics. A prospective study in a population-based cohort

International Journal of Obesity volume 35, pages 14421449 (2011) | Download Citation

Abstract

Objective:

Relatively unexplored contributors to the obesity and diabetes epidemics may include sleep restriction, increased house temperature (HT), television watching (TW), consumption of restaurant meals (RMs), use of air conditioning (AC) and use of antidepressant/antipsychotic drugs (ADs).

Design and Subjects:

In a population-based cohort (n=1597), we investigated the possible association among these conditions, and obesity or hyperglycemia incidence at 6-year follow-up. Subjects with obesity (n=315) or hyperglycemia (n=618) at baseline were excluded, respectively, 1282 and 979 individuals were therefore analyzed.

Results:

At follow-up, 103/1282 became obese; these subjects showed significantly higher body mass index, waist circumference, saturated fat intake, RM frequency, TW hours, HT, AC and AD use, and lower fiber intake, metabolic equivalent of activity in h per week (METS) and sleep hours at baseline. In a multiple logistic regression model, METS (odds ratio=0.94; 95% confidence interval (CI) 0.91–0.98), RMs (odds ratio=1.47 per meal per week; 1.21–1.79), being in the third tertile of HT (odds ratio=2.06; 1.02–4.16) and hours of sleep (odds ratio=0.70 per h; 0.57–0.86) were associated with incident obesity. Subjects who developed hyperglycemia (n=174/979; 17.8%) had higher saturated fat intake, RM frequency, TW hours, HT, AC and AD use at baseline and lower METS and fiber intake. In a multiple logistic regression model, fiber intake (odds ratio=0.97 for each g per day; 0.95–0.99), RM (1.49 per meal per week; 1.26–1.75) and being in the third tertile of HT (odds ratio=1.95; 1.17–3.26) were independently associated with incident hyperglycemia.

Conclusions:

Lifestyle contributors to the obesity and hyperglycemia epidemics may be regular consumption of RM, sleep restriction and higher HT, suggesting potential adjunctive non-pharmacological preventive strategies for the obesity and hyperglycemia epidemics.

Introduction

Obesity and diabetes have reached epidemic proportions in the United States and Europe. Changes in dietary habits and reductions in physical activity are the two most common explanations for the increasing burden of these diseases. Recently, attention has been directed to other contributors, that may influence the balance between energy expenditure and intake.1, 2, 3, 4, 5, 6, 7, 8 Furthermore, dietary and exercise habits are not easily modifiable, and the search for risk factors that are more amenable to change might be potentially interesting and relatively unexplored in European cohorts.

The average amount of sleep per night has been reported to be declining slightly,9 and an increased incidence of obesity and hyperglycemia has been noted along with this decrease in sleep, particularly in younger cohorts.10 This finding is probably because of the profound metabolic hormonal changes exerted by sleep debt and increased fatigue, leading to increased caloric intake and reduced energy expenditure, respectively.2, 3, 10 Other features of modern societies include psychosocial stress, indoor heating during cold seasons and air conditioning (AC) during warm seasons.

The use of antidepressants and atypical antipsychotics that are associated with weight gain has increased substantially in the last decade.1 These conditions, together with increasing television watching (TW) and consuming foods that have been prepared outside the home, might promote a decline in physical activity and overeating.4, 6

In a population-based cohort, we investigated possible associations among sleep restriction, house temperature (HT), hours of TW, consumption of restaurant foods, use of AC, use of antidepressant/antipsychotic drugs (ADs) and the risk of obesity and hyperglycemia, after an average follow-up period of 6 years.

Subjects and methods

All 1877 Caucasian patients aged 45–64 years of six family physicians were invited to participate in a metabolic screening between 2001 and 2003. These subjects were representative of the Local Health Units of the province of Asti (northwestern Italy) as reported previously.11 In total, 1658 patients (88.3%) agreed to participate by written informed consent, whereas 219 declined. Both participants and non-participants showed the same gender distribution, level of education, prevalence of known diabetes and subjects living in rural areas as the resident population of the corresponding age group in the same area.11

Clinics were held in the morning after fasting overnight; for each patient, weight, height, waist circumference (measured by a plastic tape meter at the level of the umbilicus) and blood pressure were measured, and a fasting blood sample was drawn. Systolic and diastolic blood pressures were measured twice with a standard mercury sphygmomanometer with the patient in a sitting position after at least 10 min of rest. The reported values are the means of two measurements.

