3.1. Search results and Study Selection
Overall, 382 records were identified by the database search, of which 188 were duplicated records. 175 records were excluded after abstract screening, and 85 further records were retrieved through hand search. The remaining 104 full-text articles were assessed in details for eligibility, of which 64 records were excluded. Forty papers were finally considered eligible for this review (Figure1).
3.2. Data extraction sheet
Table 2 summarizes characteristics of 40 included papers. These papers were published between 2007 and 2020, with the majority published after 2014. The number of studies included in the reviews ranged between 6 to 88. Most of the original studies were conducted in developed countries (Table 2).
3.3. Summary of results and Quality Assessment
The T2DM risk estimates with 95% CI for the included studies are summarised in Tables 3- 12. The methodological quality of studies was assessed using the AMSTAR score of all included studies with a high score ranging from 8 to 11, indicating the high methodological quality. However, many of the included studies lacked a priori published study objectives, registered study protocol, or ethics approval. Also, a majority of included studies did not provide a list of excluded studies. The results on the impact of factors affecting risk of T2DM based on the included studies' results have been summarised in the tables 3-12.
3.3.1. Socioeconomic factors
Only one meta-analysis of 23 cohort studies on the association between socio-economic position (SEP) (measured by educational level, occupation or income) and T2DM risk was found (7). Pooled results showed compared with the highest SEP groups, the low educational level, low occupation, and low income groups were associated with a 42%, 31%, and 36% increased risk of T2DM, respectively. A moderate heterogeneity was observed for all three indicators. Publication bias was observed when pooling studies on educational level and income, but not obvious for occupation (Table 3) (7). Educational level, occupation and income were defined and classified differently across studies due to different birth cohorts and different geographical settings. Educational level and income varied significantly between countries due to differences in educational systems and income level of countries. Occupation was organized with regard to some characteristics including social standing, physical and psychological work environment, structure and composition of workforce. Thus, to prevent difficulties when combining data and making international comparisons, the SEP groups were divided into two extreme categories, high and low in order to capture a sense of SEP irrespective of time and place (7).
There is strong evidence for a negative relationship between low socioeconomic position (SEP) and the risk of T2DM. The considerably high significant increased risk estimates (31%, 42%, and 36%) in low socio-economic groups ensure that the increased risk of T2D remains persisted even if decreasing the estimates to 2–6%. The strength of the association measured by low SEP was consistent in high-income countries. Although increased risks were observed with lower SEP in middle- and low-income countries, data from these economies was very limited. Further well-designed studies are therefore needed to characterize the association between T2DM incidence and SEP in middle-and low-income countries (7).
3.3.2. Lifestyle factors
Nine groups of lifestyle factors were identified, including alcohol use, smoking, stress, physical activity, sleep, coffee and tea consumption, dietary factors, combined lifestyle habits and anthropometric measures. The retrieved evidence on the impact of each lifestyle factor on the risk of T2DM is summarised in the following sections.
3.3.2.1. Alcohol consumption
Three meta-analysis studies on the association between alcohol consumption and the risk of T2DM were found (8-10). Results suggested that low to moderate alcohol intake reduces the risk of T2DM, whereas heavy alcohol intake is associated with an increased risk of T2DM (8- 10). Overall, research confirms a U-shaped relationship indicating a protective effect of moderate alcohol consumption and the risk of diabetes in both men and women (8-10). In line with this conclusion, the pooled RRs for different alcoholic beverages showed that all of wine, beer or spirits consumption were associated with a decreased risk of T2DM. Wine consumption was associated with a robust significantly decreased risk of T2DM, whereas beer or spirits consumption showed a slight trend of decreasing the risk of T2DM. Thus, wine consumption as the most prevalent alcoholic beverage worldwide appeared to be more beneficial for decreasing the risk of T2DM (9). Accordingly, several studies have realized the potential protective role of wine consumption in T2DM (54-55) (Table 4).
3.3.2.2. Smoking
Three records of prospective cohort studies on the association between smoking and the risk of T2DM were found (11- 13), demonstrating strong evidence for a positive relationship between smoking and risk of T2DM. Results from a meta-analysis of 84 prospective studies with 96 estimates of almost 6 million participants, and results from a meta- analysis of 25 prospective studies with more than l million participants indicated a robust positive association between current cigarette smoking and T2DM risk, by 37% (12) and by 44% (13), respectively, for active smokers compared with non-smokers with statistical heterogeneity of RRs across studies and no evidence of publication bias. The high heterogeneity was likely driven by the extremely large overall number of participants (> l million) in both analyses. Furthermore, the increased risk of T2DM is statistically significant for both sexes, by 35% for current male smokers and 27% for current female smokers in compare with non- smokers (11). In addition, the positive association persists statistically significant across a number of sensitivity analyses and also subgroup analyses stratified by various studies and participants characteristics. The beneficial effects of smoking cessation on diabetes risk were not significantly different between sexes. Although both active and passive smoking are associated with an increased risk of incident of T2DM, passive smoking still had a lower risk than active smoking and higher risk than never smokers, supporting the benefits of smoking cessation to reduce T2DM risk, if we assume that the association is causal (11-13) (Table 5).
