OBJECTIVE

We investigated whether serum magnesium (Mg2+) was prospectively associated with macro- or microvascular complications and mediated by glycemic control (hemoglobin A1c [HbA1c]), in type 2 diabetes (T2D).

RESEARCH DESIGN AND METHODS

We analyzed in 4,348 participants the association of serum Mg2+ with macrovascular disease and mortality (acute myocardial infarction [AMI], coronary heart disease [CHD], heart failure [HF], cerebrovascular accident [CVA], and peripheral arterial disease [PAD]), atrial fibrillation (AF), and microvascular complications (chronic kidney disease [CKD], diabetic retinopathy, and diabetic foot) using Cox regression, adjusted for confounders. Mediation analysis was performed to assess whether HbA1c mediated these associations.

RESULTS

The average baseline serum Mg2+ concentration was 0.80 ± 0.08 mmol/L. During 6.1 years of follow-up, serum Mg2+ was inversely associated with major macrovascular, 0.87 (95% CI 0.76; 1.00); HF, 0.76 (95% CI 0.62; 0.93); and AF, 0.59 (95% CI 0.49; 0.72). Serum Mg2+ was not associated with AMI, CHD, CVA, and PAD. During 5.1 years of follow-up, serum Mg2+ was inversely associated with overall microvascular events, 0.85 (95% CI 0.78; 0.91); 0.89 (95% CI 0.82; 0.96) for CKD, 0.77 (95% CI 0.61; 0.98) for diabetic retinopathy, and 0.85 (95% CI 0.78; 0.92) for diabetic foot. HbA1c mediated the associations of serum Mg2+ with HF, overall microvascular events, diabetic retinopathy, and diabetic foot.

CONCLUSIONS

Serum Mg2+ concentration is inversely associated with the risk to develop HF and AF and with the occurrence of CKD, diabetic retinopathy, and foot complications in T2D. Glycemic control partially mediated the association of serum Mg2+ with HF and microvascular complications.

Vascular complications are the primary cause of morbidity and mortality in people with type 2 diabetes (T2D) (1). The consequences of prolonged hyperglycemia include macrovascular complications (peripheral arterial disease [PAD] and cerebrovascular accident [CVA]) and microvascular complications (chronic kidney disease [CKD], retinopathy, and complications of the foot) (13). Macrovascular complications are largely explained by the formation of atherosclerotic plaques as a consequence of dyslipidemia and endothelial damage by reactive oxygen species (2). Microvascular complications are generally considered to be a direct result of hyperglycemia (2).

Hypomagnesemia (serum Mg2+ <0.7 mmol/L) is frequently observed in people with T2D. The prevalence of hypomagnesemia is high in T2D, with a range between 11.0 and 47.7% compared with those without diabetes (2.5–15%) (4). Hypomagnesemia is strongly associated with high glucose and hemoglobin A1c (HbA1c) levels, factors that are defined as poorly controlled diabetes (5). Being a cofactor of >600 enzymes, Mg2+ plays a role in many metabolic pathways, including glycolysis, β-oxidation, and insulin signaling (4,6). Cohort studies have repeatedly demonstrated that reduced serum Mg2+ levels are associated with poor glycemic control and insulin resistance in people with T2D (5,6). In line with this, oral Mg2+ supplementation has been shown to improve glucose and insulin levels in people with T2D (7).

Although hypomagnesemia is common in T2D, serum Mg2+ levels are not routinely determined in a clinical setting. Perhaps because of a limited number of clinical studies that have examined the long-term risks of hypomagnesemia, the benefits of Mg2+ supplementation have only been examined in small cohorts (7). Nevertheless, low serum Mg2+ levels were shown to predict the progression from prediabetes to diabetes in the general population (8). Moreover, hypomagnesemia results in a faster decline of renal function in T2D (3) and is linked to arrhythmias and hypertension (4,9), all important determinants of macro- and microvascular complications. The association of serum Mg2+ with macrovascular disease and/or mortality has not been thoroughly investigated in people with T2D, and only a few studies have been performed in populations without diabetes or in populations with heart failure (HF) or CKD, with inconsistent results (1017). Individuals with T2D have a 35% higher risk of developing atrial fibrillation (AF) compared with the general population, and AF is an established independent risk factor for CVA, HF, and mortality (18,19). Interestingly, AF is associated with low serum Mg2+ levels (20), although Mg2+ supplementation has failed to lower AF risk in the general population (21). Only a limited number of studies report risk associations between Mg2+ and microvascular complications in T2D, although these studies had a small sample size and used a cross-sectional study design (22,23). Better understanding of the cardiovascular disease (CVD) risks associated with hypomagnesemia in T2D is therefore warranted.

In this prospective study, we investigated whether the serum Mg2+ concentration is prospectively associated with major macrovascular disease and mortality (acute myocardial infarction [AMI], coronary heart disease [CHD], HF, CVA, PAD), AF, and major microvascular (CKD, diabetic retinopathy, and diabetic foot) complications in a large and well-phenotyped cohort of individuals with T2D. Moreover, we determined whether glycemic control, defined with HbA1c levels, mediates the association of serum Mg2+ with macro- and microvascular end points.

Study Population

The Hoorn Diabetes Care System (DCS) cohort started recruiting people with T2D living in the West-Friesland region of the Netherlands in 1998. People were included in the DCS cohort based on the following criteria: 1) at least one symptom of excessive thirst, polyuria, weight loss, hunger, or pruritus combined with fasting plasma glucose ≥7.0 mmol/L or random plasma glucose ≥11.1 mmol/L, or 2) two elevated plasma glucose concentrations on two different occasions in the absence of symptoms. People aged 40 or younger were excluded in case of insulin dependence within 4 weeks after the diagnosis of diabetes (24). As part of routine diabetes care, participants receive a standardized annual assessment with measurement of diabetes-related risk factors and complications. The DCS cohort is an on-going open cohort and grew to ∼13,000 individuals in the year 2018.

Biobanking was done in the years 2008–2009 and 2012–2014 and resulted in urine, serum, and citrate plasma samples from 5,500 people. Serum samples of 4,445 people were measured for Mg2+, of which 45 were stored in EDTA tubes or had insufficient volumes and were therefore excluded. In addition, 52 people were excluded due to diabetes type, including type 1 diabetes (n = 31), maturity-onset diabetes of the young (n = 9), latent autoimmune diabetes of adulthood (n = 7), secondary diabetes (n = 1), or gestational diabetes (n = 1), and 3 individuals because confirmation of the type of diabetes was missing (Supplementary Fig. 1A). The total population of T2D used for the current study analysis consisted of 4,348 participants. Individuals were excluded from the analysis if they already had experienced a macro- or microvascular event at baseline or were missing baseline data for each outcome separately: major macrovascular (n = 917), AMI (n = 324), CHD (n = 662), HF (n = 110), CVA (n = 142), PAD (n = 183), and AF (n = 306); major microvascular (n = 1,791), CKD (n = 665), diabetic retinopathy (n = 67), and diabetic foot (n = 1,401). Finally, individuals were excluded from the analysis because of missing follow-up for each outcome separately: major macrovascular (n = 49), AMI (n = 110), CHD (n = 108), HF (n = 109), CVA (n = 119), PAD (n = 129), and AF (n = 57), and microvascular events of CKD (n = 1), diabetic retinopathy (n = 468), and diabetic foot (n = 59) (Supplementary Fig. 1B–D). Ethical approval was obtained from the VU University Medical Center Ethical Review Committee (09/07/2009, ref: NL27783.029.09), Amsterdam. All participants gave written informed consent before participation.

