OBJECTIVE

Emerging epidemiological evidence suggests that higher magnesium intake may reduce diabetes incidence. We aimed to examine the association between magnesium intake and risk of type 2 diabetes by conducting a meta-analysis of prospective cohort studies.

RESEARCH DESIGN AND METHODS

We conducted a PubMed database search through January 2011 to identify prospective cohort studies of magnesium intake and risk of type 2 diabetes. Reference lists of retrieved articles were also reviewed. A random-effects model was used to compute the summary risk estimates.

RESULTS

Meta-analysis of 13 prospective cohort studies involving 536,318 participants and 24,516 cases detected a significant inverse association between magnesium intake and risk of type 2 diabetes (relative risk [RR] 0.78 [95% CI 0.73–0.84]). This association was not substantially modified by geographic region, follow-up length, sex, or family history of type 2 diabetes. A significant inverse association was observed in overweight (BMI ≥25 kg/m2) but not in normal-weight individuals (BMI <25 kg/m2), although test for interaction was not statistically significant (Pinteraction = 0.13). In the dose-response analysis, the summary RR of type 2 diabetes for every 100 mg/day increment in magnesium intake was 0.86 (95% CI 0.82–0.89). Sensitivity analyses restricted to studies with adjustment for cereal fiber intake yielded similar results. Little evidence of publication bias was observed.

CONCLUSIONS

This meta-analysis provides further evidence supporting that magnesium intake is significantly inversely associated with risk of type 2 diabetes in a dose-response manner.

Type 2 diabetes has been a growing public health burden across the world, particularly in the developing countries (1). In China, for instance, diabetes has reached epidemic proportions, affecting ∼92.4 million people aged ≥20 years (9.7% of the adult population) (2). Thus, there is an urgent need to develop primary prevention strategies aimed at controlling this epidemic.

Diet is widely believed to play an important role in the development of type 2 diabetes (3,4). Magnesium, an important component of many foods, such as whole grains, nuts, and green leafy vegetables, is an essential cofactor for enzymes involved in glucose metabolism (5). Magnesium has received considerable interest for its potential in improving insulin sensitivity and preventing diabetes (6,7). A number of prospective cohort studies (819) of magnesium intake and diabetes incidence have been conducted, but the results are inconsistent. One previous meta-analysis (14) of eight studies reported a significant inverse association between magnesium intake and risk of type 2 diabetes. Another meta-analysis (20) of seven studies quantified a dose-response relationship for this association. However, it remains unclear whether overweight, an established risk factor of type 2 diabetes (21,22), affects the association and whether other factors highly correlated to magnesium intake, such as cereal fiber (14) and calcium (23), are responsible for the observed association.

During the past few years, the number of subsequent primary studies on this topic has nearly doubled. With mounting evidence, we conducted a meta-analysis of prospective cohort studies for the following purposes: 1) to update the epidemiologic evidence on the association between magnesium intake and type 2 diabetes; 2) to examine this association according to characteristics of study designs and populations; and 3) to quantify a dose-response pattern of magnesium intake on diabetes risk.

Literature search

We attempted to report this study in accordance with the Meta-analysis of Observational Studies in Epidemiology guidelines (24) for meta-analyses of observational studies. We searched the PubMed database through January 2011, using the key word magnesium in combination with diabetes, and examined the reference lists of the obtained articles. No restriction was imposed. When necessary, we contacted authors of original studies for additional data.

Study selection

Studies were included if they met the following criteria: the study design was a prospective cohort study; the exposure of interest was intake of dietary magnesium or total magnesium (dietary and supplemental); the outcome of interest was incidence of type 2 diabetes; and the risk estimate of type 2 diabetes related to magnesium intake and associated 95% CI were reported.

Data extraction and quality assessment

We used a standardized data collection form to extract the following information: author name; publication year; study population, location, and length of follow-up; number of cases and participants; assessment of exposure and outcome; most fully adjusted risk estimate from multivariable model for the highest versus the lowest category of magnesium intake with corresponding 95% CI; and statistical adjustment for the main confounding factors of interest. In one study (8), we used data from the multivariable model with full control for potential confounders but not for insulin and glucose because magnesium plays an important role in insulin action and glucose homeostasis (5), and adjustment for insulin and glucose, therefore, could represent overadjustment for variables on the causal pathway. Two authors (J.-Y.D. and L.-Q.Q.) independently performed the literature search, study selection, and data extraction. Any disagreements were resolved by discussion.

Statistical analysis

The relative risk (RR) was used as the common measure of association across studies, and the hazard ratio and incidence rate ratio were considered directly as RR. Homogeneity test was performed with the use of Q statistic at the P < 0.10 level of significance. We also calculated the I2 statistic, a quantitative measure of inconsistency across studies (25). Because of the presence of heterogeneity across studies, a random-effects model, which considered both within- and between-study variation, was used to compute the summary risk estimate.

