Background: Magnesium deficiency is common in type 2 diabetes and may have a negative impact on glucose homeostasis and insulin resistance, as well as on the evolution of complications such as retinopathy, thrombosis and hypertension.
Objective: To assess the dietary magnesium intake of patients with type 2 diabetes in Zurich, Switzerland and to compare the magnesium intake of diabetic and non-diabetic subjects.
Design: The magnesium intake of 97 randomly selected patients with type 2 diabetes and 100 healthy, non-diabetic controls matched for age and sex was estimated using a diet history method. During winter and summer periods, mean daily magnesium intakes were calculated from detailed information given by the test subjects about their eating habits over the previous 2 months. The calculations were performed using EBIS, a computer program based on a German nutrient data base (BLS 2.3), with food items specific to Switzerland added or directly analysed when necessary.
Results: The mean±s.d. daily magnesium intake of the male diabetic and male control subjects was 423.2±103.1 and 421.1±111.0 mg, respectively. The mean daily magnesium intake of the female diabetic and female control subjects was 419.1±109.7 and 383.5±109.7 mg, respectively. There were no significant differences in daily magnesium intake between the diabetic and the non-diabetic subjects and mean intakes in both groups exceeded Swiss recommended dietary intakes.
Conclusions: Dietary intake of magnesium appears sufficient in Swiss adults with type 2 diabetes and is unlikely to contribute to the aetiology of magnesium deficiency.
Sponsorship: The Swiss Federal Institute of Technology, Zurich, Switzerland.
Magnesium deficiency is a common characteristic of type 2 diabetes mellitus. In the USA, 25–39% of outpatient diabetics have low levels of serum magnesium (Nadler, 1995). Low serum magnesium levels in patients with type 2 diabetes have also been reported in several European countries, eg Austria, Germany, Italy, France and Sweden (Schnack et al, 1992; De Lenardis, 1999; Paolisso et al, 1988; Schlienger et al, 1988; Sjögren et al, 1988). Magnesium depletion has a negative impact on glucose homeostasis and insulin resistance in people with type 2 diabetes (Durlach & Rayssiguier, 1983; Nadler et al, 1993), as well as on the evolution of complications such as retinopathy (McNair et al, 1978), thrombosis (Nadler et al, 1993), and hypertension. Low magnesium intake may play a role in the development of diabetes (Colditz et al, 1992) and in the development of insulin resistance in non-diabetic adults (Humphries et al, 1999).
The reasons why magnesium deficiency is commonly found in diabetics are not clear, but may include increased losses of urinary magnesium, lower dietary intake of magnesium, or lower magnesium absorption compared to healthy individuals. Increased urinary losses due to glucosuria and osmotic diuresis are one of the causes of a low magnesium status in patients with diabetes (McNair et al, 1982; Fujii et al, 1982). Low dietary intake may also contribute to low magnesium status in diabetics (Durlach & Rayssiguier, 1983; Sheehan, 1991; White & Campbell, 1993). Patients with type 2 diabetes are often overweight, and may consume a diet higher in fat and energy and lower in magnesium than non-diabetics. However, if diabetics follow current dietary guidelines for diabetes which emphasise whole grains and vegetables, their intakes of magnesium may be higher than the general population. Studies that have reported magnesium intake in type 2 diabetes are equivocal (Schmidt et al, 1994; Ma et al, 1995). Schmidt et al (1994) found that diabetics had a low intake of magnesium (216.5 mg/day in women, 336.8 mg/day in men); however, the results were not compared with the intake of a control group. In contrast, data from the Atherosclerosis Risk in Communities (ARIC) study (Ma et al, 1995) showed a higher magnesium intake per MJ in diabetics than in disease-free subjects (eg in white men, 42 mg/MJ (175 mg/1000 kcal) vs 38 mg/MJ (157 mg/ 1000 kcal), respectively).