Demographic and Laboratory Measurements

Demographic data, laboratory results, medication use, and self-reports were collected during visits for annual examinations performed at the DCS. Annual measurements were conducted using fasting blood: HbA1c, blood glucose levels, triglycerides (TG), total cholesterol, HDL cholesterol, LDL cholesterol, albumin and creatine (serum and urine), and estimated glomerular filtration rate (eGFR). At every annual visit, weight and height were measured to calculate BMI by dividing weight (kg) by the square of height (m). Systolic (SBP) and diastolic blood pressure (DBP) were measured twice using a random-zero sphygmomanometer. Education level, marital status, ethnic background and smoking status information were obtained by self-reports. Smoking status was categorized as current, former, or never smoker (24). Medication use was registered annually and checked by dispensing labels. Medication use was categorized by the use of glucose-lowering drugs (Anatomical Therapeutic Chemical Classification System [ATC] code A10) and CVD medication use (ATC codes for cardiac therapy [C01], antihypertensives [C02], diuretics [C03], peripheral vasodilators [C04], vasoprotectives [C05], β-blocking agents [C07], calcium channel blockers [C08], and agents on the renin-angiotensin system [C09]). In total, 4,445 serum samples collected in the years 2008, 2009, 2012, 2013, and 2014 were stored at −80°C until the measurement for Mg2+ in 2019 at the Laboratory Medicine Department (Radboud University Medical Center) using a calibrated standardized colorimetric assay with a coefficient of variation of 1.98% (cobas C8000; Roche Diagnostics, Risch-Rotkreuz, Switzerland).

Macro- and Microvascular Outcome Measurements

Participants were monitored from the time of biobanking until occurrence of the first macrovascular event or mortality, microvascular complication, or until the end of follow-up, whichever came first. A major macrovascular event was defined by first morbidity and mortality of AMI, CHD, CVA, HF, or PAD during follow-up. Additionally, these macrovascular outcomes were analyzed separately. AF was analyzed separately as a macrovascular outcome and defined by the first occurrence during follow-up. A major microvascular event was defined by the first CKD, diabetic retinopathy, or diabetic foot complication during follow-up, which were also analyzed separately.

The macrovascular events were annually self-reported and validated against the electronic patient registration from the regional hospital and general practitioner. The validation resulted in a sensitivity of 86% and specificity of 90%. The positive and negative predictive values were 90% and 87%, respectively (24). Macrovascular events were coded according to the International Classification of Diseases Ninth Revision (ICD-9). Macrovascular-related deaths were determined by biannual check-ups using the national population registry, general practitioner records, and hospital medical records. The eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration, and CKD was defined as an eGFR <60 mL/min/1.73 m2.

The grade of retinopathy was determined using fundus photography and scored according to the EURODIAB classification scale: grade 0 is no retinopathy, grade 1 is minimal nonproliferative retinopathy, grade 2 is moderate nonproliferative retinopathy, grade 3 is severe nonproliferative (preproliferative retinopathy), grade 4 is photocoagulated retinopathy, and grade 5 is proliferative retinopathy (24,25). Referable diabetic retinopathy was considered in case of a EURODIAB grade ≥2. Diabetic foot complications were screened by multiple criteria: 1) dermatological and musculoskeletal inspection; 2) check for skin pressure and foot deformity; 3) neurological assessment, including test of protective sensation using a 10-g monofilament and test of vibratory sensation using a 128-Hz tuning fork; 4) assessing presence of PAD by foot pulses; and 5) limited joint mobility. Results of this screening were used to define foot complications as loss of protective sensibility, PAD, and/or incidence of ulcer or amputation.

Statistical Analyses

We examined whether serum Mg2+ concentrations per 0.1 mmol/L increment were associated with the occurrence of macro- and microvascular complications using Cox proportional hazards models. Linearity of the associations was visually inspected between Mg2+ and end points by modeling Mg2+ in quartiles (Fig. 1). The proportional hazard assumption was checked by visual inspection of the log minus log plot of Mg2+ quartiles for CVD, with no deviations detected. Multivariable Cox models were used to estimate the hazard ratios (HRs) and 95% CIs of serum Mg2+ for the occurrence of macro- and microvascular events.

Figure 1

HR plots (95% CI) in Mg2+ quartiles 1–4 and fatal and nonfatal HF (A), AF (B), and major microvascular (C), CKD (D), diabetic retinopathy (E), and diabetic foot (F) complications, adjusted for model 2 (age, sex, CVD confounders; duration of diabetes, SBP, TG, LDL cholesterol, albumin-to-creatinine ratio, education, smoking, glucose-lowering drugs, and CVD medication). Median Q1 = 0.70 mmol/L Mg2+, median Q2 = 0.78 mmol/L Mg2+, median Q3 = 0.83 mmol/L Mg2+, and median Q4 = 0.89 mmol/L Mg2+.

Figure 1

HR plots (95% CI) in Mg2+ quartiles 1–4 and fatal and nonfatal HF (A), AF (B), and major microvascular (C), CKD (D), diabetic retinopathy (E), and diabetic foot (F) complications, adjusted for model 2 (age, sex, CVD confounders; duration of diabetes, SBP, TG, LDL cholesterol, albumin-to-creatinine ratio, education, smoking, glucose-lowering drugs, and CVD medication). Median Q1 = 0.70 mmol/L Mg2+, median Q2 = 0.78 mmol/L Mg2+, median Q3 = 0.83 mmol/L Mg2+, and median Q4 = 0.89 mmol/L Mg2+.

Close modal

Mg2+ per 0.1 mmol/L was added as the central determinant to the regression model, followed by confounders in three nested models. The first model (model 1) consisted of age (years) and sex. The second model (model 2) included model 1 plus CVD risk factors: duration of diabetes (years), SBP (mmHg), TG (mmol/L), LDL cholesterol (mmol/L), albumin-to-creatinine ratio (mg/mmol), education (low, medium, high), smoking (current, former, never), glucose-lowering drugs (yes/no), and CVD medication (yes/no). Antihypertensives (ATC code C02), diuretics (C03), and insulin (A10A) were included separately as confounders in the model in a sensitivity analysis due to their association with kidney failure and Mg2+ homeostasis. Potential mediation by glycemic control was assessed by adding HbA1c per 5 mmol/mol as a covariate in a third model (model 3). Mediation was confirmed by using the product-of-coefficients estimator to calculate the indirect effect using a and b as coefficients. The a coefficient was modeled as the association of serum Mg2+ per 0.1 mmol/L with the mediator, and b represents the association of the mediator HbA1c per 5 mmol/mol with CVD outcomes.