Prespecified subgroup analyses according to geographic region, length of follow-up, sex, family history of diabetes, and BMI were performed to examine the impacts of these factors on the association. To test the robustness of the association, we also performed sensitivity analyses restricted to studies on dietary magnesium only and studies with further adjustment for cereal fiber intake. In addition, we examined the influence of a single study on the summary risk estimate by omitting one study in each turn.

We next quantified a potential linear dose-response relationship of magnesium intake to diabetes risk because most individual studies have detected a significant linear trend. We first calculated an RR for every 100 mg/day increment in magnesium intake for each study based on the method proposed by Greenland and Longnecker (26). These RRs across studies were then combined to obtain a summary estimate.

Potential publication bias was assessed by Begg rank correlation test and Egger linear regression test (27,28). All analyses were performed using STATA version 11.0 (StataCorp, College Station, TX). All statistical tests were two-sided and P < 0.05 was considered statistically significant, except where otherwise specified.

Study characteristics

We identified 13 prospective cohort studies (819) of magnesium intake and risk of type 2 diabetes—one article (11) included two separate cohort studies—involving 536,318 participants and 24,516 incident cases. All studies reported type 2 diabetes as the primary outcome except one study (17) that did not specify diabetes type (the great majority of participants had type 2 diabetes). Characteristics of the selected studies are presented in Table 1. The 13 prospective cohort studies were published between 1999 and 2010. Of them, eight studies were conducted in the United States, three in Asia, one in Europe, and one in Australia. The number of cases diagnosed in the primary studies ranged from 330 to 8,587, and the number of participants ranged from 4,497 to 85,060. Both men and women were included in seven studies, one study consisted of men only, and five studies consisted of women only. Of the studies, nine reported results on dietary magnesium intake, two (11) on total (dietary and supplemental) magnesium intake, and two (12,17) on both dietary and total magnesium intake. Intake of supplemental magnesium was assessed from use of magnesium or multivitamin supplements; however, dietary magnesium accounted for the majority of total magnesium intake.

Table 1

Characteristics of 13 prospective cohort studies of magnesium and type 2 diabetes

