American Journal of Epidemiology, Vol. 94 August, 1971 No. 2
Abbreviation: CHD, coronary heart
1Dept. of Epidemiology and Community Medicine, University of Ottawa, Ottawa, Ontario, Canada.
2School of Hygiene, University of Toronto, Toronto, Ontario.
This research was supported in part by grant #605-7-609 from the Department of National Health and Welfare, Federal Government, Canada.
Neri, L. C. (Dept. Epidemiology and Community
Medicine, University of Ottawa, Ontario, Canada), D. Hewitt and
J. S. Mandel. Risk of sudden death in soft water areas.
Amer J Epidem 94: 101-104, 1971.—This paper
continues discussion of a reported association between
suddenness of death from coronary heart disease and residence
in soft water areas. An attempt has been made to establish, for
the city of Ottawa, the extent to which coroner-certified
deaths correspond to a classification of sudden/nonsudden. It
is suggested that the predictive value of water hardness may be
related to trace metal contents.
arteriosclerosis; coronary heart disease; death, sudden; epidemiology; water supply
It is widely recognized that coronary heart disease (CHD) is not, from the epidemiologic point of view, a single entity. For example, both cigarette smoking and physical inactivity (1) have been shown to relate mainly to the risk of sudden death from CHD rather than to the risk of more chronic forms of the disease. It was recently suggested by Anderson (2) that excess CHD mortality of residents in soft water areas may also take the form of sudden death. The basis for this suggestion was a statistical analysis of deaths occurring during 1967 in the Province of Ontario whose counties had, for the purpose of the study, been grouped into three ranges of average water hardness. Separation of sudden from nonsudden deaths was made on the basis of whether or not the death certificate was signed by a coroner. The reason for using the coroner/noncoroner classification was simply one of convenience as it can be determined from machine-readable records, whereas the more direct indications of suddenness can only be retrieved by search of the original death certificates. It was shown that, during 1967, the percentage of certificates signed by a coroner was highest (47 per cent) in the counties with softest water, and lowest (33 per cent) in those with the hardest water. No comparable trend was detected for other causes of death, the percentage of coroners' certificates being 17 per cent in soft water counties and 14 per cent in hard water counties.
Selected items of information concerning all deaths of Canadian residents are retained on magnetic tape in Ottawa. Copies of the tapes for 1960, 1961 and 1962—containing no individual identification—were kindly provided by the Vital Statistics Section of the Dominion Bureau of Statistics. They have been sorted by age, sex, residence and certified cause of death to obtain numerators for mortality rates based on 1961 Census populations. An interim report of this study has already been presented (3) and fuller details will appear elsewhere. We are concerned here only with data corresponding to those analyzed by Anderson (2). Our units of study were individual municipalities (all having a municipal water supply) of which there are over 200 in Ontario, compared to only 54 counties, and which therefore offer a better means of separating out the populations exposed to water of different degrees of hardness. Table 1 shows, for the three years 1960-1962, the percentage of coroners' certificates among all certificates citing CHD as the underlying cause of death of male residents in municipalities whose water supplies fell into each of three ranges of hardness. As was previously found by Anderson, this percentage is appreciably higher (by a factor of 1.5) for the soft water areas in Ontario (<100 ppm), than for the hard water areas (>200 ppm), but the figure for the intermediate range of hardness, instead of falling between the values for hard and soft water areas, here shows as the lowest of the three. A more important difference between the present figures and those previously reported is that the percentage of coroners' certificates among all certificates citing other causes of death (noncardiac, non-accident) is also appreciably higher (by a factor of 1.77) in soft than in hard water areas. Thus the Ontario correlation between soft water and a high percentage of death certificates signed by a coroner probably has to be explained in terms more general than those of any postulated relationship between soft water and the risk of sudden death from CHD. A likely partial explanation is that in Ontario, as in Canada generally, average hardness and coroner utilization rates are both related to size of community.
Anderson had it in mind to counter this possible explanation for his findings when he showed that similar results could be obtained from the 1967 figures with or without the inclusion of cities over 100,000. However, as table 2 shows, this exclusion does not suffice to remove the confounding effect of community size.
Interpretation of figures based on a single year or a period as short as three years is made difficult by the fact that utilization of coroners' services in Ontario (and Canada) has not remained steady over the years. Though the general level appears to have been much the same in 1967 as it was during 1960-1962, we know that as recently as 1950-1952 the percentage of coroners' certificates was lower by more than a half.
Another interpretative problem arises from uncertainty experienced by various investigations (4) about the extent to which a classification of certificates as coroners/noncoroners corresponds to any classification as sudden/nonsudden. We were able to test this in Ottawa for the year 1966.
Out of 2,500 death certificates reviewed, there were 316 male deaths at ages 35-74 that fell in the category 420 + 422 (CHD). Statistical evaluation is complicated by the fact that a proportion of certificates contained no statement concerning the interval between onset and death, but clearly the frequency of sudden death defined here as within 15 minutes of onset, in this group could not be less than 0 per cent or more than 100 per cent. Applying each of these extreme assumptions in turn, it can be computed from the figures in table 3 that a coroner-certified death from coronary disease was 2.1 to 2.3 times as likely to have been "sudden" as one certified by a noncoroner. Compared with coronary deaths as a whole the relative likelihood that coroner-certified death would have been sudden was only about 1.2 to 1.3.