The Monte Carlo method was used to calculate the 95% CI of the indirect effect and the associated HR. A significant indirect effect was considered present when the null hypothesis (ab = 0) of the coefficient fell outside of the 95% CI, indicating mediation by HbA1c (26). The percentage of mediation was calculated using the associated HR. Sex as an effect modifier was assessed by including the cross-product of sex and Mg2+ in the fully adjusted model 2. Potential confounders of migration background (Western/non-Western), BMI (kg/m2), serum creatinine (µmol/L), and AF at baseline (yes/no) were included in the model in a sensitivity analysis. AF at baseline (yes/no) was included as a potential mechanistic risk factor for macrovascular morbidity and mortality. We also excluded 25 participants using Mg2+ supplement at baseline in a sensitivity analysis. Missing values for the confounders (<5%) for duration of diabetes, SBP, TG, LDL cholesterol, albumin-to-creatinine ratio, education (low, medium, high), smoking (current, former, never), serum creatinine, migration background, and BMI were imputed using predictive mean matching combing 10 iterations and 5 imputation sets into 1 imputation model. The determinant Mg2+, glucose-lowering, and CVD medication groups, and macro- and microvascular end points were added to the imputation model as predictors. P values ≤0.05 were considered statistically significant. All statistical analyses were performed using IBM SPSS Statistics for Windows 25 software (IBM Corp, Armonk, NY).

Study Population

The mean age of the participants was 65 ± 11 years, duration of diabetes was 7 ± 6 years, and 56% were men. The mean baseline Mg2+ concentration was 0.80 ± 0.08 mmol/L. The prevalence of hypomagnesemia (serum Mg2+ <0.7 mmol/L) was 9.1% (n = 397) of 4,348 individuals with T2D. During the median follow-up of 6.1 years, 328 major macrovascular events occurred, of which AMI occurred in 91 individuals, CHD in 169, HF in 155, CVA in 118, and PAD in 113. During the median of 6.2 years of follow-up, 157 AF incidents occurred. During the median follow-up of 5.1 years, 1,285 major microvascular complications occurred, of which 1,093 individuals had CKD (eGFR <60 mL/min/1.73 m2), 114 were referable for diabetic retinopathy, and 1,147 had (risk factors for) diabetic foot complications (Supplementary Table 1). The participants in the lowest Mg2+ quartile (<0.75 mmol/L Mg2+) were more often women compared with the other quartiles (Table 1). There was a clear upward trend from the highest to the lowest Mg2+ quartile regarding duration of diabetes, HbA1c, fasting glucose levels, and medication use (Table 1) as well as the incidence of macrovascular morbidity and mortality, AF, and microvascular complications (Supplementary Table 1).

Table 1

Baseline characteristics of individuals with T2D in the Hoorn, DCS Study (N = 4,348)

Quartile 1Quartile 2Quartile 3Quartile 4
Mg2+ <0.75 n = 1,153Mg2+ 0.75–0.80 n = 1,049Mg2+ 0.80–0.85 n = 1,124Mg2+ ≥0.80 n = 1,022
Demographics     
 Men 574 (50) 610 (58) 667 (59) 588 (58) 
 Age (years) 65 ± 11 64 ± 11 65 ± 11 66 ± 11 
 Duration diabetes (years) 6.9 (3.6–11.3) 5.6 (2.8–10.0) 4.7 (2.3–9.0) 4.3 (2.3–8.4) 
Education level     
 Low 478 (42) 408 (39) 445 (41) 415 (41) 
 Intermediate 449 (39) 441 (42) 447 (40) 415 (41) 
 High 165 (14) 157 (15) 179 (16) 156 (15) 
Smoking     
 Current 223 (20) 178 (17) 213 (19) 141 (14) 
 Former 554 (48) 515 (49) 564 (50) 530 (52) 
 Never 358 (31) 351 (34) 339 (30) 345 (34) 
Metabolic variables     
 BMI (kg/m230.3 (27.1–34.5) 29.5 (26.6–33.3) 29.2 (26.5–32.5) 28.9 (26.1–32.3) 
 SBP (mmHg) 144 ± 21 141 ± 20 140 ± 20 140 ± 19 
 DBP (mmHg) 78 ± 9 78 ± 9 78 ± 9 78 ± 9 
 Cholesterol     
  Total (mmol/L) 4.4 ± 1.0 4.5 ± 1.0 4.6 ± 1.0 4.7 ± 1.1 
  LDL (mmol/L) 2.4 ± 0.9 2.5 ± 0.9 2.6 ± 0.9 2.6 ± 0.9 
  HDL (mmol/L) 1.1 (1.0–1.4) 1.2 (1.0–1.4) 1.2 (1.0–1.5) 1.2 (1.0–1.5) 
 TG (mmol/L) 1.7 (1.2–2.4) 1.5 (1.2–2.1) 1.5 (1.1–2.1) 1.5 (1.1–2.0) 
 HbA1c (%) 6.7 (6.3–7.5) 6.6 (6.1–7.2) 6.4 (6.0–6.9) 6.3 (5.9–6.7) 
 HbA1c (mmol/mol) 50.0 (45.1–58.0) 48.6 (43.2–55.2) 47.0 (43.0–52.0) 45.0 (42.0–50.0) 
 Fasting glucose (mmol/L) 8.2 (7.1–9.7) 7.9 (7.0–9.1) 7.6 (6.8–8.4) 7.3 (6.5–8.2) 
 Albumin-to-creatine ratio (mg/mmol) 0.8 (0.3–2.0) 0.5 (0.0–1.3) 0.4 (0.0–1.1) 0.4 (0.0–0.8) 
 Serum creatinine 75 (63–88) 77.0 (65–90) 78 (67–91) 81 (70–93) 
 eGFR (mL/min/1.73 m281.1 ± 18.7 82.2 ± 17.6 80.5 ± 18.2 76.5 ± 19.0 
Medication use     
 Insulin (%) 27 24 17 16 
 Glucose-lowering medication (%)     
   Excluding insulins 90 81 73 60 
  Pooled group1 95 87 79 67 
 Cardiac therapy (%) 10 10 
 Antihypertensives (%) 
 Diuretics (%) 36 28 27 73 
 Peripheral vasodilators (%) <1 <1 <1 <1 
 Vasoprotectives (%) 48 36 37 36 
 β-Blocking agents (%) 22 15 17 18 
 Calcium channel blockers (%) 58 48 49 53 
 CVD medication (pooled group) (%)2 92 86 86 84 
Quartile 1Quartile 2Quartile 3Quartile 4
Mg2+ <0.75 n = 1,153Mg2+ 0.75–0.80 n = 1,049Mg2+ 0.80–0.85 n = 1,124Mg2+ ≥0.80 n = 1,022
Demographics     
 Men 574 (50) 610 (58) 667 (59) 588 (58) 
 Age (years) 65 ± 11 64 ± 11 65 ± 11 66 ± 11 
 Duration diabetes (years) 6.9 (3.6–11.3) 5.6 (2.8–10.0) 4.7 (2.3–9.0) 4.3 (2.3–8.4) 
Education level     
 Low 478 (42) 408 (39) 445 (41) 415 (41) 
 Intermediate 449 (39) 441 (42) 447 (40) 415 (41) 
 High 165 (14) 157 (15) 179 (16) 156 (15) 
Smoking     
 Current 223 (20) 178 (17) 213 (19) 141 (14) 
 Former 554 (48) 515 (49) 564 (50) 530 (52) 
 Never 358 (31) 351 (34) 339 (30) 345 (34) 
Metabolic variables     
 BMI (kg/m230.3 (27.1–34.5) 29.5 (26.6–33.3) 29.2 (26.5–32.5) 28.9 (26.1–32.3) 
 SBP (mmHg) 144 ± 21 141 ± 20 140 ± 20 140 ± 19 
 DBP (mmHg) 78 ± 9 78 ± 9 78 ± 9 78 ± 9 
 Cholesterol     
  Total (mmol/L) 4.4 ± 1.0 4.5 ± 1.0 4.6 ± 1.0 4.7 ± 1.1 
  LDL (mmol/L) 2.4 ± 0.9 2.5 ± 0.9 2.6 ± 0.9 2.6 ± 0.9 
  HDL (mmol/L) 1.1 (1.0–1.4) 1.2 (1.0–1.4) 1.2 (1.0–1.5) 1.2 (1.0–1.5) 
 TG (mmol/L) 1.7 (1.2–2.4) 1.5 (1.2–2.1) 1.5 (1.1–2.1) 1.5 (1.1–2.0) 
 HbA1c (%) 6.7 (6.3–7.5) 6.6 (6.1–7.2) 6.4 (6.0–6.9) 6.3 (5.9–6.7) 
 HbA1c (mmol/mol) 50.0 (45.1–58.0) 48.6 (43.2–55.2) 47.0 (43.0–52.0) 45.0 (42.0–50.0) 
 Fasting glucose (mmol/L) 8.2 (7.1–9.7) 7.9 (7.0–9.1) 7.6 (6.8–8.4) 7.3 (6.5–8.2) 
 Albumin-to-creatine ratio (mg/mmol) 0.8 (0.3–2.0) 0.5 (0.0–1.3) 0.4 (0.0–1.1) 0.4 (0.0–0.8) 
 Serum creatinine 75 (63–88) 77.0 (65–90) 78 (67–91) 81 (70–93) 
 eGFR (mL/min/1.73 m281.1 ± 18.7 82.2 ± 17.6 80.5 ± 18.2 76.5 ± 19.0 
Medication use     
 Insulin (%) 27 24 17 16 
 Glucose-lowering medication (%)     
   Excluding insulins 90 81 73 60 
  Pooled group1 95 87 79 67 
 Cardiac therapy (%) 10 10 
 Antihypertensives (%) 
 Diuretics (%) 36 28 27 73 
 Peripheral vasodilators (%) <1 <1 <1 <1 
 Vasoprotectives (%) 48 36 37 36 
 β-Blocking agents (%) 22 15 17 18 
 Calcium channel blockers (%) 58 48 49 53 
 CVD medication (pooled group) (%)2 92 86 86 84 