StudyPopulation (cases)Duration (years)Dietary assessment methodCase ascertainmentMagnesium intake (highest vs. lowest, mg/day)Adjusted RR (95% CI)Adjustment for potential confounders
Kao et al., 1999 (811,896 adults aged 45–64 years, U.S. (1,106) Validated FFQ Glucose levels, use of diabetic medication, and self-report 361 vs. 154 White: 1.08 (0.78–1.49); black: 0.98 (0.57–1.72) Age, BMI, sex, education, family history, WHR, sports index, diuretic use, and intakes of alcohol, calcium, and potassium 
Meyer et al., 2000 (935,988 women aged 55–69 years, U.S. (1,141) Validated FFQ Self-report 362 vs. 220 0.67 (0.55–0.82) Age, BMI, education, smoking, WHR, physical activity, intakes of total energy, alcohol, whole grains, and cereal fiber 
Hodge et al., 2004 (1034,641 adults aged 40–69 years, Australia (365) FFQ Confirmed self-report 773 vs. 230 0.55 (0.32–0.97) Age, BMI, sex, education, country of birth, family history, WHR, weight change, physical activity, and intakes of total energy and alcohol 
Lopez-Ridaura et al., 2004 (1142,872 men aged 40–75 years, U.S. (1,333) 11 Validated FFQ Confirmed self-report 457 vs. 270 0.72 (0.58–0.89) Age, BMI, family history, hypertension, hypercholesterolemia, smoking, physical activity, and intakes of total energy, alcohol, glycemic load, PUFA, TFA, processed meat, and cereal fiber 
Lopez-Ridaura et al., 2004 (1185,060 women aged 30–55 years, U.S. (4,085) 17 Validated FFQ Confirmed self-report 374 vs. 222 0.73 (0.65–0.82) Age, BMI, family history, hypertension, hypercholesterolemia, smoking, physical activity, and intakes of total energy, alcohol, glycemic load, PUFA, TFA, processed meat, and cereal fiber 
Song et al., 2004 (1238,025 women aged ≥45 years, U.S. (918) Validated FFQ Confirmed self-report 399 vs. 252 0.89 (0.71–1.10) Age, BMI, family history, smoking, physical activity, and intakes of total energy and alcohol 
van Dam et al., 2006 (1341,186 women aged 21–69 years, U.S. (1,964) Validated FFQ Confirmed self-report 244 vs. 115 0.65 (0.54–0.78) Age, BMI, education, family history, smoking, physical activity, and intakes of total energy, alcohol, coffee, sugar-sweetened drinks, red meat, processed meat, and calcium 
Schulze et al., 2007 (1425,067 adults aged 35–65 years, Germany (844) Validated FFQ Confirmed self-report 377 vs. 268 0.99 (0.78–1.26) Age, BMI, sex, education, sports activity, cycling, occupational activity, smoking, WC, and intakes of total energy, alcohol, carbohydrate, PUFA-to-SFA ratio, MUFA-to-SFA ratio, and cereal fiber 
Villegas et al., 2009 (1564,191 women aged 40–70 years, China (2,270) 6.9 Validated FFQ Confirmed self-report 318 vs. 214 0.80 (0.68–0.93) Age, BMI, WHR, smoking, physical activity, income, education, occupation, hypertension, and intakes of total energy and alcohol 
Hopping et al., 2010 (1675,512 men aged 45–75 years, U.S. (8,587) 14 FFQ Glucose levels, use of diabetic medication, and self-report 370 vs. 260 Men: 0.77 (0.70–0.85); women: 0.84 (0.76–0.93) BMI, physical activity, education, ethnicity, and total energy intake 
Kim et al., 2010 (174,497 adults aged 18–30 years, U.S. (330) 20 Validated diet history questionnaire Glucose levels and use of diabetic medication 403 vs. 200 0.53 (0.32–0.86) Age, BMI, sex, ethnicity, study center, education, smoking, physical activity, family history, systolic blood pressure, and intakes of total energy, alcohol, saturated fat, and crude fiber 
Kirii et al., 2010 (1817,592 adults aged 40–65 years, Japan (459) Validated dietary questionnaire Confirmed self-report 303 vs. 158 0.64 (0.44–0.94) Age, BMI, family history, smoking, hours of walking and sports participation, and intakes of total energy, alcohol, green tea, and coffee 
Nanri et al., 2010 (1959,791 men aged 45–75 years, Japan (1,114) Validated FFQ Confirmed self-report 348 vs. 213 Men: 0.86 (0.63–1.16); women 0.92 (0.66–1.28) Age, BMI, study area, smoking, family history, leisure time physical activity, hypertension, and intakes of total energy, alcohol, coffee, and calcium 
StudyPopulation (cases)Duration (years)Dietary assessment methodCase ascertainmentMagnesium intake (highest vs. lowest, mg/day)Adjusted RR (95% CI)Adjustment for potential confounders
Kao et al., 1999 (811,896 adults aged 45–64 years, U.S. (1,106) Validated FFQ Glucose levels, use of diabetic medication, and self-report 361 vs. 154 White: 1.08 (0.78–1.49); black: 0.98 (0.57–1.72) Age, BMI, sex, education, family history, WHR, sports index, diuretic use, and intakes of alcohol, calcium, and potassium 
Meyer et al., 2000 (935,988 women aged 55–69 years, U.S. (1,141) Validated FFQ Self-report 362 vs. 220 0.67 (0.55–0.82) Age, BMI, education, smoking, WHR, physical activity, intakes of total energy, alcohol, whole grains, and cereal fiber 
Hodge et al., 2004 (1034,641 adults aged 40–69 years, Australia (365) FFQ Confirmed self-report 773 vs. 230 0.55 (0.32–0.97) Age, BMI, sex, education, country of birth, family history, WHR, weight change, physical activity, and intakes of total energy and alcohol 
Lopez-Ridaura et al., 2004 (1142,872 men aged 40–75 years, U.S. (1,333) 11 Validated FFQ Confirmed self-report 457 vs. 270 0.72 (0.58–0.89) Age, BMI, family history, hypertension, hypercholesterolemia, smoking, physical activity, and intakes of total energy, alcohol, glycemic load, PUFA, TFA, processed meat, and cereal fiber 
Lopez-Ridaura et al., 2004 (1185,060 women aged 30–55 years, U.S. (4,085) 17 Validated FFQ Confirmed self-report 374 vs. 222 0.73 (0.65–0.82) Age, BMI, family history, hypertension, hypercholesterolemia, smoking, physical activity, and intakes of total energy, alcohol, glycemic load, PUFA, TFA, processed meat, and cereal fiber 
Song et al., 2004 (1238,025 women aged ≥45 years, U.S. (918) Validated FFQ Confirmed self-report 399 vs. 252 0.89 (0.71–1.10) Age, BMI, family history, smoking, physical activity, and intakes of total energy and alcohol 
van Dam et al., 2006 (1341,186 women aged 21–69 years, U.S. (1,964) Validated FFQ Confirmed self-report 244 vs. 115 0.65 (0.54–0.78) Age, BMI, education, family history, smoking, physical activity, and intakes of total energy, alcohol, coffee, sugar-sweetened drinks, red meat, processed meat, and calcium 
Schulze et al., 2007 (1425,067 adults aged 35–65 years, Germany (844) Validated FFQ Confirmed self-report 377 vs. 268 0.99 (0.78–1.26) Age, BMI, sex, education, sports activity, cycling, occupational activity, smoking, WC, and intakes of total energy, alcohol, carbohydrate, PUFA-to-SFA ratio, MUFA-to-SFA ratio, and cereal fiber 
Villegas et al., 2009 (1564,191 women aged 40–70 years, China (2,270) 6.9 Validated FFQ Confirmed self-report 318 vs. 214 0.80 (0.68–0.93) Age, BMI, WHR, smoking, physical activity, income, education, occupation, hypertension, and intakes of total energy and alcohol 
Hopping et al., 2010 (1675,512 men aged 45–75 years, U.S. (8,587) 14 FFQ Glucose levels, use of diabetic medication, and self-report 370 vs. 260 Men: 0.77 (0.70–0.85); women: 0.84 (0.76–0.93) BMI, physical activity, education, ethnicity, and total energy intake 
Kim et al., 2010 (174,497 adults aged 18–30 years, U.S. (330) 20 Validated diet history questionnaire Glucose levels and use of diabetic medication 403 vs. 200 0.53 (0.32–0.86) Age, BMI, sex, ethnicity, study center, education, smoking, physical activity, family history, systolic blood pressure, and intakes of total energy, alcohol, saturated fat, and crude fiber 
Kirii et al., 2010 (1817,592 adults aged 40–65 years, Japan (459) Validated dietary questionnaire Confirmed self-report 303 vs. 158 0.64 (0.44–0.94) Age, BMI, family history, smoking, hours of walking and sports participation, and intakes of total energy, alcohol, green tea, and coffee 
Nanri et al., 2010 (1959,791 men aged 45–75 years, Japan (1,114) Validated FFQ Confirmed self-report 348 vs. 213 Men: 0.86 (0.63–1.16); women 0.92 (0.66–1.28) Age, BMI, study area, smoking, family history, leisure time physical activity, hypertension, and intakes of total energy, alcohol, coffee, and calcium 