If low magnesium intake contributes to low magnesium status in type 2 diabetes, magnesium intakes could be increased by dietary emphasis on foods which are rich in magnesium or by magnesium supplements. Our objective was therefore to assess the dietary magnesium intake of patients with type 2 diabetes in Zurich, Switzerland and to compare the magnesium intake of diabetic and non-diabetic subjects.
Subjects and methods
Ninety-seven type 2 diabetics and 100 healthy non-diabetic subjects matched for age and sex participated in the study. The mean age (range) of the 54 diabetic and control men was 62.3 (39–75) and 62.4 y (46–74), respectively. The mean age (range) of the 43 diabetic women and the 46 control women was 63.3 (45–76) and 61.8 y (45–80), respectively (Table 1). All subjects lived in the Zurich metropolitan area, were of Central European origin, and lived in their own home (ie were not institutionalised). Sample size was set to allow detection of a difference of intake of 50 mg or more between the two groups with a significance level of 0.05 and a power of 90%.
The 97 diabetic patients were randomly selected from the register of the diabetic outpatient clinic at the University Hospital of Zurich. The response rate was 32%. Thirty-seven were using insulin, 35 were taking oral hypoglycaemics, 11 were using both, and 14 were not prescribed any antidiabetic drugs. The 100 non-diabetic subjects (54 men, 46 women) were recruited from two different sources. The city of Zurich provided 600 random addresses of Swiss nationals aged 45–75 y from which we randomly selected 342 matched by age and sex to the diabetic group. From this group, 54 generally healthy non-diabetic subjects were recruited. The remaining 46 subjects were recruited from the programme for older adults at the University of Zurich and from the staff of the Swiss Federal Institute of Technology Zurich.
The dietary habits of the subjects were assessed by recording the dietary history using a computer-aided interview. We used the EBIS computer software, a nutrient database analysis software developed at the Robert-Bosch-Krankenhaus in co-operation with the University of Hohenheim, Stuttgart, Germany. The diet history portion of the program presents questions about the different meals and their components in the form of a tree structure, starting with the single meals and ending with details of amounts of individual food items (Landig et al, 1998). The program is based on the German Food and Nutrient Database (Bundeslebensmittelschlüssel) BLS 2.3, an electronic database developed by the Federal Health Department, for use in nutrition epidemiology and dietary assessment. The BLS includes approximately 11 000 food items and recipes with 166 nutrients. The nutrient values come from different European food tables and from analyses of the Federal Research Institutes. About 90% of the analysed data derive from the German food table of Souci et al (1994). Missing values for which no analysed data are available were estimated from similar foods (Häussler et al, 1990).
The interviews were carried out during two different periods of the year, from November to December 1999 and from April to June 2000, to take seasonal variations into consideration. The subjects were asked to give detailed information about their eating habits during the previous 2 months. Breakfast, lunch, dinner and snacks were discussed in turn to determine which foods were consumed and how often. Detailed descriptions of all foods, beverages and supplements consumed, including cooking methods and brand names, were recorded. A photo catalogue was used to estimate portion sizes and a few pre-weighed samples of often consumed foods were showed to the subjects (eg slice of bread, piece of cheese). Cups and glasses with known volumes were also used to estimate quantities. The interviews were performed by five different interviewers. All of them were carefully trained by the same person to ensure uniform data collection. If interviewers had difficulty coding a food item, it was recorded in written form and coded later by the person doing the evaluation of all the interviews. The time required for an interview was between 1½ and 2 h. The EBIS diet history method for determining magnesium intake has been validated against weighed food records with good results (Landig et al, 1998).