Data are presented as n (%), mean ± SD, or median (interquartile range) unless indicated otherwise. Mg2+ in mmol/L. Median Q1 = 0.70 mmol/L Mg2+, median Q2 = 0.78 mmol/L Mg2+, median Q3 = 0.83 mmol/L Mg2+, and median Q4 = 0.89 mmol/L Mg2+.

1

Glucose-lowering medication (ATC code A10: insulins [A10A] and glucose-lowering medication excluding insulins [A10B]).

2

CVD medication use (ATC codes cardiac therapy [C01], antihypertensives [C02], diuretics [C03], peripheral vasodilators [C04], vasoprotectives [C05], β-blocking agents [C07], calcium channel blockers [C08], and agents on renin-angiotensin system ([C09]).

Risk Association of Serum Mg2+ With Macrovascular Disease and Mortality

In the age- and sex-adjusted model, serum Mg2+ concentration was inversely related to the risk of major macrovascular events (0.81, 95% CI 0.70; 0.93), which attenuated to borderline significance after adjustment for CVD risk factors (0.87, 95% CI 0.76; 1.00). Considering the individual macrovascular diseases (AMI, CHD, HF, CVA, and PAD), serum Mg2+ only demonstrated a strong association with fatal and nonfatal HF (0.72, 95% CI 0.59; 0.87). This association attenuated slightly after CVD risk factors were included in the model (0.76, 95% CI 0.62; 0.93) but remained significant. Serum Mg2+ was not associated with AMI, CHD, CVA, and PAD and associated mortality. Figure 1A shows the HR (95% CI) per Mg2+ quartile with fatal and nonfatal HF, adjusted for model 2. The HR plot indicates that the lowest Mg2+ quartile is at higher risk of fatal and nonfatal HF.

Sensitivity analysis, including migration background, BMI, and serum creatinine as potential confounders to the model and separating medication groups (antihypertensives, diuretics, and insulin use), did not affect our results, and neither did the exclusion of 25 participants who were using Mg2+ supplements at baseline (Supplementary Table 2). Adjusting for HbA1c as a potential mediator slightly attenuated the association with HF from 0.76 (95% CI 0.62; 0.93) to 0.78 (95% CI 0.64; 0.96) (Table 2). No effect modification was observed by sex in the association of serum Mg2+ with macrovascular associations (P > 0.10). Mediation analyses showed a significant indirect effect of HbA1c of −0.054 (95% CI −0.107; −0.003) in the association of serum Mg2+ with fatal and nonfatal HF (Table 3), translating to 5% mediation by glycemic control.

Table 2

Cox regression analyses of the linear association of Mg2+ with macrovascular disease and mortality and with microvascular complications