FFQ, food-frequency questionnaire; MUFA, monounsaturated fatty acid; PUFA, polyunsaturated fatty acid; SFA, saturated fatty acid; TFA, trans fatty acid; WC, waist circumference; WHR, waist-to-hip ratio.

The mean length of follow-up ranged from 4 to 20 years. Most studies used validated food-frequency questionnaires in dietary assessment. Diabetes ascertainment was largely based on self-reports of physician diagnosis, but the majority of cases were confirmed in validation studies. The major adjusted confounders included age, BMI, physical activity, total energy intake, alcohol consumption, smoking, education, and family history of diabetes. Adjusted dietary confounders were varied across individual studies, whereas cereal fiber intake was adjusted for in five studies (9,11,14,17).

Main analysis

The multivariable-adjusted RRs of type 2 diabetes for each study and all studies combined for the highest versus the lowest category of magnesium intake are shown in Figure 1. Of the 13 selected studies, 9 found a statistically significant inverse association between magnesium intake and diabetes risk. The summary RR of type 2 diabetes was 0.78 (95% CI 0.73–0.84), comparing the highest with the lowest category of magnesium intake. Substantial heterogeneity was observed across studies (Pheterogeneity = 0.04, I2 = 42.4%).

Figure 1

Forest plot of prospective cohort studies examining magnesium intake and risk of type 2 diabetes. M, male; F, female.

Figure 1

Forest plot of prospective cohort studies examining magnesium intake and risk of type 2 diabetes. M, male; F, female.

Close modal

Subgroup and sensitivity analyses

Table 2 presents the results of subgroup analyses stratified by characteristics of study designs and populations. Overall, the inverse association between magnesium intake and risk of type 2 diabetes was not substantially modified by geographic region, follow-up length, sex, or family history of diabetes. Notably, the observed inverse association was more pronounced among participants with BMI ≥25 kg/m2 (RR 0.73 [95% CI 0.66–0.81]), and there was little evidence of heterogeneity (Pheterogeneity = 0.21, I2 = 28.1%). However, no association was observed among those with BMI <25 kg/m2 (RR 1.09 [0.76–1.56]).

Table 2

Subgroup analyses relating magnesium to type 2 diabetes by characteristics of study designs and populations