3.3.2.3. Stress
Two studies including four variables have performed meta-analysis of cohort studies on the association between work-related stress and T2DM risk (14-15). Results did not confirm the association between the risk of T2DM and work-related stress in the form of high demands, poor decision latitude, job strain, or long working hours. When stratified by sex, results showed job strain was a risk factor for T2DM in women, but not for men. Further research is warranted to confirm the findings on this association (14-15) (Table 6).
3.3.2.4. Physical activity
Four systematic reviews and meta-analyses of cohort studies assessed the association between physical activity and the risk of T2DM (16-19). There is strong evidence for negative relationship between physical activity and the risk of T2DM. All subtypes of physical activity appear to be beneficial. High versus low total physical activity, leisure-time activity, low, moderate and vigorous intensity activity, resistance exercise, occupational activity and walking, resistance exercise, occupational physical activity, cardiorespiratory fitness were each associated with a statistically significant reduction in the risk of T2DM ranging from 15% to 55% (16-19). Cardiorespiratory fitness was responsible for the highest effect with 55 % reduction in risk of T2DM, while walking and occupational activities were both associated for 15% decrease in risk of T2DM (16). Vigorous activity with 39% risk reduction, the second highest risk reduction variable, appeared to be more strongly associated with reduced T2DM risk than walking (with 15%), occupational physical activity (with 15%), leisure-time activity (with 26%), suggesting further benefit at higher physical activity levels. Regular walking had a stronger effect on diabetes risk reduction than walking. Those who engaged in regular walking had 30% lower risk of T2DM compared with sedentary individuals (16-17). A similar decrease in diabetes risk was observed for physical activity of moderate intensity. Even after adjustment for BMI, the reduction in diabetes risk remained substantial (17%) for both regular moderately intense activity and walking. Also, the dose–response meta-analysis indicated that higher cardiorespiratory fitness and higher muscular strength were inversely associated with the risk of T2DM (18). An additional metabolic equivalent task (MET) increase in cardiorespiratory fitness and each one SD increase in muscular strength are associated with an 8% and 13% lower risk of T2DM (16). Therefore, physical activities that enhance cardiorespiratory fitness and/or muscular strength are of great importance to reducing the risk of T2DM. Furthermore, an assessment of the health benefits from a unit (= 11.25 MET h/week) increase in physical activity levels for an inactive individual is associated with a reduction of risk for T2DM incidence by 26%, independent of body weight (19) (Table 8).
3.3.2.5. Sleep
Four meta-analyses of prospective cohort studies exploring the association between T2DM risk and sleep-related factors were identified (20-23). Results of the associations between measures of quantity and quality of sleep, and common sleep disturbances and T2DM risk show a considerable and consistent pattern of increased risk of developing T2DM. Obstructive sleep apnea (OSA), difficulty in maintaining sleep, and difficulty initiating sleep had the largest effect varied from 55% to 102%, and then poor sleep quality and shift work with the same risk estimate of 40% were in the second place (20-23). The increased risk associated with either end of the distribution of sleep duration (short sleepers (generally 6 h/d), long sleepers (generally 9 h/night)) and long daytime napping (>1 hour/day) varies from 28% to 48% (21-22). Almost all of the studies included in the meta-analyses of sleep disturbances had estimated the risk of incident diabetes after adjustment for age, sex, BMI as well as multiple other potential confounders, conferring further strength to the results. Furthermore, the high statistical power, which tended to increase with the duration of study follow-up, and the absence of publication bias provide strong evidence for an important role of sleep disturbances in the risk of T2DM incidence. Data from more than a million participants revealed that comparing the increased risk of diabetes imparted by common sleep disturbances versus that of well-established risk factors listed in all guidelines for the screening and preventing of T2DM, OSA, difficulty in initiating and maintaining sleep had effect sizes only slightly smaller than having a family history of diabetes or being overweight but clearly larger than being physically inactive (22). Also, the impact of short or long sleep duration, poor sleep quality and shift work on diabetes risk is comparable to that of being physically inactive. Considering the dramatic increase in the prevalence of sleep disorders including OSA over the past two decades, the retrieved evidence emphasizes the importance of considering "sleep" in clinical recommendations for diabetes screening and prevention (Table 10).