Model 11Model 21Model 31
HR (95% CL)P valueHR (95% CL)P valueHR (95% CL)P value
Major macrovascular (AMI, CHD, HF, CVA, PAD) 0.81 (0.70, 0.93) 0.002 0.87 (0.76, 1.00) 0.054 0.89 (0.77, 1.03) 0.121 
 AMI 0.83 (0.64, 1.07) 0.141 0.89 (0.68, 1.17) 0.402 0.93 (0.71, 1.22) 0.594 
 CHD 0.89 (0.73, 1.08) 0.222 0.97 (0.79, 1.18) 0.733 1.00 (0.81, 1.22) 0.974 
 HF 0.72 (0.59, 0.87) 0.001 0.76 (0.62, 0.93) 0.008 0.78 (0.64, 0.96) 0.018 
 CVA 0.77 (0.62, 0.96) 0.022 0.83 (0.66, 1.04) 0.105 0.84 (0.66, 1.06) 0.132 
 PAD 0.92 (0.73, 1.16) 0.489 1.02 (0.80, 1.30) 0.880 1.01 (0.79, 1.29) 0.946 
 AF 0.58 (0.49, 0.70) <0.001 0.59 (0.49, 0.72) <0.001 0.59 (0.49, 0.72) <0.001 
Major microvascular (CKD, diabetic retinopathy, diabetic foot)2 0.82 (0.76, 0.88) <0.001 0.85 (0.78, 0.91) <0.001 0.86 (0.80, 0.93) <0.001 
 CKD2 0.86 (0.80, 0.93) <0.001 0.89 (0.82, 0.96) 0.003 0.88 (0.81, 0.96) 0.002 
 Diabetic retinopathy2 0.65 (0.52, 0.81) <0.001 0.77 (0.61, 0.98) 0.031 0.92 (0.72, 1.18) 0.506 
 Diabetic foot2 0.82 (0.76, 0.88) <0.001 0.85 (0.78, 0.92) <0.001 0.86 (0.79, 0.93) <0.001 
Model 11Model 21Model 31
HR (95% CL)P valueHR (95% CL)P valueHR (95% CL)P value
Major macrovascular (AMI, CHD, HF, CVA, PAD) 0.81 (0.70, 0.93) 0.002 0.87 (0.76, 1.00) 0.054 0.89 (0.77, 1.03) 0.121 
 AMI 0.83 (0.64, 1.07) 0.141 0.89 (0.68, 1.17) 0.402 0.93 (0.71, 1.22) 0.594 
 CHD 0.89 (0.73, 1.08) 0.222 0.97 (0.79, 1.18) 0.733 1.00 (0.81, 1.22) 0.974 
 HF 0.72 (0.59, 0.87) 0.001 0.76 (0.62, 0.93) 0.008 0.78 (0.64, 0.96) 0.018 
 CVA 0.77 (0.62, 0.96) 0.022 0.83 (0.66, 1.04) 0.105 0.84 (0.66, 1.06) 0.132 
 PAD 0.92 (0.73, 1.16) 0.489 1.02 (0.80, 1.30) 0.880 1.01 (0.79, 1.29) 0.946 
 AF 0.58 (0.49, 0.70) <0.001 0.59 (0.49, 0.72) <0.001 0.59 (0.49, 0.72) <0.001 
Major microvascular (CKD, diabetic retinopathy, diabetic foot)2 0.82 (0.76, 0.88) <0.001 0.85 (0.78, 0.91) <0.001 0.86 (0.80, 0.93) <0.001 
 CKD2 0.86 (0.80, 0.93) <0.001 0.89 (0.82, 0.96) 0.003 0.88 (0.81, 0.96) 0.002 
 Diabetic retinopathy2 0.65 (0.52, 0.81) <0.001 0.77 (0.61, 0.98) 0.031 0.92 (0.72, 1.18) 0.506 
 Diabetic foot2 0.82 (0.76, 0.88) <0.001 0.85 (0.78, 0.92) <0.001 0.86 (0.79, 0.93) <0.001 

CL, confidence limits.

1

Model 1 = age- and sex-adjusted model; model 2 = model 1 plus confounders duration of diabetes (years), SBP (mmHg), TG (mmol/L), LDL cholesterol (mmol/L), albumin-to-creatinine ratio (mg/mmol), education (low, medium, high), smoking (current, former, never), glucose-lowering drugs (yes/no), CVD medication (yes/no); model 3 = model 1 plus model 2 plus adjusted for glycemic control as HbA1c (per 5 mmol/mol).

2

CKD is defined as an eGFR <60 mL/min/1.73 m2, diabetic retinopathy is grade ≥2, and diabetic foot complications are grade ≥1. Major macrovascular (n = 3,382), AMI (n = 3,914), CHD (n = 3,578), HF (n = 4,129), CVA (n = 4,087), PAD (n = 4,043), and AF (n = 3,985); major microvascular (n = 2,557), CKD (n = 3,682), diabetic retinopathy (n = 3,813), and diabetic foot (n = 2,888).

Table 3

Mediation analysis of HbA1c with macrovascular disease and mortality and with microvascular complications

Indirect affect—HbA1c (per 5 mmol/mol) (ab coefficient)1Coefficient (95% CI)HR (95% CI)
Major macrovascular −0.047 (−0.084; −0.011) 0.954 (0.920; 0.989) 
 HF −0.054 (−0.107; −0.003) 0.947 (0.899; 0.997) 
 AF 0.007 (−0.053; 0.067) 1.007 (0.949; 1.070) 
Major microvascular −0.032 (−0.054; −0.009) 0.969 (0.948; 0.991) 
 CKD 0.013 (−0.012; 0.037) 1.013 (0.988; 1.038) 
 Diabetic retinopathy −0.181 (−0.227; −0.139) 0.834 (0.797; 0.870) 
 Diabetic foot −0.030 (−0.054; −0.011) 0.971 (0.947; 0.989) 
Indirect affect—HbA1c (per 5 mmol/mol) (ab coefficient)1Coefficient (95% CI)HR (95% CI)
Major macrovascular −0.047 (−0.084; −0.011) 0.954 (0.920; 0.989) 
 HF −0.054 (−0.107; −0.003) 0.947 (0.899; 0.997) 
 AF 0.007 (−0.053; 0.067) 1.007 (0.949; 1.070) 
Major microvascular −0.032 (−0.054; −0.009) 0.969 (0.948; 0.991) 
 CKD 0.013 (−0.012; 0.037) 1.013 (0.988; 1.038) 
 Diabetic retinopathy −0.181 (−0.227; −0.139) 0.834 (0.797; 0.870) 
 Diabetic foot −0.030 (−0.054; −0.011) 0.971 (0.947; 0.989) 
1

The indirect effect is calculated by using the product-of-coefficients estimator using model 2 (age, sex, CVD confounders; duration of diabetes, SBP, TG, LDL cholesterol, albumin-to-creatinine ratio, education, smoking, glucose-lowering drugs and CVD medication). HF (n = 4,129), AF (n = 3,985), major microvascular (n = 2,557), CKD (n = 3,682), diabetic retinopathy (n = 3,813), and diabetic foot (n = 2,888).

Risk Association of Serum Mg2+ With AF

In the age- and sex-adjusted model, serum Mg2+ was inversely associated with AF (0.58, 95% CI 0.49; 0.70) (Table 2). Including CVD risk factors in the model did not affect the association of serum Mg2+ with AF (0.59, 95% CI 0.49; 0.72) (Table 2). Figure 1B shows that the risk of AF decreases linearly throughout all four Mg2+ quartiles. No effect modification was observed by sex in the association of serum Mg2+ with AF (P > 0.10). Mediation analysis indicated no indirect effect by glycemic control (Table 3). To examine whether AF is a risk factor in the relation of Mg2+ with CVD outcomes, and in particular HF, the presence of AF at baseline was added as a confounder in the model and did not alter the association outcomes (Supplementary Table 2).