GroupNumber of studiesRR (95% CI)PPheterogeneityI2 (%)Pinteraction
Total 13 0.78 (0.73–0.84) <0.001 0.04 42.4  
Geographic area      0.64 
 U.S. 0.77 (0.71–0.83) <0.001 0.04 49.9  
 Asia 0.81 (0.71–0.91) 0.001 0.53  
Duration (years)      0.93 
 ≤6 0.79 (0.73–0.85) <0.001 0.05 48.9  
 >6 0.78 (0.75–0.82) <0.001 0.12 40.2  
Sex      0.95 
 Men 0.77 (0.73–0.81) <0.001 0.85  
 Women 0.76 (0.70–0.83) <0.001 0.06 47.1  
Family history of diabetes      0.83 
 Yes 0.73 (0.62–0.85) <0.001 0.42  
 No 0.71 (0.62–0.80) <0.001 0.31 15.7  
BMI (kg/m2     0.13 
 <25 1.09 (0.76–1.56) 0.76 0.12 52.9  
 ≥25 0.73 (0.66–0.81) <0.001 0.21 28.1  
GroupNumber of studiesRR (95% CI)PPheterogeneityI2 (%)Pinteraction
Total 13 0.78 (0.73–0.84) <0.001 0.04 42.4  
Geographic area      0.64 
 U.S. 0.77 (0.71–0.83) <0.001 0.04 49.9  
 Asia 0.81 (0.71–0.91) 0.001 0.53  
Duration (years)      0.93 
 ≤6 0.79 (0.73–0.85) <0.001 0.05 48.9  
 >6 0.78 (0.75–0.82) <0.001 0.12 40.2  
Sex      0.95 
 Men 0.77 (0.73–0.81) <0.001 0.85  
 Women 0.76 (0.70–0.83) <0.001 0.06 47.1  
Family history of diabetes      0.83 
 Yes 0.73 (0.62–0.85) <0.001 0.42  
 No 0.71 (0.62–0.80) <0.001 0.31 15.7  
BMI (kg/m2     0.13 
 <25 1.09 (0.76–1.56) 0.76 0.12 52.9  
 ≥25 0.73 (0.66–0.81) <0.001 0.21 28.1  

To test the robustness of our findings, we conducted sensitivity analyses. Restricting analysis to studies of dietary magnesium intake yielded an RR of 0.80 (95% CI 0.70–0.86; n = 11). Restricting analysis to studies that controlled for cereal fiber yielded an RR of 0.74 (0.68–0.80; n = 5). Further analyses investigating the influence of a single study on the overall risk estimate by omitting one study at each turn yielded a narrow range of RRs from 0.77 (0.72–0.82) to 0.79 (0.74–0.85). In addition, no single study substantially contributed to the heterogeneity across studies.

Dose-response analysis

In the dose-response analysis of the 13 primary studies, the summary RR of type 2 diabetes for every 100 mg/day increment in magnesium intake was 0.86 (95% CI 0.82–0.89), with evidence of heterogeneity among studies (Pheterogeneity = 0.02, I2 = 48.4%).

Publication bias

There was little evidence of publication bias with regard to magnesium intake in relation to risk of type 2 diabetes, as indicated by Begg rank correlation test (P = 0.99) and Egger linear regression test (P = 0.95).

In the past decade, the role of magnesium in development of type 2 diabetes has been increasingly recognized. The present updated meta-analysis of 13 prospective cohort studies involving 536,318 participants and 24,516 cases determined a significant inverse association between magnesium intake and risk of type 2 diabetes in a dose-response manner.

Substantial heterogeneity was observed across studies, which was expected given the between-study variation, such as inconsistent data collecting methods, various nutrient databases, and different ethnic populations. In our subgroup analyses, the results were not substantially affected by geographic region, follow-up length, sex, or family history of diabetes, whereas the association tended to be stronger in overweight than in normal-weight individuals. The absence of heterogeneity in the stratified analyses by BMI indicated that BMI may, at least partially, contribute to the overall between-study variation. More important, BMI may serve as an effect modifier of the magnesium and diabetes association. All included cohort studies have adjusted for BMI, but only six studies (11,12,1719) have assessed the impacts of BMI on the association between magnesium intake and diabetes risk. Among them, three studies (12,17,18) found more pronounced associations in overweight individuals than in normal-weight ones, yet only one (12) reached statistical significance in interaction tests. In our stratified analyses, the effect of BMI on the association was apparent, although the test for interaction was not statistically significant (Pinteraction = 0.13), which was likely the result of insufficient statistical power. Of note, it is plausible that high magnesium intake may have greater effects on improving insulin sensitivity in overweight individuals who are prone to insulin resistance (12). In this regard, waist circumference and waist-to-hip ratio, which are suggested to be better predictors of insulin resistance than BMI (29), may have more pronounced effects on the magnesium and diabetes association.

Because magnesium intake highly correlated to other healthy lifestyle and dietary factors, such as cereal fiber and calcium intakes, it is difficult to isolate the effect of magnesium intake on diabetes risk from other factors. Consequently, the potential influences of these factors deserve consideration while interpreting the results. Higher cereal fiber intake has been shown to be associated with a reduced diabetes risk (14). In our analysis restricted to five studies (9,11,14,17) with control for cereal fiber intake, the inverse association between magnesium intake and diabetes risk persisted and remained statistically significant. In addition, limited evidence suggests calcium intake may play a role in preventing diabetes (23). However, in the Black Women’s Health Study (13) that mutually adjusted for magnesium and calcium, the association between calcium intake and risk of type 2 diabetes disappeared, whereas an inverse association with magnesium intake remained. Moreover, physical activity (11,15) and smoking (15,18), two important and modifiable risk factors for diabetes (30,31), appeared not to have appreciable impacts on the association. Taken together, existing evidence to date from observational cohort studies supports an independent, protective role of magnesium intake against type 2 diabetes.