3.3.2.6. Coffee and tea consumption
Five eligible studies associated with the effect of coffee and tea consumption on T2DM risk were included (24-28). The evidence overall confirms an inverse association between coffee, caffeine, and tea (>3cup/day) consumption and subsequent risk of T2DM . The association was stronger for women than that for men (24- 28). Both Carlstrom et al and Jiang et al indicated that habitual coffee consumption reduces the risk of developing T2D by approximately 30%; however, Huxley et al found a lower T2DM risk by 24% daily for coffee consumption (25-27). Dose–response analysis indicated 5% to 10% lower risk of incident T2DM for every additional cup of coffee per a day and a 12% reduction in T2DM risk for every 2 cups/day increment in coffee intake with the assumption of linearity (25). According to findings, both caffeinated and decaffeinated coffees have favourable metabolic effects, although the risk reduction of T2DM appears to be somewhat stronger with caffeinated coffee and slighter for tea consumption. Results obtained for tea consumption were slightly different between three retrieved meta-analyses studies. Overall, all three studies confirmed no statistically significant association between tea consumption and risk of T2DM (24, 27, 28). However, there was a difference between men and women with regards to tea consumption and T2DM risk as the reduced risk was bigger for men than for women and a significant association was found for men in the study by Yang et al (24). All three studies indicated that compared with the non/lowest of tea consumption group, tea consumption ≥3- 4 cups/day was associated with reduced T2DM risk. Jing et al showed that ≥4 cups/day of tea consumption was associated with a 20% reduced T2DM risk; both Yang et al and Huxley et al showed that daily tea consumption (≥3–4 cups/day) was associated with a lower T2DM risk by approximately 16% (24, 27, 28). Although the distinction among green tea, black tea, and oolong tea might be relevant for risk of T2DM, few studies provided information on tea types; therefore, no stratified analysis on tea types was conducted. Also, few studies were adjusted for additives (milk, cream, and sugar), which is of great importance (Table 7).
3.3.2.7. Combined Lifestyle factors
Two eligible studies on the combination of multiple healthy lifestyle factors (i.e. following a healthy diet, avoiding smoking and avoiding harmful alcohol drinking, maintaining healthy body weight, and exercising daily for at least 30 min) and the risk of incident T2DM were retrieved (29- 30). Results from the pooled analysis of 15 estimates from 14 cohort studies involving 970,170 participants indicated that adopting a combination of multiple healthy lifestyles is associated with a substantially lower risk of T2DM with 75% risk reduction. The associations remained persistent in all analyses stratified by populations from different socioeconomic backgrounds and baseline characteristics and no between-group differences were found (29). The results from a meta-analysis of six studies involving 96,175 participants, confirm a positive strong association between breakfast skipping and risk of T2DM. This association persists significant after adjustment for BMI though the effect was slightly attenuated (30) (Table 9).
3.3.2.8. Anthropometric measures
Only one eligible study exploring the relationship between birth weight and the risk of T2DM was retrieved (31). Results suggest a negative association between birth weight (BW) and future risk of diabetes. Using the normal BW of 2,500–4,000 grams as a reference, the results show that the risk of diabetes increases with Low BW, but not with High BW, indicating a negative association between BW and future risk of diabetes (31) (Table 9.1).
3.3.2.9. Food groups and Dietary Factors
Overall, ten eligible studies exploring the association between food and dietary factors and T2DM risk were retrieved (32-41). Diets according to the "healthy" dietary patterns (DPs) have a strong potential for preventing diabetes. In four meta-analyses of prospective studies, the DPs characterized by "healthy" DPs vs. "unhealthy" DPs were significantly associated with diabetes risk. The available data provide evidence of a negative significant association between adherence to healthy DPs and the risk of T2DM with the decreased risk ranged from 14% to 21%, as well as a positive significant association between adherence to unhealthy DPs and the risk of T2DM with the increased risk ranged from 30% to 44% (32, 35, 38, 39, and 41). The smallest risk estimates, a 14% increase in risk and a 30% decrease in risk by healthy and unhealthy DPs, came from the Maghsoudi study which included more estimates in its meta-analysis than the other studies (35). Even if we consider the least amounts of change in risk, the risk estimates are still large and emphasises the importance of good dietary practices for the prevention of T2DM. Furthermore, a considerable and consistent pattern of lower risk of T2DM was observed through greater adherence to plant-based DP, Mediterranean DP, DASH DP (Dietary Approaches to Stop Hypertension), AHEI DP (Alternative Healthy Eating Index), vegetarian DP, and Healthy/Prudent DP, suggesting the protective role of these DPs in reducing the risk of T2DM at a reasonable magnitude by 23%, , 17%, 18%, 21%, 35.6%, and 15% respectively(34, 36, 37, 39). The risk reduction remained significant even after adjustment for BMI. In most studies, subgroup analyses revealed that especially long-term follow-up (> 10 years) showed a prominent and significant decrease of T2DM risk, suggesting a beneficial long-term effect of DPs and food groups on T2DM risk (33- 37, 39) (Table 11).