Risk Association of Serum Mg2+ With Microvascular Complications

In the age- and sex-adjusted model, serum Mg2+ was associated with major microvascular events combined (0.82, 95% CI 0.76; 0.88) and with each of the components: CKD (0.86, 95% CI 0.80; 0.93), diabetic retinopathy (0.65, 95% CI 0.52; 0.81), and diabetic foot complications (0.82, 95% CI 0.76; 0.88). Serum Mg2+ remained associated with major microvascular (0.85, 95% CI 0.78; 0.91), CKD (0.89, 95% CI 0.82; 0.96), diabetic retinopathy (0.77, 95% CI 0.61; 0.98), and diabetic foot complications (0.85, 95% CI 0.78; 0.92) after adjusting for CVD risk factors (Table 3). Figure 1C–F shows the HR (95% CI) per Mg2+ quartile with major microvascular, CKD, diabetic retinopathy, and diabetic foot complications, adjusted for model 2. The HR plots indicate that the lowest Mg2+ quartile is at higher risk for microvascular complications. Sensitivity analysis adjusting for migration background, BMI, and serum creatinine and separating CVD medication groups did not affect these results, and neither did exclusion of 25 participants who were using Mg2+ supplements at baseline (Supplementary Table 2). No effect modification was observed by sex in the association of serum Mg2+ with microvascular associations (P > 0.10). Mediation analyses showed significant indirect effects for major microvascular events together, −0.032 (95% CI −0.054; −0.009); diabetic retinopathy, −0.181 (95% CI −0.227; −0.139); and diabetic foot complications, −0.030 (95% CI −0.054; −0.011) (Table 3), indicating partial mediation by HbA1c. This translates into ∼3.1% mediation for major microvascular events, 16.6% mediation for diabetic retinopathy, and only 2.9% for diabetic foot complications by HbA1c.

In this study, we demonstrated that serum Mg2+ concentrations were prospectively and inversely associated with fatal and nonfatal HF and AF but not with other macrovascular end points (AMI, CHD, CVA and PAD). The association of serum Mg2+ with fatal and nonfatal HF was independent of the incidences of AF. In addition, serum Mg2+ was inversely associated with the risk for all microvascular events combined as well as for the individual components of CKD, diabetic retinopathy, and foot complications after correcting for CVD risk factors. Poor glycemic control, defined as HbA1c, a key feature of T2D, acts as a mediator in the association of serum Mg2+ with fatal and nonfatal HF and microvascular complications.

To date, only studies in the general population, individuals with CKD, and HF cohorts have assessed the association of serum Mg2+ with macrovascular outcomes (1017,20). A large meta-analysis study showed a reduced risk of major macrovascular events by 30% per 0.2 mmol/L Mg2+ increment, which is comparable to our 13% per 0.1 mmol/L Mg2+ increment (12). This study did not focus on T2D, contained limited heterogeneity, and did not fully distinguish between different forms of CVD. The association of serum Mg2+ with fatal and nonfatal AMI has only been studied in small trials or in combination with other therapies, and no prospective data are currently available (13). Large prospective studies that have investigated the relation between plasma Mg2+ with CHD reported contradictory results. In healthy women, higher Mg2+ was associated with reduced CHD risk, independent of CVD risk factors, whereas plasma Mg2+ and CHD incidence and death were not associated in the Prevention of Renal and Vascular End-Stage Disease study (14,15). Regarding CVA, the association between serum Mg2+ and the risk of stroke was not significant in a 15-year follow-up study after adjusting for CVD risk factors (16). A study of >13,000 individuals showed that hypomagnesemia was associated with an increased prevalence of PAD (17). The association of circulating Mg2+ with HF has been established in multiple prospective studies in the healthy population and HF patients and remained significant even after correcting for covariables (2730). Low serum Mg2+ was associated with the development of AF in patients who underwent coronary artery bypass surgery and in individuals without CVD disease (20). Surprisingly, our study is compared with previous studies showing an association with AF not only in the lowest Mg2+ quartile but also in the entire cohort. Our study is partly in accordance with the above-described studies in the general population, demonstrating that Mg2+ predominantly decreases the association of fatal and nonfatal HF and AF in people with T2D but not in other CVD types (AMI, CHD, CVA, and PAD).

To our knowledge, this is the first prospective study that has assessed the association of Mg2+ with the development of microvascular complications. The significant association of serum Mg2+ with microvascular complications is in correspondence with previous cross-sectional studies. For instance, low serum Mg2+ concentrations were associated with diabetic retinopathy, diabetic nephropathy, and diabetic neuropathy in hospitalized patients with diabetes (22). Regarding CKD, low serum Mg2+ is independently associated with a decline of the eGFR rate that is shown to be greater in people with diabetes (3). In line with these results, Mg2+ was inversely associated with albuminuria and retinopathy (23).

The association of serum Mg2+ with fatal and nonfatal HF may be related to microvascular outcomes. The types of HF, namely, HF with reduced (HFrEF) or preserved ejection fraction(HFpEF), are characterized by different biomarker profiles. HFpEF is mainly caused by hyperinsulinemia and microvascular endothelial inflammation, whereas HFrEF is explained by lipotoxicity and hyperglycemia (31). The association of serum Mg2+ with fatal and nonfatal HF was mediated with a mean of only 5% by glycemic control. People with diabetes are more prone to suffer from HFpEF, and interestingly, the likelihood of having HFpEF is increased with the number of microvascular complications (32). It is, therefore, likely that a large part of the HF complications in our study is HFpEF, which is driven by microvascular endothelial inflammation and not strongly mediated by glycemic control. This would suggest that Mg2+ has a protective effect on HF that is partially mediated by glycemic control and also by other factors such as endothelial inflammation.

AF is suggested to facilitate the development of HF by several mechanisms, but in our study, the protective effect of serum Mg2+ with AF was independent. Interestingly, our study shows that serum Mg2+ is associated with a reduced risk of AF throughout all Mg2+ quartiles, whereas previous studies showed an association in the low Mg2+ quartile only (20). The causal relation between T2D and AF is still debatable, but several studies indicate that inflammation, such as C-reactive protein and cytokines, have a role in the development of AF (33). Interestingly, Mg2+ supplementation improves endothelial function (34,35), although the effect of Mg2+ on inflammation markers (C-reactive protein and cytokines) is debated (36). Hence, inflammation may be the cause of the independent, prospective associations of serum Mg2+ with fatal and nonfatal HF and AF. Follow-up studies are needed to confirm whether inflammation has a causal role in the relation of serum Mg2+ with fatal and nonfatal HF, AF, and microvascular risk.

HbA1c partially mediated the association of Mg2+ with microvascular end points. Because hyperglycemia is considered to be a direct cause of microvascular complications (37), our finding is in line with previous research demonstrating that hypomagnesemia is associated with poor glycemic control (4). In addition, it is widely accepted that controlling hyperglycemia particularly improves microvascular outcomes in people with diabetes. The UK Prospective Diabetes Study (UKPDS) has shown that controlling glucose levels, by use of different types of sulfonylureas or insulin, decreases the risk of microvascular complications (retinopathy, vitreous hemorrhage, and/or fatal or nonfatal renal failure) but not macrovascular disease (38). The role of hyperglycemia in macrovascular disease is relatively small because other factors more prominently contribute to the disease development. Until recently, improved glycemic control had limited effect on the development of major CVD outcomes and did not seem to be related to the way glucose was lowered. Some recent drug classes, particularly glucagon-like peptide 1 receptor agonists and sodium–glucose cotransporter 2 (SGLT-2) inhibitors, however, have shown beneficial CVD effects beyond glycemic control (39). Interestingly, SGLT-2 inhibitors slightly increase Mg2+ levels, but the cardioprotective effects of SGLT-2 inhibitors may be independent of increased Mg2+ levels.