Indeed, experimental and metabolic studies have provided convincing evidence in support of direct effects of magnesium intake on insulin resistance and type 2 diabetes. Intracellular magnesium deficiency may result in disorders of tyrosine kinase activity during insulin signaling and glucose-induced insulin secretion, leading to impaired insulin sensitivity in muscle cells and adipocytes (32,33). In animal studies, magnesium supplementation has been shown to protect against fructose-induced insulin resistance (6) and reduce the development of diabetes in rat models (7). Furthermore, evidence from randomized controlled studies suggests that magnesium supplementation may exert beneficial effects on glucose control in patients with type 2 diabetes (34) and improve insulin sensitivity in nondiabetic subjects (3537).

Our findings confirmed results from previous meta-analyses (14,20). With the accumulative evidence, we were able to enhance the precision of the risk estimates and perform subgroup analyses to explore sources of heterogeneity, thereby increasing the clinical relevance of our findings (38). Also, all included studies used a prospective cohort design, which minimized the likelihood of recall and selection biases. In addition, the presence of dose-response relationship strengthened the association of magnesium intake with risk of type 2 diabetes.

Several limitations should be acknowledged. First, observational studies cannot establish causal association, and residual confounding remains a concern. A wide range of potential confounders, including demographic and lifestyle factors, were adjusted for in the primary studies, whereas dietary factors were not sufficiently considered. Therefore, we could not fully exclude the possibility that unmeasured or inaccurately measured dietary factors are responsible for the observed association. Second, misclassification bias may have weakened the strength of the association. Because dietary assessment was based on self-administered questionnaires, misclassification of dietary magnesium intake is inevitable. Misclassification of diabetes cases is also likely to occur given that diabetes ascertainments in most studies were based on self-reports. Few studies (11,15,17,19) repeated dietary assessment during the follow-up period, which may also result in misclassification. Third, we could not completely rule out the influences of changes in diet on the risk estimate in those with subclinical diabetes at baseline. Sensitivity analyses that excluded cases diagnosed within the initial several years of follow-up would help examine these influences and achieve more reliable results, but they were performed in few studies (15). Finally, publication bias could affect results of meta-analyses. However, minimal evidence of this bias was detected in our meta-analysis.

Our findings have implications for further research and are of public health significance. Apart from adequate control for potential confounders, particularly dietary factors, subsequent cohort studies of magnesium intake and type 2 diabetes should take the potential effect modification by BMI into consideration. Individual-level meta-analyses (39) with more power to detect effect modification than study-level meta-analyses may also be pursued. To date, large scale, randomized, placebo-controlled trials, which provide the strongest evidence for establishing a causal relation (40), have not been carried out to directly evaluate the effect of magnesium on diabetes incidence. Given the compelling evidence from the observational studies, such trials are anticipated to draw definitive conclusions. As for public health, increased consumption of magnesium-rich foods, such as whole grains, nuts, and green leafy vegetables, may bring considerable benefits in diabetes prevention, especially in those at high risk.

In conclusion, the present meta-analysis of prospective cohort studies provides further evidence in support of a significant inverse association between magnesium intake and risk of type 2 diabetes, consistent with a dose-response relationship.

This study was funded by a grant from the National Natural Science Foundation of China (30771808).

No potential conflicts of interest relevant to this article were reported.

J.-Y.D. researched data, contributed to discussion, wrote the manuscript, and reviewed and edited the manuscript. P.X., K.H., and L.-Q.Q. contributed to discussion and reviewed and edited the manuscript.

