In: Magnesium in Health and Disease, Y Itokawa and J. Durlach, Eds., John Libbey Press, London, Paris, 1989; 565-571
The belief that the mineral content of drinking water may influence cardiovascular death rates has attracted a great deal of attention since Kobayashi (1957) and Schroeder (1960) published their original observations.
It remains possible that some of the correlations that have been observed between death rates and water hardness are nothing more than statistical artefacts, but this discussion will be restricted to the Province of Ontario, where the evidence in favor of a cause-and-effect relationship is reasonably strong, and where magnesium (Mg) appears to be the substance responsible.
Ontario covers a large area, stretching approximately 1000 miles in both north-south and east-west directions. In the north the water is very soft (i.e. low in minerals) and generally contains no more than 2-3 mg of Mg/per liter, while in the extreme south there is an area with water that is very hard due to high concentrations of both calcium (Ca) and Mg. Here the drinking water may contain over 100 mg/per liter of Ca, and up to 40 mg/per liter of Mg. Since the average North American diet supplies about 200 mg of Mg per day (Schroeder, 1969), an intake of 1-2 liters of some of these southern Ontario water supplies could therefore increase daily intake by as much as 20-40%, which could be crucially important, if, as Seelig (1964) has claimed, the usual dietary intake of Mg is barely adequate.
In our studies we have attempted to identify which of the several links in the chain of events leading to death from heart attack is responsible for the increased death rate seen in the soft-water areas; a second line of attack has been to look for substance in the blood or tissues that shows a variation in concentration across the province parallel to that seen in local water supplies.
Our first study (Anderson et al., 1969) was prompted by the observation (Crawford and Crawford, 1967) that despite a large difference in the death rates from heart disease in Glasgow (soft water) and London (hard water), the frequency of coronary atherosclerosis was similar in both places. Since the excitability of muscles and nerves is known to vary with the concentration of Ca and Mg (the major "hardness" minerals), we examine the Ontario data for evidence of an excess of sudden deaths (due to abnormal electrical rhythms) in the soft-water areas.
We divided the province into three areas of water hardness, less than 100 ppm, 100-200 ppm, and greater than 200 ppm, and examined mortality data over a 3-year period, 1965-1967. The only index of sudden death available on the official data cards was whether the death had been certified by a coroner, rather than by a private physician. Having first established that there was little or no gradient across the three areas in the proportion of noncardiac deaths certified by coroners. we examined ischemic heart disease (IHD) deaths and found that there was indeed a steep gradient, with sudden IHD death rates being almost twice as high in the soft-water area as in the hard-water area (Fig. 1).
We could find no evidence that this excessive sudden-death rate was related to the more severe winters in the soft (northern) part of the province (Anderson and La Riche, 1970) or to a number of socioeconomic variables (Anderson and La Riche, 1971). Furthermore, a detailed investigation of all deaths occurring over a 3-year period in two cities near the extremes of the water hardness spectrum, confirmed that the proportion of IHD deaths occurring suddenly was higher in the soft-water town (Anderson and Le Riche. 1971), and the difference was apparent in both sexes and at all ages (Fig. 2).
If a cardiac water factor really exists, it is presumably something beneficial in hard water or something toxic in soft water. In either case, one might expect to find evidence of a difference in blood or tissue levels of the responsible substance between residents of soft and hard-water areas. The work of Crawford and Crawford (1967) and Bierenbaum et al. (1973) encouraged us to believe that the link was more likely to a beneficial hard-water factor (probably Ca. but possibly Mg) than a toxic soft-water factor, and we therefore arranged for serum samples to be collected from middle-aged residents in two Ontario cities, one with very hard water, the other with very soft water, and for the samples to be analyzed for Ca and Mg. In addition we compared x-rays of the hand in the two areas to see if there was any difference in the degree of bone mineralization. The findings in this study were entirely negative (Fig. 3); there were identical mean serum levels for both Ca and Mg, and no difference in the x-ray measurements.
Up to this time we had favored Ca as the more likely protective element of the major "hardness" cations. However, since the differences in serum levels were negligible, intracellular concentrations might be more important, and Mg is a much more plentiful ion within the cell (particularly in heart muscle) than is Ca. We had also become aware of Seelig's evidence (1964) that Western diets are barely adequate in Mg, so that the relatively small water-borne intake might be crucial. The findings of Heggtveit et al. (1969) that the myocardium was low in Mg in persons dying of heart attacks also lent strength to the idea that Mg, rather than Ca, was a more likely candidate for the Ontario "water factor."
We therefore arranged with pathologists in a number of softand-hard-water cities in Ontario for a portion of heart muscle to be retained from each routine autopsy performed over a 4-month period (Anderson et al., 1975). The apex of the heart was examined in each case. This was done to ensure uniformity of sampling site and to decrease the likelihood of obtaining an area of very recent (and therefore inapparent) myocardial infarction.
Myocardial samples were obtained from 54 violent deaths in 5 soft-water cities, and from 29 violent deaths in 3 hard-water cities. The mean Mg concentration was 918 mcg per g (dry tissue) in the soft-water residents, some 7% lower than the mean of 982 mcg per g in the hard-water residents. This difference was statistically significant at P < 0.01. No significant differences were seen in the concentration of Ca (soft 224, hard 232 mcg per g), zinc (103, 102), copper (15.7, 16.4). chromium (0.9, 0.13), or cadmium (0.14, 0.14). Lead was detectable at a level of 0.01 mcg per g or greater in only 26% of the cases in the soft- water area, and 21% in the hard-water area.