All patients answered a questionnaire at the health screening. The following data were collected for each subject: smoking habits, alcohol consumption, education level, health conditions, drugs used, sleep duration, mean HT during autumn/winter, mean daily number of hours of TW, mean weekly number of meals consumed in restaurants (also considering fast-food restaurants and pizzerias) and regular use (>2 days per week) of AC during the summer season. Sleep duration was defined as self-reported time in bed (calculated from bedtime to rise time) minus sleep latency.

All subjects completed the validated, semiquantitative food-frequency questionnaire used in the European Prospective Investigation into Cancer and nutrition studies12 and the Minnesota leisure time physical activity questionnaire.13 A dietician who was blinded to the study details checked all questionnaires for completeness, internal coherence and plausibility. Each nutrient was adjusted for total energy using the residual method.14 The leisure physical activity level was calculated as the product of the duration and frequency of each activity (in h per week), weighted by an estimate of the metabolic equivalent of the activity (METS), and summed for all activities performed.

From January to November 2008, patients were contacted for follow-up visits. Deaths occurred in 61/1658 (3.7%) subjects during the follow-up period. All the remaining 1597 patients had weight, waist circumference and blood pressure measurements taken, and a blood sample was drawn for the determination of fasting metabolic parameters.

All procedures were in accordance with the Declaration of Helsinki. The study was approved by the local Ethics Committee.

Laboratory methods have been described previously.11 Diabetes and impaired fasting glucose were defined in accordance with guidelines.15

Statistical analyses

When analyzing the association between the incidence of obesity and baseline variables, subjects with obesity (body mass index (BMI) 30 kg m−2) at baseline (n=315/1597; 19.7%). were excluded. When analyzing the association between the incidence of hyperglycemia and baseline variables, subjects with hyperglycemia (fasting glucose 5.6 mmol l−1) at baseline (n=618/1597; 38.7%) were excluded. Therefore, analyses were carried out in 1282 and 979 subjects, respectively.

Because of the low number of expected incident cases of type 2 diabetes, our study did not have sufficient statistical power to detect differences in baseline variables between patients with and without incident diabetes. Therefore, incident impaired fasting glucose and diabetes were combined into one category, termed incident hyperglycemia.

A Student's t-test (normal distribution) or Mann–Whitney test (skewed distribution) and a χ2-test were performed to assess raw differences in baseline continuous and categorical variables, respectively. A logistic regression analysis was performed to estimate adjusted odds ratios among baseline fiber intake, saturated fat intake, METS, degrees centigrade (°C) of HT, hours of sleep, hours of TW, mean number of RMs per week, antidepressant/AD use, AC use and incident obesity and incident hyperglycemia, after controlling for sex, baseline BMI, education level and (in cases of incident hyperglycemia) baseline glucose values and alcohol intake. A multiple linear regression model was conducted to evaluate the association among these variables and continuous values of BMI and fasting glucose at follow-up.

Because of the suspect of a non-linear relationship between incremental HT and BMI and fasting glucose at follow-up, tertiles of HT were used as dummy variables, using the lowest tertile as a reference.

Results

Incident obesity

Baseline characteristics according to obesity development are reported in Table 1. At follow-up, 103/1282 (8.0%; 95% confidence interval (CI) 6.5–9.5) subjects had become obese; those subjects showed significantly higher values of BMI, waist circumference, saturated fat intake, frequency of RMs, hours of TW, HT, use of AC and antidepressant/ADs at baseline. They also had significantly lower fiber intake, METS and hours of sleep.

Table 1: Baseline characteristics according to obesity development at follow-up

BMI values at follow-up showed an overall linear increase with increasing numbers of RMs, hours of television and reduced hours of sleep, fiber intake and METS (Figure 1). The mean BMI at follow-up was highest in the small group of subjects consuming 4 per RMs per week: 1.72 kg m−2 (corresponding to a 3 kg increase in weight and a 4 cm increase in waist circumference).