Because certain Swiss foods differ from German foods, the BLS nutrient database was adapted to Swiss customs and special food items were added to the database. The adaptation of the database to Swiss foods included the choice of food items from the BLS (only about 1100 food items from the total BLS were used for the interview) and the addition of missing foods and dishes. Missing foods were added to the database when the exact nutrient values (eg for foods found in other food tables) were known, otherwise they were composed as recipes using unprocessed food items from the BLS. Added foods included different mineral waters on the Swiss market and missing food items for which data were found in other food composition tables. To estimate the magnesium value for ingested tap water, the mean magnesium content of Zurich's water supply was used. Recipes were composed for fortified foods on the Swiss market (breakfast cereals, fortified milk products, milky breakfast beverages, fortified flours, cereal bars, bakery products, fruit beverages) and for special Swiss dishes. Because Swiss recipes for breads differ from German ones, and because of their importance in the Swiss diet, breads were analysed for their magnesium content by atomic absorption spectroscopy (SpectrAA 400, Varian, Mulgrave, Australia). The fat content of many Swiss dairy products (yoghurts, curds, cheese), sausages and meat products differs from the fat content of German products. These products were carefully chosen from the total BLS or adapted by composing recipes.
The mean daily magnesium intake of the diabetic group was compared to both the mean intake of the control group and to the recommended dietary intakes (RDI) for Switzerland (D-A-CH Referenzwerte für die Nährstoffzufuhr, 2000). The proportion of subjects not meeting the RDI was determined and compared between the two groups. The probability approach (Anderson et al, 1982; Gibson, 1990) was also used to compare magnesium intakes with the RDI. This method predicts the number of individuals within a group with nutrient intakes below their own requirements and hence provides an estimate of the prevalence of inadequate intakes. The contribution of magnesium supplements to daily intake and the contribution of the different food groups to the magnesium intake were calculated and compared between the two groups. The extent of under-reporting in this study was estimated by comparing the reported individual energy intakes with estimated energy requirements (FAO/WHO/UNO, 1985; James & Schofield, 1990), with the physical activity level set at 1.55 for men and 1.56 for women based on a sedentary lifestyle or light activity.
Data processing and statistical analysis were done using Excel 97 (Microsoft, Seattle, WA, USA) and SPSS for Windows 10.0 (SPSS Inc., Chicago, IL, USA). Normally distributed data were expressed as means and standard deviations (s.d.) and the unpaired Student's t-test was used for comparison. Variables not normally distributed were expressed as medians and ranges and compared with the Mann–Whitney test. Differences were considered statistically significant at P<0.05.
Mean magnesium intakes of the diabetic and the control group were not significantly different (Table 2). The mean±s.d. daily magnesium intake in the male diabetic and male control subjects was 423.2±103.1 and 421.1±111.0 mg, respectively. The mean daily magnesium intake in the female diabetic and female control subjects was 419.1±109.7 and 383.5±109.7 mg, respectively. The recommended dietary intakes for Switzerland (350 mg for adult men, 300 mg for adult women) were not met by 24.8% of the diabetics and 26.0% of the control subjects. Using the probability approach (Gibson, 1990), 5.4% of the diabetic group and 9.1% of the control group were predicted to have intakes of magnesium below their individual requirements. In both diabetics and controls, magnesium intake was also compared between quartiles of age, but no significant differences were found either in men or in women.
Diabetics reported consuming less energy than the control subjects and foods with a significantly higher magnesium density (per 1 MJ) than the control subjects (Table 2). The mean magnesium density±s.d. in the food consumed by the male diabetic and male control subjects was 50±10 mg/MJ (209±42 mg/1000 kcal) and 43±9 mg/MJ (181±38 mg/1000 kcal), respectively (P<0.001). The mean magnesium density±s.d. of the female diabetic and female control subjects was 55±12 mg/MJ (229±49 mg/1000 kcal) and 47±10 mg/MJ (198±41 mg/1000 kcal), respectively (P=0.001). Women (diabetics and controls) consumed foods with a higher magnesium density than men (51±11 mg/MJ (213±47 mg/1000 kcal) vs 47±10 mg/MJ (195±42 mg/1000 kcal), P=0.005).