The results of our study stress the need for intervention studies to assess whether Mg2+ supplements decrease the risk of HF, AF, and microvascular complications in T2D. The HR plots in Mg2+ quartiles suggest that only the lowest quartile is at higher risk of HF, suggesting that maintaining adequate Mg2+ levels may suffice to reduce CVD risks. It has been established that an increased Mg2+ intake improves glucose and lipid profiles, consequently suggesting a beneficial effect on vascular health (7). On the other hand, it is important to maintain serum Mg2+ levels in a physiological range, because oversupplementation can cause severe health problems such as diminished tendon reflexes and hypotension (9). So far, dietary Mg2+ studies have mainly focused on macrovascular end points such as stroke or CHD, with only a limited number of HF studies, while leaving microvascular end points unaddressed (12). Mg2+ supplementation has failed to reduce AF risk in the general population (21) and may have more potential in a cohort with T2D. Because randomized controlled trials with Mg2+ supplements in T2D are lacking, Mg2+ supplementation is not listed in current treatment guidelines. Future intervention studies will be needed to assess the effect of Mg2+ supplementation on HF, AF, and microvascular complications in T2D.

Our study has a few limitations. First, we defined CKD as a low eGFR rate (<60 mL/min/1.73 m2), whereas CKD is often classified as a low eGFR rate in combination with albuminuria. Besides, we used the albumin-to-creatinine ratio as a confounder in our analysis. A consistent low eGFR rate might indicate kidney disease but must persist for at least 3 months, a time factor that we did not take into account (40). Another alternative would be to apply the official grading of nephropathy such as the H.S. Lee glomerular grading system. Secondly, HF was registered using the following ICD-9 codes: poor left ventricle function (428), 428.1 (left HF), and 428.9 (HF, unspecified). Consequently, we were not able distinguish between the two different types of HF, because ICD-9 codes 428.2 (HFrEF) or 428.3 (HFpEF) were not used.

The strengths of this study include the large sample size combined with a prospective design of 4–11 years and comprehensive evaluation of different CVD end points. The detailed phenotyping and extensive recording of laboratory measurements and complications allowed us to adjust for a complete set of CVD confounders. The participants of the Hoorn DCS cohort are treated by primary care in the Netherlands and, therefore, present very well the natural course of T2D. Most studies have included individuals with T2D who are hospitalized with longstanding diabetes. This might be the reason for the relatively low prevalence of hypomagnesemia of 9.1%, compared with previous T2D cohort studies (4,5).

In conclusion, the serum Mg2+ concentration is prospectively and inversely associated with the risk of fatal and nonfatal HF and AF but not with all of the other macrovascular end points (AMI, CHD, CVA, and PAD). Furthermore, the serum Mg2+ concentration is prospectively and inversely associated with the occurrence of the microvascular end points of CKD and diabetic foot complications in T2D. The association of serum Mg2+ with fatal and nonfatal HF and microvascular complications is partly mediated by glycemic control (HbA1c). Altogether, these results suggest that Mg2+ is an independent risk factor for the development of HF, AF, and microvascular complications in T2D. Long-term studies of oral Mg2+ supplementation are needed to determine whether maintaining Mg2+ levels in a physiological range can decrease HF, AF, and microvascular risk in people with T2D.

This article contains supplementary material online at https://doi.org/10.2337/figshare.14573631.

Acknowledgments. This study was a collaboration with the Diabetes Care System West-Friesland. We would like to thank the participants of this study as well as research staff of the Diabetes Care System West-Friesland.

Funding. This research was funded by grants from the Netherlands Organization for Scientific Research (NWO Veni 016.186.012), the Dutch Diabetes Research Foundation (2017-81-014), and the NIGRAM2+ consortium. The NIGRAM2+ collaboration project is co-funded by the PPP Allowance made available by Health∼Holland Top Sector Life Sciences & Health, to stimulate public-private partnerships (LSHM17034) and the Dutch Kidney Foundation (16TKI02). J.W.J.B. is supported by a ZonMW NWO-Vidi grant (91 71 8304).