Figure 1
Figure 1

Baseline environmental characteristics and BMI at follow-up, by group of increment of the variables. : Upper–lower values, : 75th–25th percentile, □: median. Restaurant foods—group 1: no consumption of restaurant foods, n=1165; group 2: 1–3 per week, n=78; group 3: 4 per week, n=39. House temperature—first tertile 18°C, n=354; second tertile >18°C to <20°C, n=401; third tertile 20°C, n=527. Television hours—group 1: 0–1 h per day, n=533; group 2: 2–3 h per day, n=546; group 3: >3 h per day, n=203. Sleep hours—group 1: 6.5 h per day, n=309; group 2: 7 h per day, n=456; group 3: >7 h per day, n=517. METS—group 1: <18 h per week, n=410; group 2: 18 h per week <24 h per week, n=411; group 3:24 h per week, n=461. Fiber intake (g per day)—group 1: first tertile <16.5 g per day, n=447; group 2: second tertile 16.5 g per day <23.5 g per day, n=413; group 3: third tertile 23.5 g per day, n=422. Saturated fat intake (% energy)—group 1: first tertile <10.5%, n=428; group 2: second tertile 10.5%<12.7%, n=426; group 3: third tertile 12.7%, n=428.

In a multiple logistic regression model, after adjusting for sex, education level, baseline BMI and all the variables listed in Table 2, the following variables were independently associated with incident obesity: reduced level of exercise, increased number of RMs (for each additional meal per week), being in the highest tertile of HT and reduced hours of sleep (Table 2).

Table 2: Association between baseline variables and obesity at follow-up in a logistic regression model: crude (left) and adjusted (right)

When using BMI at follow-up as a continuous variable, both a reduced level of exercise (β=−0.02; 95% CI −0.03 to −0.01, P<0.001) and an increased number of RMs (β=0.27; 95% CI 0.19–0.35, P<0.001 for each additional meal per week) remained significantly associated with BMI at follow-up, after carrying out a multiple linear regression model that was adjusted for all the variables listed in Table 2.

Incident hyperglycemia

Baseline characteristics according to hyperglycemia development are reported in Table 3. At follow-up, 174/979 (17.8%; 95% CI 15.4–20.2) subjects had developed hyperglycemia; those subjects were more frequently male and had significantly higher values of waist circumference, fasting glucose, saturated fat intake, alcohol intake, mean number of RMs per week, hours of TW, HT and use of AC and antidepressant/ADs at baseline. They also showed significantly lower fiber intake and exercise levels.

Table 3: Baseline characteristics according to hyperglycemia development at follow-up

Fasting glucose values at follow-up increased with increasing numbers of RMs per week and hours of TW as well as reduced fiber intake (Figure 2). The greatest increase in fasting glucose values during follow-up was 1.2 mmol l−1 in the group eating 4 RMs per week (incident hyperglycemia=61%; 95% CI 43–79), whereas the lowest (−0.02 mmol l−1) was in the group with the highest fiber intake (incident hyperglycemia=14.5%; 95% CI 10.6–18.4).

Figure 2
Figure 2

Baseline environmental characteristics and fasting glucose values at follow-up, by group of increment of the variables. : Upper–lower values, : 75th–25th percentile, □: median. Restaurant foods—group 1: no consumption of restaurant foods, n=893; group 2: 1–3 per week, n=55; group 3: 4 per week, n=31. House temperature—first tertile 18°C, n=250; second tertile >18°C to <20°C, n=325; third tertile 20°C, n=404. Television hours—group 1: 0–1 h per day, n=413; group 2: 2–3 h per day, n=427; group 3: >3 h per day, n=139. Sleep hours—group 1: 6.5 h per day, n=252; group 2: 7 h per day, n=320; group 3: >7 h per day, n=407. METS—group 1: <18 h per week, n=312; group 2: 18 h per week <24 h per week, n=318; group 3:24 h per week, n=349. Fiber intake (g per day)—group 1: first tertile <16.5 g per day, n=334; group 2: second tertile 16.5 g per day <23.5 g per day, n=335; group 3: third tertile 23.5 g per day, n=310. Saturated fat intake (% energy)—group 1: first tertile <10.5%, n=316; group 2: second tertile 10.5%<12.7%, n=321; group 3: third tertile 12.7%, n=342.

In a multiple logistic regression analysis, after adjusting for sex, education level, alcohol intake, baseline BMI and glucose, and all the variable listed in Table 4, the following variables were independently associated with incident hyperglycemia: reduced fiber intake, increased number of RMs (for each additional meal per week) and being in the highest tertile of HT (Table 4). When using glucose level at follow-up as a continuous variable, both reduced fiber intake (β=−0.008; 95% CI −0.012 to −0.004, P=0.001) and increased number of RMs per week (β=0.22; 95% CI 0.18–0.26, P<0.001 for each additional meal per week) remained significantly associated with glucose values at follow-up after performing a multiple linear regression model that was adjusted for all the variables reported in Table 4.