1.
Wild
S
,
Roglic
G
,
Green
A
,
Sicree
R
,
King
H
.
Global prevalence of diabetes: estimates for the year 2000 and projections for 2030
.
Diabetes Care
2004
;
27
:
1047
1053
[PubMed]
2.
Yang
W
,
Lu
J
,
Weng
J
, et al
;
China National Diabetes and Metabolic Disorders Study Group
.
Prevalence of diabetes among men and women in China
.
N Engl J Med
2010
;
362
:
1090
1101
[PubMed]
3.
Colditz
GA
,
Manson
JE
,
Stampfer
MJ
,
Rosner
B
,
Willett
WC
,
Speizer
FE
.
Diet and risk of clinical diabetes in women
.
Am J Clin Nutr
1992
;
55
:
1018
1023
[PubMed]
4.
Knowler
WC
,
Barrett-Connor
E
,
Fowler
SE
, et al
;
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
[PubMed]
5.
Belin
RJ
,
He
K
.
Magnesium physiology and pathogenic mechanisms that contribute to the development of the metabolic syndrome
.
Magnes Res
2007
;
20
:
107
129
[PubMed]
6.
Balon
TW
,
Jasman
A
,
Scott
S
,
Meehan
WP
,
Rude
RK
,
Nadler
JL
.
Dietary magnesium prevents fructose-induced insulin insensitivity in rats
.
Hypertension
1994
;
23
:
1036
1039
[PubMed]
7.
Balon
TW
,
Gu
JL
,
Tokuyama
Y
,
Jasman
AP
,
Nadler
JL
.
Magnesium supplementation reduces development of diabetes in a rat model of spontaneous NIDDM
.
Am J Physiol
1995
;
269
:
E745
E752
[PubMed]
8.
Kao
WH
,
Folsom
AR
,
Nieto
FJ
,
Mo
JP
,
Watson
RL
,
Brancati
FL
.
Serum and dietary magnesium and the risk for type 2 diabetes mellitus: the Atherosclerosis Risk in Communities Study
.
Arch Intern Med
1999
;
159
:
2151
2159
[PubMed]
9.
Meyer
KA
,
Kushi
LH
,
Jacobs
DR
 Jr
,
Slavin
J
,
Sellers
TA
,
Folsom
AR
.
Carbohydrates, dietary fiber, and incident type 2 diabetes in older women
.
Am J Clin Nutr
2000
;
71
:
921
930
[PubMed]
10.
Hodge
AM
,
English
DR
,
O’Dea
K
,
Giles
GG
.
Glycemic index and dietary fiber and the risk of type 2 diabetes
.
Diabetes Care
2004
;
27
:
2701
2706
[PubMed]
11.
Lopez-Ridaura
R
,
Willett
WC
,
Rimm
EB
, et al
.
Magnesium intake and risk of type 2 diabetes in men and women
.
Diabetes Care
2004
;
27
:
134
140
[PubMed]
12.
Song
Y
,
Manson
JE
,
Buring
JE
,
Liu
S
.
Dietary magnesium intake in relation to plasma insulin levels and risk of type 2 diabetes in women
.
Diabetes Care
2004
;
27
:
59
65
[PubMed]
13.
van Dam
RM
,
Hu
FB
,
Rosenberg
L
,
Krishnan
S
,
Palmer
JR
.
Dietary calcium and magnesium, major food sources, and risk of type 2 diabetes in U.S. black women
.
Diabetes Care
2006
;
29
:
2238
2243
[PubMed]
14.
Schulze
MB
,
Schulz
M
,
Heidemann
C
,
Schienkiewitz
A
,
Hoffmann
K
,
Boeing
H
.
Fiber and magnesium intake and incidence of type 2 diabetes: a prospective study and meta-analysis
.
Arch Intern Med
2007
;
167
:
956
965
[PubMed]
15.
Villegas
R
,
Gao
YT
,
Dai
Q
, et al
.
Dietary calcium and magnesium intakes and the risk of type 2 diabetes: the Shanghai Women’s Health Study
.
Am J Clin Nutr
2009
;
89
:
1059
1067
[PubMed]
16.
Hopping
BN
,
Erber
E
,
Grandinetti
A
,
Verheus
M
,
Kolonel
LN
,
Maskarinec
G
.
Dietary fiber, magnesium, and glycemic load alter risk of type 2 diabetes in a multiethnic cohort in Hawaii
.
J Nutr
2010
;
140
:
68
74
[PubMed]
17.
Kim
DJ
,
Xun
P
,
Liu
K
, et al
.
Magnesium intake in relation to systemic inflammation, insulin resistance, and the incidence of diabetes
.
Diabetes Care
2010
;
33
:
2604
2610
[PubMed]
18.
Kirii
K
,
Iso
H
,
Date
C
,
Fukui
M
,
Tamakoshi
A
;
JACC Study Group
.
Magnesium intake and risk of self-reported type 2 diabetes among Japanese
.
J Am Coll Nutr
2010
;
29
:
99
106
[PubMed]
19.
Nanri
A
,
Mizoue
T
,
Noda
M
, et al
;
Japan Public Health Center-based Prospective Study Group
.
Magnesium intake and type II diabetes in Japanese men and women: the Japan Public Health Center-based Prospective Study
.
Eur J Clin Nutr
2010
;
64
:
1244
1247
[PubMed]
20.
Larsson
SC
,
Wolk
A
.
Magnesium intake and risk of type 2 diabetes: a meta-analysis
.
J Intern Med
2007
;
262
:
208