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

Author Contributions. L.J.O., A.A.W.A.v.d.H., E.A,V., and J.W.J.B. contributed to statistical analysis and data interpretation. L.J.O., A.A.W.A.v.d.H., C.J.T., J.W.J.B., and J.H.F.d.B contributed to study design. L.J.O., A.A.W.A.v.d.H., J.W.J.B., and J.H.F.d.B wrote the manuscript. L.J.O., C.B., S.K., and M.v.B. took blood measurements. A.A.W.A.v.d.H., P.J.M.E., R.C.S., and J.W.J.B. contributed to cohort data collection. J.G.J.H., C.J.T., J.W.J.B., and J.H.F.d.B. supervised the study and interpreted the data. All authors reviewed and approved the final version of the manuscript. J.H.F.d.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Fan
W
.
Epidemiology in diabetes mellitus and cardiovascular disease
.
Cardiovasc Endocrinol
2017
;
6
:
8
16
2.
Cade
WT
.
Diabetes-related microvascular and macrovascular diseases in the physical therapy setting
.
Phys Ther
2008
;
88
:
1322
1335
3.
Pham
PC
,
Pham
PM
,
Pham
PT
,
Pham
SV
,
Pham
PA
,
Pham
PT
.
The link between lower serum magnesium and kidney function in patients with diabetes mellitus Type 2 deserves a closer look
.
Clin Nephrol
2009
;
71
:
375
379
4.
Pham
P-CT
,
Pham
P-MT
,
Pham
SV
,
Miller
JM
,
Pham
P-TT
.
Hypomagnesemia in patients with type 2 diabetes
.
Clin J Am Soc Nephrol
2007
;
2
:
366
373
5.
Kurstjens
S
,
de Baaij
JHF
,
Bouras
H
,
Bindels
RJM
,
Tack
CJJ
,
Hoenderop
JGJ
.
Determinants of hypomagnesemia in patients with type 2 diabetes mellitus
.
Eur J Endocrinol
2017
;
176
:
11
19
6.
Chutia
H
,
Lynrah
KG
.
Association of serum magnesium deficiency with insulin resistance in type 2 diabetes mellitus
.
J Lab Physicians
2015
; 
7
:
75
78
7.
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
8.
Kieboom
BCT
,
Ligthart
S
,
Dehghan
A
, et al
.
Serum magnesium and the risk of prediabetes: a population-based cohort study
.
Diabetologia
2017
; 
60
:
843
853
9.
Topf
JM
,
Murray
PT
.
Hypomagnesemia and hypermagnesemia
.
Rev Endocr Metab Disord
2003
;
4
:
195
206
10.
Angkananard
T
,
Anothaisintawee
T
,
Eursiriwan
S
, et al
.
The association of serum magnesium and mortality outcomes in heart failure patients: a systematic review and meta-analysis
.
Medicine (Baltimore)
2016
;
95
:
e5406
11.
Silva
PA
,
Fragoso
A
,
Silva
C
, et al
.
Magnesium and mortality in patients with diabetes and early chronic kidney disease
.
J Diabetes Metab
2014
;
5
:
1000347
12.
Del Gobbo
LC
,
Imamura
F
,
Wu
JHY
,
de Oliveira Otto
MC
,
Chiuve
SE
,
Mozaffarian
D
.
Circulating and dietary magnesium and risk of cardiovascular disease: a systematic review and meta-analysis of prospective studies
.
Am J Clin Nutr
2013
;
98
:
160
173
13.
DiNicolantonio
JJ
,
Liu
J
,
O’Keefe
JH
.
Magnesium for the prevention and treatment of cardiovascular disease
.
Open Heart
2018
;
5
:
e000775
14.
Chiuve
SE
,
Sun
Q
,
Curhan
GC
, et al
.
Dietary and plasma magnesium and risk of coronary heart disease among women
.
J Am Heart Assoc
2013
;
2
:
e000114
15.
Joosten
MM
,
Gansevoort
RT
,
Mukamal
KJ
, et al.;
PREVEND Study Group
.
Urinary and plasma magnesium and risk of ischemic heart disease
.
Am J Clin Nutr
2013
;
97
:
1299
1306
16.
Ohira
T
,
Peacock
JM
,
Iso
H
,
Chambless
LE
,
Rosamond
WD
,
Folsom
AR
.
Serum and dietary magnesium and risk of ischemic stroke: the Atherosclerosis Risk in Communities Study
.
Am J Epidemiol
2009
;
169
:
1437
1444
17.
Sun
X
,
Zhuang
X
,
Huo
M
, et al
.
Serum magnesium and the prevalence of peripheral artery disease: the Atherosclerosis Risk in Communities (ARIC) study
.
Atherosclerosis
2019
;
282
:
196
201
18.
Seyed Ahmadi
S
,
Svensson
AM
,
Pivodic
A
,
Rosengren
A
,
Lind
M
.
Risk of atrial fibrillation in persons with type 2 diabetes and the excess risk in relation to glycaemic control and renal function: a Swedish cohort study
.
Cardiovasc Diabetol
2020
;
19
:
9
19.
Bordignon
S
,
Chiara Corti
M
,
Bilato
C
.
Atrial fibrillation associated with heart failure, stroke and mortality
.
J Atr Fibrillation
2012
;
5
:
467
20.
Khan
AM
,
Lubitz
SA
,
Sullivan
LM
, et al
.
Low serum magnesium and the development of atrial fibrillation in the community: the Framingham Heart Study
.
Circulation
2013
;
127
:
33
38
21.
Kotecha
D
.
Magnesium for atrial fibrillation, myth or magic?
Circ Arrhythm Electrophysiol
2016
;
9
:
1
3
22.
Zhang
Y
,
Li
Q
,
Xin
Y
,
Lv
W
,
Ge
C
.
Association between serum magnesium and common complications of diabetes mellitus
.
Technol Health Care
2018
;
26
(
Suppl. 1
):
379
387
23.
Lu
J
,
Gu
Y
,
Guo
M
,
Chen
P
,
Wang
H
,
Yu
X
.
Serum magnesium concentration is inversely associated with albuminuria and retinopathy among patients with diabetes
.
J Diabetes Res
2016
;
2016
:
1260141
24.
van der Heijden
AAWA
,
Rauh
SP
,
Dekker
JM
, et al
.
The Hoorn Diabetes Care System (DCS) cohort. A prospective cohort of persons with type 2 diabetes treated in primary care in the Netherlands
.
BMJ Open
2017
;
7
:
e015599
25.
Aldington
SJ
,
Kohner
EM
,
Meuer
S
,
Klein
R
,
Sjølie
AK
.
Methodology for retinal photography and assessment of diabetic retinopathy: the EURODIAB IDDM complications study
.
Diabetologia
1995
;
38
:
437
444
26.
Selig
J
,
Preacher
K
.
Monte Carlo method for assessing mediation: an interactive tool for creating confidence intervals for indirect effects
.
2008
.
Accessed 28 January 2021. Available from http://quantpsy.org/medmc/medmc.htm
27.
Kunutsor
SK
,
Khan
H
,
Laukkanen
JA
.
Serum magnesium and risk of new onset heart failure in men: the Kuopio Ischemic Heart Disease Study
.
Eur J Epidemiol
2016
;
31
:
1035
1043
28.
Reffelmann
T
,
Dörr
M
,
Ittermann
T
, et al
.
Low serum magnesium concentrations predict increase in left ventricular mass over 5 years independently of common cardiovascular risk factors
.
Atherosclerosis
2010
;
213
:
563
569
29.
Wannamethee
SG
,
Papacosta
O
,
Lennon
L
,
Whincup
PH
.
Serum magnesium and risk of incident heart failure in older men: the British Regional Heart Study
.
Eur J Epidemiol
2018
;
33
:
873
882
30.
Adamopoulos
C
,
Pitt
B
,
Sui
X
,
Love
TE
,
Zannad
F
,
Ahmed
A
.
Low serum magnesium and cardiovascular mortality in chronic heart failure: a propensity-matched study
.
Int J Cardiol
2009
;
136
:
270
277
31.
Paulus
WJ
,
Dal Canto
E
.
Distinct myocardial targets for diabetes therapy in heart failure with preserved or reduced ejection fraction
.
JACC Heart Fail
2018
;
6
:
1
7
32.
Tromp
J
,
Lim
SL
,
Tay
WT
, et al.;
ASIAN-HF Investigators
.
Microvascular disease in patients with diabetes with heart failure and reduced ejection versus preserved ejection fraction
.
Diabetes Care
2019
;
42
:
1792
1799
33.
Sun
Y
,
Hu
D
.
The link between diabetes and atrial fibrillation: cause or correlation?
J Cardiovasc Dis Res
2010
;
1
:
10
11
34.
Darooghegi Mofrad
M
,
Djafarian
K
,
Mozaffari
H
,
Shab-Bidar
S
.
Effect of magnesium supplementation on endothelial function: a systematic review and meta-analysis of randomized controlled trials
.
Atherosclerosis
2018
;
273
:
98
105
35.
Shechter
M
,
Sharir
M
,
Labrador
MJP
,
Forrester
J
,
Silver
B
,
Bairey Merz
CN
.
Oral magnesium therapy improves endothelial function in patients with coronary artery disease
.
Circulation
2000
;
102
:
2353
2358
36.
Nielsen
FH
.
Magnesium deficiency and increased inflammation: current perspectives
.
J Inflamm Res
2018
;
11
:
25
34
37.
Bakker
W
,
Eringa
EC
,
Sipkema
P
,
van Hinsbergh
VWM
.
Endothelial dysfunction and diabetes: roles of hyperglycemia, impaired insulin signaling and obesity
.
Cell Tissue Res
2009
;
335
:
165
189
38.
UK Prospective Diabetes Study (UKPDS) Group
.
Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33)
.
Lancet
1998
;
352
:
837
853
39.
Carbone
S
,
Dixon
DL
,
Buckley
LF
,
Abbate
A
.
Glucose-lowering therapies for cardiovascular risk reduction in type 2 diabetes mellitus: state-of-the-art review
.
Mayo Clin Proc
2018
;
93
:
1629
1647
40.
Delanaye
P
,
Glassock
RJ
,
Pottel
H
,
Rule
AD
.
An age-calibrated definition of chronic kidney disease: rationale and benefits
.
Clin Biochem Rev
2016
;
37
:
17
26
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/content/license.