Magnesium intake from different foods are shown in Table 3. The contribution of the different food groups to the magnesium intake was compared between diabetic patients and controls. Among the five major sources of magnesium in the diet (cereals and cereal products, milk products, vegetables, fruit and non-alcoholic beverages), there was no significant difference in the percentage magnesium intake from these sources in diabetics and non-diabetics. However, the magnesium intake of the diabetics from meat (6.6 vs. 5.1%), mineral water (6.1 vs 3.1%) and soups and sauces (2.6 vs 2.1%) was significantly higher than the control group. In contrast, the magnesium intake of the diabetics was significantly lower than the control group from alcoholic beverages (0.5 vs 3.0%), nuts and seeds (0.8 vs 2.0%), and sugars, chocolate and sweets (0.8 vs 1.7%). No significant differences were found when the other food groups were compared.
Magnesium supplements accounted for 3.4% of the total magnesium intake when diabetic and non-diabetic subjects were combined. Of the 40 subjects taking magnesium-containing supplements, 24 were taking supplements containing only magnesium and 14 were taking magnesium in combination with other minerals and vitamins (two subjects were taking other medical preparations containing magnesium). Diabetics took more magnesium supplements than controls (23.7 vs 17.0%) and women took more supplements than men (29.2 vs 13.0%). Consumption of multivitamin and/or multimineral supplements did not differ between the diabetics and controls, but supplements containing magnesium alone were more frequently consumed by diabetics (14.4 in diabetics vs 10.0% in controls) and by women (20.2 in women vs 5.6% in men).
Potential under-reporting was compared between the diabetic and the control subjects. In the diabetic patients, mean±s.d. reported energy intake (8478 kJ (2026 kcal) in men, 7268 kJ (1737 kcal) in women) was 21.6±18.7% lower than the mean estimated energy requirement (10 908 kJ (2607 kcal) in men, 9386 kJ (2243 kcal) in women). In the non-diabetic subjects, however, mean±s.d. reported energy intake (9693 kJ (2317 kcal) in men, 7798 kJ (1864 kcal) in women) was only 6.9±19.7% lower than the mean estimated energy requirement (10 459 kJ (2500 kcal) in men, 8411 kJ (2010 kcal) in women). There were no significant differences between men and women in both groups.
The magnesium intake of the patients with type 2 diabetes and healthy controls was considerably higher than the magnesium intakes reported in the ARIC study (Ma et al, 1995) and the study of Schmidt et al (1994), both from the USA. Schmidt et al, using a 3 day food record in 50 type 2 diabetics, found an intake of 336.8 mg/day for men and 216.5 mg/day for women (mean age 57.2±10.2 y). The ARIC study investigated dietary magnesium intakes in 15 248 participants using a food frequency questionnaire. In white male and female diabetics they found a magnesium density of 42 mg/MJ (175 mg/1000 kcal) and 42 mg/MJ (174 mg/1000 kcal), respectively, compared to 38 mg/MJ (157 mg/1000 kcal) and 41 mg/MJ (170 mg/1000 kcal) in white disease-free subjects. Our findings agree with the ARIC study, in that diabetics consumed foods with a higher magnesium density than controls, but the overall density was considerably higher in our Swiss population.
The differences between our results and data from the US may reflect differences in overall dietary habits between the inhabitants of the two countries. The USA's 1977–1978 Nationwide Food Consumption Survey (Morgan et al, 1985) found that the magnesium intakes for all age/sex classes were below the 1989 US RDA (Food and Nutrition Board, National Research Council, 1989). Mean daily magnesium intake of adult men and women was 301 and 223 mg, respectively. Data from the Total Diet Study 1982–1989 in the USA (Pennington & Young, 1991) produced similar results: means of 259 mg in 60–65-y-old men and 195 mg in 60–65-y-old women. In Switzerland, estimates based on disappearance data from 1994/1995 suggest that the magnesium intake is adequate: 406 mg/day in adults >15 y (Fourth Swiss Nutrition Report, 1998). Differences in dietary habits might be due to a higher consumption of magnesium from cereal products in Switzerland. In the USA, cereals account for 17–18% of magnesium intake (US Department of Health and Human Services, 1989; Pennington & Young, 1991) compared to 23% in our study.