No significant differences were found in the distribution according to sex, age, or history of alcohol abuse in the softand hard-water areas (Table 1). Similarly, suicides, as well as accidental deaths, and traumatic, as well as "chemical" deaths (e.g. drug overdose or carbon monoxide poisoning), showed essentially the same geographical differences.
Samples of diaphragm and pectoralis major muscles were also analyzed, as examples of another continuously active muscle and a typical skeletal muscle. Neither of these muscles showed a statistically significant difference in Mg concentration between the two areas.
Samples were also obtained from a total of 40 IHD deaths. As previously reported by Heggtveit et al. (1969), the myocardial Mg levels were much lower than in the accident cases, with mean values of 697 and 744 mcg per g in the soft- and hard-water areas, respectively; over 20% below the corresponding values in accidental deaths.
Since these results were published, an apparently contradictory geographic pattern has been reported from England (Chipperfield et al., 1976). In this study, heart muscle samples from non-cardiac deaths were compared from two cities, Hull (hard water) and Burnley (soft water), and a 19% difference found in the "wrong" direction. However, by Ontario standards the Mg content of both water supplies was very low (Hull 5 and Burnley 2 mg per liter), and since the difference between them represents barely 2% of probable dietary intake, water is unlikely to make any difference to total supply. (Magnesium levels in most British water supplies are relatively low compared to Ontario, so that if there is a real "water factor" in Britain it is probably something other than Mg.) The reason for the differences in myocardial Mg reported by Chipperfield et al. is unclear, but some procedural difference is a distinct possibility, since an even larger difference (37%) was found in their myocardial potassium concentrations.
At least some of the geographical variation in heart disease mortality in Ontario may be related to a marginally inadequate dietary intake of Mg. In the hard-water area of the province, water-borne Mg increases total daily intake by at least 20% so that residents of this area are less likely to be Mg deficient. This hypothesis is supported by the finding that the concentration of Mg in the myocardium tends to be higher in residents of the hard-water area, and it is consistent with the observation that fatal cardiac arrhythmias are less common than in the soft-water area.
Death rates from IHD in the Province of Ontario show an inverse correlation with hardness of local water supply, due to an excess of sudden deaths in the soft-water area. An autopsy comparison of accident victims in the soft- and hard-water areas showed that the mean myocardial Mg concentration was 7% lower (P < 0.01) in the soft-water area. No significant geographical differences were seen in Mg concentrations in diaphragm or pectoral muscles, or (in previous studies) serum concentrations or skeletal mineralization. The autopsy comparison also confirmed the previously reported low levels of myocardial Mg in victims of heart attacks. It would appear that, in Ontario at least, the contribution of water-borne Mg to total dietary intake may be critical, and that some residents of soft-water areas are in a state of subclinical Mg deficiency. If these individuals suffer a myocardial infarction, they may then be at an increased risk of developing a fatal cardiac arrhythmia, leading to sudden death.
Anderson. T.W., Le Riche, W.H.. and MacKay. J.S. Sudden death and Ischemic heart disease: Correlation with hardness of local water supply. New England J. of Med. 280, 805-807 (1969).
Anderson, T.W. and Le Riche, W.H. Cold weather and myocardial infarction. Lancet, 291-296 (1970).
Anderson, T.W. and Le Riche, W.H. Sudden death from ischemic heart disease in Ontario and its correlation with water hardness and other factors. Can. Med. Assoc. J. 105, 155-160 (1971).
Anderson, T.W. Serum electrolytes and skeletal mineralization in hard and soft-water areas. Can. Med. Assoc. J. 107, 36-37 (1972).
Anderson, T.W., Neri, L.C., Schreiber, G.B., Talbot, F.D.F., and Zdrojewski, A. Ischemic heart disease, water hardness and myocardial magnesium. Can. Med. Assoc. J. 113, 199-203 (1975).
Bierenbaum, M.L., Fleischman, A.I., Dunn, J.P., Hayton, T., Pattison, D.C., and Watson, P. Serum parameters in hard and soft water communities. Am. J. Public Health 63, 169-173 (1973).
Chipperfield, B., Chipperfield, J.R., Behr, G., and Burton, P. Magnesium and potassium content of normal heart muscle in areas of hard and soft water. Lancet 7, 121-123 (1976).
Crawford, T. and Crawford. M.D. Prevalence and pathological changes of ischemic heart-disease in a hard-water and in a soft-water area. Lancet 1, 229-232 (1967).
Duruisseau, J. P., Lapointe, J., and Laurendeau, E. Teneur en magnésium de la diète servie à l'Hôpital Notre-Dame de la Merci.. Can. J. Public Health 67, 30-32 (1976).
Heggtveit, H.A., Tanser, P., and Hunt, B. Magnesium content of normal and ischemic hearts. 7th International. Congress. of Pathol., Montreal, July 13-19 (1969), p. 53.
Kobayashi, J. On geographical relationship between the chemical nature of river water and death-rate from apoplexy. Bar. Ohara Inst. Landwirt. Biol. 11, 12-21 (1957).
Schroeder, H.A. Relation between mortality from cardiovascular disease and treated water supplies. J. Am. Med. Assoc. 172. 1902-1908 (1960).
Schroeder, H.A., Nason, A.P., and Tipton, I.H. Essential metals in man: Magnesium. J. Chronic Dis. 21, 815-841 (1969).
Seelig, M.S. The requirement of magnesium by the normal adult. Am. J. Clin. Nutr. 14, 342-390 (1964).
This page was first uploaded on June 19, 1996