Table 4: Association between baseline variables and hyperglycemia at follow-up in a logistic regression model: crude (left) and adjusted (right)

A sensitivity analysis was performed in all patients including the 61 subjects who had died during the follow-up period, applying two extreme scenarios, assuming that either none of them or all of them had developed obesity or hyperglycemia at the follow-up. These results were consistent with those obtained when analyzing the living individuals only.

Discussion

Alternative independent contributors to the obesity and hyperglycemia epidemic, other than physical inactivity and increased energy intake, may be the regular consumption of RMs, sleep restriction and higher home temperature. Other factors, such as TW and the use of antidepressant/ADs and AC, may have a lesser impact.

Sleep restriction

Sleep duration has declined from 8 to 9 h per night to 7 h or less per night in the last 50 years, largely as a consequence of voluntary sleep restriction (watching television, using the internet and getting more work done and so on).1 Sleep debt is associated with decreased rates of glucose clearance, insulin response and glucose effectiveness, increased sympathetic nervous system activity and impaired glucose regulation by reduced lipolytic effects.16 During sleep restriction, plasma leptin levels are decreased, whereas ghrelin, cortisol and orexin secretion is increased.1, 16 Thus, the link between sleep debt and hormones implicated in feeding regulation explains the observed increase in appetite and food intake,17 particularly for energy-dense, high-carbohydrate foods.16 This finding and the observed decrease in daytime physical activity after sleep loss18 may contribute to the documented increased risk of obesity.2, 19 However, these associations have been criticized because the epidemiological evidence is weak and the risk seems very small, and develops over many years in very short sleepers (around 5 h).20, 21 The relationships between sleep duration and incident diabetes are contrasting, as large United States epidemiological studies have found associations between both short and long sleep duration and diabetes.3, 22, 23 Associations have been found to be significant only in a subset of diabetic patients with severe symptoms, and not for short sleepers after adjusting for BMI,22 in men but not in women,24 and other studies failed to find any association.25 On the other hand, a growing number of epidemiological studies and meta-analyses have provided evidence of an association between short-duration sleep and the risk of obesity, as reviewed.1, 16 In accordance with the literature, in our cohort, sleep restriction was associated with obesity at follow-up, but not with incident hyperglycemia at follow-up.

Indoor temperature

The ability of brown adipose tissue to burn rather than store calories depends on its mitochondrial uncoupling proteins.26 Cold temperature can activate brown adipose tissue in adult humans, irrespective of age and gender.27 Over 30 years ago, research suggested that obesity could be treated by exercise in the cold.28 On the other hand, in a hot environment, the propensity for feeding is diminished,5 and AC may contribute to rising obesity because the body expends less energy in temperature ranges associated with climate-controlled settings (via postural adjustments and evaporative cooling).1, 29 Furthermore, the rising trend in central AC could provide an incentive for people to remain indoors and exercise less.7 We found a nonlinear relationship between mean HT and BMI and fasting glucose levels at follow-up; a twofold increased risk for both incident obesity and hyperglycemia was estimated in subjects living at an indoor temperature >20°C. It might be hypothesized that metabolic processes are favorably affected by an ambient temperature within the thermal neutral zone, that is, not requiring energy expenditure to be allocated to maintaining a constant body temperature.1 However, no evidence exists to support this and socioeconomic factors might confound these associations.

In our cohort, AC use, although associated with more than threefold higher incidence of both obesity and hyperglycemia, showed a lower impact on these conditions compared with other risk factors.

Diet and exercise

Several dietary factors that increase the risk for obesity and diabetes have been identified; among them, a reduced fiber intake has shown to have a strong predictive role for the incidence of type 2 diabetes,30 in line with our results. Meal consumption in restaurants was unusual in this middle-aged cohort, as <10% of the study subjects regularly consumed food away from home. Those who did consume food away from home did so with the following distribution: 47% pizzerias, 35% full-service restaurants and 18% fast-food restaurants. Compared with food prepared at home, restaurant food has high energy density, more fat and high glycemic load, and portion sizes are usually larger.6 It has been shown that a higher ratio of fast-food to full-service restaurant density was associated with higher BMI and risk of obesity.31 Furthermore, people find it difficult to estimate the caloric contents of food items at restaurants and tend to underestimate their energy content.32 Because of the low number of subjects regularly consuming meals in restaurants, our study did not have sufficient statistical power to detect risk differences by restaurant type. Nevertheless, the associations between the incremental number of meals consumed away from home, and incident obesity and hyperglycemia were strong and largely independent of other potentially confounding lifestyle factors. The incremental rises in BMI and glucose levels at follow-up were greatest in patients eating ≥4 RMs per week, and the values were comparable with the data obtained for fast-food consumption.33 These results are of potential interest for public programs aimed at reducing the diabetes and obesity epidemics.