Although we excluded the studies focusing on single foods or nutrients, a systematic review and meta-analysis study on the associations of 12 a priori defined food groups with T2DM risk was included in the current review (40). The findings of this study are in line with previous meta-analyses conducted mostly on single food groups. Taken together, an inverse association with T2DM risk was identified for the consumption of whole grains, dairy products, and fruits; on the other hand, for processed meat, red meat, and sugar sweetened beverage consumption, the association was positive, while no significant linear association for the intake of eggs, fish, nuts, vegetables, legumes, and refined grains was found (40). The available evidence on the association between dietary macronutrient intake (total fat, carbohydrate, protein) and T2DM risk indicates that a high intake of total dietary carbohydrate may increase T2DM risk; however, total fat and total proteins were not significantly associated with T2DM risk (33) (Table 11).
3.3.3. Environmental contaminants
Three main groups for environmental contaminants were identified, including heavy metal exposure, air pollutants, and exposure to noise traffic. The retrieved evidence on the impact of each environmental factor on the risk of T2DM has been summarised in the following sections.
3.3.3.1. Exposure to Noise Traffic
Only one meta-analysis of prospective cohort studies, all from developed countries, including 441,820 participants on the association between T2DM and exposure to noise traffic was retrieved (42). In summary, available evidence suggests a slight positive association between T2DM risk and exposure to noise traffic. Findings indicate a 4% increase in the risk of diabetes mellitus associated to 5 dB increases in noise exposure. An earlier meta-analysis reported a 22% (9%, 37%) higher risk of diabetes mellitus in those exposed to noise ≥ 64 dB compared with those with noise exposure < 64 db. As the results for subgroup analyses were not based on prospective cohort studies, we did not include them in the table of results. However, the strongest associations were observed for air traffic noise followed by road and then transportation noise. Considering the limited number of studies exploring the association between T2DM risk and noise exposure, further research is needed to confirm the available evidence on this association (42) (Table 12).
3.3.3.2. Air pollutants
A limited number of studies, all from developed countries (US, Canada, and Europe), have reported the association between air pollutants and T2DM incidence (43-45). The available data demonstrated a modest but statistically significant increase in risk of diabetes associated with exposure to major air pollutants. The findings from three retrieved meta-analyses (Balti et al, Eze et al, and Janghorbani et al) indicated that both gaseous pollutants [NO2 and NOx] and particulate matter [PM2.5 and PM10] were associated with T2DM risk; the association with gaseous pollutants, particularly NO2, was stronger (43-45). Both Eze et al and Balti et al found similar increase in T2DM risk associated with PM2.5, at 10% and 11%, respectively (43, 45). NO2 was associated with slightly higher increase in risk of T2DM by 13% increase (43). There was minimal heterogeneity in both NO2 and PM2.5 analyses which strengthens the results. Also, the subsequent increased risk of T2DM associated with air pollution, a complex mixture of pollutants including both gaseous pollutants and particulate matter, was 8% (44). Although the increased risk for T2DM due to air pollution for an individual would seem small compared with the effect of established diabetes risk factors such as physical inactivity and dietary factors, given the enormous number of people who are likely exposed to air pollution, even conservative risk estimates would result into a substantial increase in the population attributable fraction of diabetes related to air pollutants. However, a limited number of primary studies have reported the association of major air pollutants and T2DM incidence. In addition, although most studies had several markers of air pollution available, considering a high probability that negative findings will not be published, or the possibility that some markers have been measured but not reported, so some selective reporting may have occurred (45). Overall, the retrieved meta- analyses suggest that exposure to major air pollutants may be a risk factor for diabetes. Additional research from both developing and developed countries is therefore needed to improve our understanding of this association (Table 12).
3.3.3.3. Heavy metal exposure
One eligible study was found (46). In the meta-analysis of associations between urinary/blood Cadmium exposure (U- Cd/ B- Cd) and risk of T2DM, no statistically significant association was found. The odds ratio (ORs) for the highest vs. lowest exposure to B- Cd obtained from pooling 2 cohort studies was 0.81 (0.31-2.11) times with the P-value of 0.663, and the pooled odds ratio (ORs) for the highest vs. lowest exposure to U-Cd resulted from only one cohort study was 0.33(0.10-1.11) times with P-value of 0.073. Heterogeneity test was not available for U- Cd concentration due to the limited number of included cohort studies (only 1 study) for this variable. For B-Cd concentration moderate heterogeneity was found. Publication bias was not done due to the relatively limited number of eligible studies. In aggregation, the available findings suggest that high Cadmium exposure may not be risk factor for T2DM in the general population (46). However, further prospective studies with large population and long follow-up duration (> 10 years) are needed to confirm this finding (Table 12).