In the opinion of the present writers, the Ontario statistics so far examined, taken as a whole, hardly sustain Anderson’s hypothesis. Nevertheless, because it appeared to us that this hypothesis had heuristic value, we analyzed data from seven of the nine other Provinces [Newfoundland had no coronary deaths certified by a coroner during the entire period 1966-1962, while in Prince Edward Island only two localities were available for the study.] in the hope that some research leads would emerge (see table 1). In three of these (Quebec, Manitoba, Saskatchewan), communities within the softest range showed a lower percentage of coroners’ certificates than one or both of the relatively hard water groups. In the two Atlantic Provinces studied, coroners did tend to certify more of the CHD deaths of residents in soft than in hard water areas, but the level of coroner utilization was so low that it was impossible to judge whether this tendency was characteristic of CHD in particular or whether, as in Ontario, it applied to deaths in general. There remain the two westernmost Provinces (Alberta, British Columbia), in each of which the percentage of "coroners' coronaries" followed the pattern noticed by Anderson in the 1967 Ontario figures, and in each of which, most satisfactorily, the corresponding percentages for other causes of death exhibited a contrasting arrangement. Could it be more than a coincidence that the report by Peterson et al, (5) is based on data collected in an area of the United States adjacent to British Columbia? These workers set out to reproduce Anderson's results as closely as possible, using the same ranges of water hardness, and the same criterion of sudden death. When this failed, they substituted another criterion (died at home or dead on arrival at hospital) that was believed, under Washington conditions, to provide a stronger presumption of sudden death than did certification by a coroner. On this basis, they felt they were able to demonstrate an association of the type sought. Yet, on examining their findings, it can be seen that the contrast between water hardness ranges was much greater for sudden deaths not attributed to arteriosclerotic heart disease (ASHD) (mortality ratio of 1.5) than for sudden deaths that were attributed to ASHD (mortality ratio of only 1.1). Hence if the phenomenon reported by Peterson et al. is a genuine one at all it certainly cannot be regarded as something related to ASHD in patients. Also, it is important to note that Washington State has no very hard water (>200 ppm) and therefore comparisons have to be made between only two ranges of water "softness". This guaranteed the authors at least a 50:50 chance of obtaining qualitative agreement with any earlier findings.
The percentage of deaths certified by coroners is, of course, a "proxy variable" whose validity as an index of the frequency of sudden death in different areas cannot be assumed without local knowledge. In all likelihood, hardness of water is another such variable, whose predictive value depends on the degree to which, in a particular geologic setting, hardness correlates with the presence or absence of some water constituent having real physiologic significance. Our statistical analyses, to be reported in detail elsewhere, suggest that in Canada generally, water hardness does not have the value as a predictor of CHD mortality rates that it has been shown to have in the United States, England, etc. (6-10). We have therefore launched a nation-wide collection of water samples and are making determinations of a number of trace elements, among which (in view of the recent report by Voors (11)) we now propose to include lithium as well as chromium, zinc, copper, cadmium, molybdenum, vanadium and magnesium.
1. Kannel WB, Castelli WP, McNamara PM: The coronary profile: 12-year follow-up in the Framingham Study. J Occup Med, 9: 611-619, 1967
2. Anderson TW, le Riche WH, MacKay JS: Sudden death and ischaemic heart disease. New Eng J Med 280: 805-807, 1969
3. Neri LC, Mandel JS: The relationship between water hardness and cardiovascular mortality in Canada (an interim report). Presented at the Society for Epidemiological Research, Minneapolis, USA, June 1970
4. Kuller L: Sudden and unexpected non-traumatic deaths in adults: A review of epidemiological and clinical studies. J Chron Dis 20: 1165-1192, 1967
5. Peterson DR, Thompson DJ, Nam J: Water hardness, arteriosclerotic heart disease and sudden death. Amer J Epidem 92: 90-93, 1970
6. Schroeder HA: Relation between mortality from cardiovascular disease and treated water supplies: Variations in states and 163 largest municipalities of the United States. JAMA 172: 1902-1908, 1960
7. Morris JN, Crawford MD, Heady JA: Hardness of local water-supplies and mortality from cardiovascular disease in the county boroughs of England and Wales. Lancet 1: 860-862, 1961
8. Biorck G, Bostrom H, Widstrom A: On the relationship between water hardness and death rate in cardiovascular disease. Acta Med Scand 178: 239-251, 1965
9. Schroeder HA: Municipal drinking water and cardiovascular death rates. JAMA 195: 81-85, 1966
10. Crawford T, Crawford MD: Lead content of bones in a soft and hard water area. Lancet 1: 699-701, 1969
11. Voors AW: Lithium in the drinking water and atherosclerotic heart death: Epidemiologic argument for protective effect. Amer S Epidem 92: 164-171, 1970
This page was first uploaded to The Magnesium Web Site on September 30, 2002