Physical activity have a central role in diabetes and obesity prevention,34 and reduced exercise at baseline predicted the incidence of obesity in our cohort. TW was positively associated with both incident obesity and hyperglycemia, but this relationship was not significant in the multivariate model, probably because of the predominant roles of reduced exercise and unhealthy dietary factors in obesity and hyperglycemia, respectively; both of these conditions are strongly associated with increased amounts of time watching television.35

Other possible contributors

Even if incident obesity and hyperglycemia were five- and twofold higher, respectively, in users of antidepressant/ADs, the associations were smaller and no longer statistically significant in the multivariate model, suggesting that the crude association was confounded by other risk factors. However, because of the low prevalence of antidepressant/AD users in our cohort (95/1597; 5.9%), the study lacked sufficient statistical power to detect small risks.

Limitations and strengths

A potential limitation of this study was the reliance on self-reported diet and other lifestyle factors. In addition, because this was an observational study, the possibility of confounding by unmeasured variables cannot be excluded. Socioeconomic status, which was not analyzed, might be a potential confounder. Nevertheless, we introduced education level into the multivariate models; this variable is a reliable indicator of socioeconomic status because it is stable, established in early adulthood and not modified by chronic disease.36 Random misclassification and measurement errors in our prospective study would result in attenuated estimates of the strengths of the association with the outcome variables.

Finally, our results were limited to middle-aged individuals, many of whom lived in rural areas and had low levels of education. However, the population-based cohort and the biological plausibility of our results, which are in accordance with previous studies each analyzing a single contributor, lends support to our conclusions.

The strengths of this study were the facts that a large proportion of subjects were enrolled from a defined community and its focus on multiple novel explanations for incident obesity and hyperglycemia at once.

Conclusions

Sleep restriction, higher home temperature and regular consumption of RMs might represent lifestyle contributors to the obesity and hyperglycemia epidemics. Avoiding these behaviors could be a potential adjunctive non-pharmacological strategy for preventing the obesity and hyperglycemia epidemics.

References

  1. 1.

    , , , , , et al. Ten putative contributors to the obesity epidemic. Critic Rev Food Sci Nutr 2009; 49: 868–913.

  2. 2.

    , , , , , et al. The association between sleep duration, body mass index and metabolic measures in the Hordaland Health Study. J Sleep Res 2007; 16: 66–76.

  3. 3.

    , , , , , et al. Association of sleep time with diabetes mellitus and impaired glucose tolerance. Arch Intern Med 2005; 165: 863–867.

  4. 4.

    , , . Physical activity and television watching in relation to risk of type 2 diabetes. Am J Epidemiol 2009; 169: 428–434.

  5. 5.

    , , . The contribution of psychosocial stress to the obesity epidemic. Horm Metab Res 2009; 41: 261–270.

  6. 6.

    , , , , . Consumption of restaurant foods and incidence of type 2 diabetes in African American women. Am J Clin Nutr 2010; 91: 465–471.

  7. 7.

    , , , , . The relationship of housing and population health: a 30-year retrospective analysis. Environ Health Perspect 2009; 117: 597–604.

  8. 8.

    , , , , . Is obesity associated with lower body temperatures? Core temperature: a forgotten variable in energy balance. Metabolism 2009; 58: 871–876.

  9. 9.

    , , , , . Trends in the prevalence of short sleepers in the USA: 1975–2006. Sleep 2010; 33: 37–45.

  10. 10.

    , . Short sleep duration and weight gain: a systematic review. Obesity 2008; 16: 643–653.

  11. 11.

    , , , , , et al. The metabolic syndrome and high C-reactive protein: prevalence and difference by sex in a southern-European population-based cohort. Diab Metab Res Rev 2005; 21: 515–524.

  12. 12.

    , , , , , , . Validation of a self-administered food-frequency questionnaire administered in the European Prospective Investigation into Cancer and Nutrition (EPIC) study: comparison of energy, protein, and macronutrient intakes estimated with the doubly labeled water, urinary nitrogen, and repeated 24-h dietary recall methods. Am J Clin Nutr 1999; 70: 439–447.

  13. 13.

    , , , , , . Questionnaire for the assessment of leisure time physical activities. J Chronic Dis 1978; 31: 741–755.

  14. 14.

    , . Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 1986; 124: 17–27.

  15. 15.

    The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Report of the expert committee on the diagnosis and classification of diabetes mellitus. Diab Care 2003; 26: S5–S20.

  16. 16.

    , , , . Effects of poor and short sleep on glucose metabolism and obesity risk. Nat Rev Endocrinol 2009; 5: 253–261.

  17. 17.

    , , , , . Acute partial sleep deprivation increases food intake in healthy men. Am J Clin Nutr 2010; 91: 1550–1559.

  18. 18.

    , , , , , et al. Short-term sleep loss decreases physical activity under free-living conditions but does not increase food intake under time-deprived laboratory conditions in healthy men. Am J Clin Nutr 2009; 90: 1476–1482.

  19. 19.

    , , , , , et al. The association between sleep duration and obesity in older adults. Int J Obes 2008; 32: 1825–1834.

  20. 20.

    , , . Is sleep duration related to obesity? A critical review of the epidemiological evidence. Sleep Med Rev 2008; 12: 289–298.

  21. 21.

    . Short sleep is a questionable risk factor for obesity and related disorders: statistical versus clinical significance. Biol Psycol 2008; 77: 266–276.

  22. 22.

    , , , , , et al. A prospective study of self-reported sleep duration and incident diabetes in women. Diab Care 2003; 26: 380–384.

  23. 23.

    , , . Sleep duration as a risk factor for the development of type 2 diabetes. Diab Care 2006; 29: 657–661.

  24. 24.

    , , . High incidence of diabetes in men with sleep complaints or short sleep duration. Diab Care 2005; 28: 2762–2767.

  25. 25.

    , , , , , et al. Sleep disturbance in midlife unrelated to 32-year diabetes incidence. Diab Care 2005; 28: 2739–2744.

  26. 26.

    . Be cool, lose weight. Nature 2009; 485: 839–840.

  27. 27.

    . Brown adipose tissue – when it pays to be inefficient. New Engl J Med 2009; 360: 1553–1556.

  28. 28.

    , , . Treatment of obesity by exercise in the cold. Can Med Assoc J 1977; 117: 773–778.

  29. 29.

    , , , , , et al. Putative contributors to the secular increase in obesity: exploring the roads less traveled. Int J Obes 2006; 30: 1585–1594.

  30. 30.

    , , , , . Whole-grain and fiber intake and the incidence of type 2 diabetes. Am J Clin Nutr 2003; 77: 622–629.

  31. 31.

    , . Weight status and restaurant availability. A multilevel analysis. Am J Prev Med 2008; 34: 127–133.

  32. 32.

    . The fattering truth about restaurant food. BMJ 2008; 337: a2229.

  33. 33.

    , , , , , et al. Fast-food habits, weight gain, and insulin resistance (the Cardia study): 15-year prospective analysis. Lancet 2005; 365: 36–42.

  34. 34.

    The Diabetes Prevention Program Research Group. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med 2002; 346: 393–403.

  35. 35.

    , , . Time spent watching television, sleep duration and obesity in adults living in Valencia, Spain. Int J Obes Relat Metab Disord 2000; 24: 1683–1688.

  36. 36.

    , , , . Own education, current conditions, parental material circumstances, and risk of myocardial infarction in a former communist country. J Epidemiol Community Health 2000; 54: 91–96.

Download references

Acknowledgements

This study was supported by a grant from the Regione Piemonte, 2008.

Author information

Affiliations

  1. Department of Internal Medicine, University of Turin, Turin, Italy

    • S Bo
    • , M Durazzo
    • , L Ghinamo
    • , P Villois
    • , S Canil
    • , R Gambino
    • , M Cassader
    •  & P Cavallo-Perin
  2. Unit of Cancer Epidemiology, University of Turin, Turin, Italy

    • G Ciccone
  3. Diabetic Clinic, Hospital of Asti, Asti, Italy

    • L Gentile

Authors

  1. Search for S Bo in:

  2. Search for G Ciccone in:

  3. Search for M Durazzo in:

  4. Search for L Ghinamo in:

  5. Search for P Villois in:

  6. Search for S Canil in:

  7. Search for R Gambino in:

  8. Search for M Cassader in:

  9. Search for L Gentile in:

  10. Search for P Cavallo-Perin in:

Competing interests

The authors declare no conflict of interest.

Corresponding author

Correspondence to S Bo.

About this article

Publication history

Received

Revised

Accepted

Published

DOI

https://doi.org/10.1038/ijo.2011.5