214
[PubMed]
21.
Chan
JM
,
Rimm
EB
,
Colditz
GA
,
Stampfer
MJ
,
Willett
WC
.
Obesity, fat distribution, and weight gain as risk factors for clinical diabetes in men
.
Diabetes Care
1994
;
17
:
961
969
[PubMed]
22.
Colditz
GA
,
Willett
WC
,
Rotnitzky
A
,
Manson
JE
.
Weight gain as a risk factor for clinical diabetes mellitus in women
.
Ann Intern Med
1995
;
122
:
481
486
[PubMed]
23.
Pittas
AG
,
Lau
J
,
Hu
FB
,
Dawson-Hughes
B
.
The role of vitamin D and calcium in type 2 diabetes. A systematic review and meta-analysis
.
J Clin Endocrinol Metab
2007
;
92
:
2017
2029
[PubMed]
24.
Stroup
DF
,
Berlin
JA
,
Morton
SC
, et al
.
Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group
.
JAMA
2000
;
283
:
2008
2012
[PubMed]
25.
Higgins
JP
,
Thompson
SG
,
Deeks
JJ
,
Altman
DG
.
Measuring inconsistency in meta-analyses
.
BMJ
2003
;
327
:
557
560
[PubMed]
26.
Greenland
S
,
Longnecker
MP
.
Methods for trend estimation from summarized dose-response data, with applications to meta-analysis
.
Am J Epidemiol
1992
;
135
:
1301
1309
[PubMed]
27.
Begg
CB
,
Mazumdar
M
.
Operating characteristics of a rank correlation test for publication bias
.
Biometrics
1994
;
50
:
1088
1101
[PubMed]
28.
Egger
M
,
Davey Smith
G
,
Schneider
M
,
Minder
C
.
Bias in meta-analysis detected by a simple, graphical test
.
BMJ
1997
;
315
:
629
634
[PubMed]
29.
Racette
SB
,
Evans
EM
,
Weiss
EP
,
Hagberg
JM
,
Holloszy
JO
.
Abdominal adiposity is a stronger predictor of insulin resistance than fitness among 50-95 year olds
.
Diabetes Care
2006
;
29
:
673
678
[PubMed]
30.
Jeon
CY
,
Lokken
RP
,
Hu
FB
,
van Dam
RM
.
Physical activity of moderate intensity and risk of type 2 diabetes: a systematic review
.
Diabetes Care
2007
;
30
:
744
752
[PubMed]
31.
Willi
C
,
Bodenmann
P
,
Ghali
WA
,
Faris
PD
,
Cornuz
J
.
Active smoking and the risk of type 2 diabetes: a systematic review and meta-analysis
.
JAMA
2007
;
298
:
2654
2664
[PubMed]
32.
Suárez
A
,
Pulido
N
,
Casla
A
,
Casanova
B
,
Arrieta
FJ
,
Rovira
A
.
Impaired tyrosine-kinase activity of muscle insulin receptors from hypomagnesaemic rats
.
Diabetologia
1995
;
38
:
1262
1270
[PubMed]
33.
Kandeel
FR
,
Balon
E
,
Scott
S
,
Nadler
JL
.
Magnesium deficiency and glucose metabolism in rat adipocytes
.
Metabolism
1996
;
45
:
838
843
[PubMed]
34.
Song
Y
,
He
K
,
Levitan
EB
,
Manson
JE
,
Liu
S
.
Effects of oral magnesium supplementation on glycaemic control in Type 2 diabetes: a meta-analysis of randomized double-blind controlled trials
.
Diabet Med
2006
;
23
:
1050
1056
[PubMed]
35.
Guerrero-Romero
F
,
Tamez-Perez
HE
,
González-González
G
, et al
.
Oral magnesium supplementation improves insulin sensitivity in non-diabetic subjects with insulin resistance. A double-blind placebo-controlled randomized trial
.
Diabetes Metab
2004
;
30
:
253
258
[PubMed]
36.
Chacko
SA
,
Sul
J
,
Song
Y
, et al
.
Magnesium supplementation, metabolic and inflammatory markers, and global genomic and proteomic profiling: a randomized, double-blind, controlled, crossover trial in overweight individuals
.
Am J Clin Nutr
2011
;
93
:
463
473
[PubMed]
37.
Mooren
FC
,
Krüger
K
,
Völker
K
,
Golf
SW
,
Wadepuhl
M
,
Kraus
A
.
Oral magnesium supplementation reduces insulin resistance in non-diabetic subjects - a double-blind, placebo-controlled, randomized trial
.
Diabetes Obes Metab
2011
;
13
:
281
284
[PubMed]
38.
Thompson
SG
.
Why sources of heterogeneity in meta-analysis should be investigated
.
BMJ
1994
;
309
:
1351
1355
[PubMed]
39.
Stewart
LA
,
Tierney
JF
.
To IPD or not to IPD? Advantages and disadvantages of systematic reviews using individual patient data
.
Eval Health Prof
2002
;
25
:
76
97
[PubMed]
40.
Concato
J
,
Shah
N
,
Horwitz
RI
.
Randomized, controlled trials, observational studies, and the hierarchy of research designs
.
N Engl J Med
2000
;
342
:
1887
1892
[PubMed]