Large surveys in other European countries have demonstrated that magnesium intakes were slightly below the recommended dietary intakes. The German NVS study which assessed 23 000 subjects using a 7 day food record, showed that mean daily magnesium intake of the adult men was at or slightly below the RDI of 350 mg (325–354 mg, depending on the age group), and magnesium intake of the adult women was below the RDI in all age groups (270–282 mg; Ernährungsbericht, 1996). Results of the SU.VI.MAX cohort in France which investigated magnesium intake in 5448 subjects using six 24 h recalls were similar. The mean magnesium intake in adult men and women was 369 and 280 mg, respectively (Galan et al, 1997). In Belgium, analysing duplicate portions of food collected over several 24 h periods, a mean magnesium intake of 271 mg/day was found (Hendrix et al, 1995). It is difficult to judge if magnesium intakes in Switzerland are truly higher than in other Euro-pean countries as only disappearance data are available for Switzerland, which are usually higher than intake data from direct dietary assessments (Schneider, 1997).
Five food sources (cereals, milk products, vegetables, fruit, non-alcoholic beverages) accounted for 64% of magnesium intake in both diabetics and non-diabetics. Magnesium intake from alcoholic beverages, and sugars, chocolate and sweets (including candies, ice-cream, honey and syrup) of the diabetics was significantly lower than the control group, although intake from these sources was low (0.5–3%). This is likely due to the fact that the diabetics in our study were advised by dieticians to avoid excess simple sugars and alcoholic beverages.
We chose the diet history method as the appropriate assessment tool for this investigation. Although 24 h recalls are easier to carry out, they evaluate food intake only on single days, and a minimum of 4–6 days are required to assess magnesium intake in adults accurately (Nelson et al, 1989). A food frequency questionnaire also assesses dietary habits, but because magnesium is widely distributed in foods, a prohibitive number of food items need to be included in the questionnaire. Although weighed food records are accurate, they are time-consuming and intensive. The diet history provides information on habitual dietary intake and has a relatively low respondent burden compared to a weighed food record (Gibson, 1993). Our sample was moderately large and only group intakes were evaluated. In general, the diet history method yields good precision when used for a group, especially over a relatively short time frame (Van Staveren et al, 1985). The reproducibility of the diet history method used in this study is very good (Reshef & Epstein, 1972; Morgan et al, 1978). In some studies, the diet history method has produced higher estimates of group mean intake than the food record (Gibson, 1990). Other investigators have found similar or lower mean nutrient intakes when comparing the diet history with food records (Van Staveren et al, 1985). Landig et al (1998) compared nutrient intake obtained by the EBIS diet history method with a weighed food record over the same observation period and has received similar results for magnesium with both methods (with EBIS mean magnesium intake was 3.0% lower than with the weighed food record). Major sources of potential errors in the diet history method are reliance upon memory, incorrect estimation of portion sizes, self-selection of participation subjects, coding errors and errors due to different interviewing techniques (Gibson, 1990). To minimise these potential sources of error, we used a photo catalogue to estimate portion sizes, a computer program to standardise the interviewing technique, and the nutrient databank was checked and corrected for mistakes.
Under-reporting occurs frequently in dietary assessment. In the present study, the energy intake reported by the diabetics was 21.6% lower than their estimated energy requirement. This could have been due to under-reporting, as patients with type 2 diabetes are often overweight and are aware they should eat less, or could reflect reduced energy intake in an attempt to lose weight. The tendency towards under-reporting appeared to be higher among the diabetics than the control subjects. This suggests that if our data include biases due to under-reporting, the measured magnesium intake of diabetics in this sample may actually be higher than reported here.
For their assistance in this project, we thank the dieticians of the Division of Endocrinology and Diabetes of the University Hospital of Zurich, and the programme for older adults at the University of Zurich.
About this article
Urinary excretion of an intravenous 26Mg dose as an indicator of marginal magnesium deficiency in adults
European Journal of Clinical Nutrition (2006)