RELATIVE DEATH-RATES OF SELF-DECLARED ABSTAINERS

AND MODERATE DRINKERS FROM THE

ACTUARIES' VIEWPOINT

By Edward Bunnell Phelps

An article under the above heading appeared in The American Underwriter Magazine and Insurance Review for June, 1915, and was published after the death of the author. Mr. Phelps had given an enormous amount of research and study to the question of the comparative longevity of abstainers and moderate drinkers. In this last article from his pen he reviews all the available evidence and points out the conclusions which it warrants.

In introducing the subject Mr. Phelps draws attention to the incompatibility between scientific data and emotional propaganda, saying:

"It is therefore unfortunate that for many years the advocates of Prohibition have persistently tried to persuade the thinking world that the actual effect of moderate drinking on human longevity may be scientifically calculated and practically reduced to exact figures by means of certain extremely technical life insurance statistics. Conclusive evidence of this world-wide effort on the part of the crusaders is to be found in innumerable books, pamphlets, papers and addresses, including many volumes bearing a scientific imprint. Of late the movement has been noticeably accelerated, as is demonstrated by the steadily increasing amount of space given in the lay magazines and newspapers and more or less technical periodicals to this phase of the prohibitionists' argument.

"No matter how much space may be given to exploiting the contention that life insurance experience has practically established the relative death-rates of those who do not, and those who do, drink alcohol-commonly classified as Abstainers

and Non-Abstainers—the proposition is bound eventually to be thrown out of the scientific court of last resort unless it can stand the ordeal of trial and critical examination in accordance with the well-known rules of procedure of that court. Every one of the alleged facts and figures cited in evidence must be substantiated, and the sequence and causal relations of the statistical data must be positively established before the sweeping conclusion so confidently drawn from them can be accepted as a scientifically established fact. If that point has been reached, the world at last has the real measure of the physiological effect of the moderate use of alcohol on the ordinary

man.

"After examining the subject and its collateral phases from the statistical viewpoint for several years, I have recently concluded a study of all the figures and text available in the discussions of the subject recorded in the transactions of the actuarial societies of Great Britain and this country; have obtained some illuminating actuarial data hitherto unpublished, and am more than ever confident that the measure of the actual effect of moderate drinking on the ordinary man is still, and in all probability for many a year will be, an unsolved problem. The facts and processes of reasoning responsible for this conclusion are below stated in detail, and the authorities for the facts in each case specified. Thus provided with the facts on which my conclusions are based, the reader will be in a position to work out his own conclusions, and I am ready to let the soundness of my conclusions be determined by the consensus of competent, unbiased conclusions based on these same facts."

Mr. Phelps deals exhaustively with the returns of various companies which have been utilized to demonstrate the relation between alcohol and the expectation of life, also quoting copiously from the statements of various actuaries of international reputation who at one time or another have had the subject before them.

To American readers, the statistics drawn from American and Canadian life insurance companies are of especial interest, and they are therefore given below without abbreviation and are followed by Mr. Phelps' summary of all the evidence cited.

A PUZZLING SHOWING OF THE MEDICO-ACTUARIAL

INVESTIGATION

MORTALITY

The results of the prolonged investigation of the mortality experience of forty-three American and Canadian life companies, from 1885 to 1909, conducted in 1909-1913 by a joint committee of the Actuarial Society of America and the Association of Life Insurance Medical Directors, and widely known as the MedicoActuarial Mortality Investigation, have been published in five volumes, and include tabulations of the 43 companies' experience with several classes of policyholders grouped from the standpoint of their affiliation with one form or other of the liquor industry, their confessed former excessive use of alcohol, or their steady free use of alcohol. In Class 18, Group K, classified under the heading of "Habits as to Alcohol, Steady Free User," about 42,000 cases were considered, and in respect to this class the report says:

The following is a synopsis of the results in Group K:

As there were known to be considerable differences in the interpretation placed by the companies on a "steady free use" of alcohol, the statistics were divided into two sections, according to whether a conservative or a liberal interpretation had been adopted. In the former section were placed the risks from those companies which considered that two or more glasses of beer or one glass of whiskey or their equivalents per day constituted a steady free use. In the latter section were placed the risks from those companies which used a standard as high as or higher than Anstie's limit of two ounces of alcohol per day. The following are the results of this investigation:

The data were divided into two groups by entry ages:

In the "Liberal" Section the death-rate from cirrhosis of the liver was five times the normal, and from diabetes, tuberculosis, pneumonia and suicide twice the normal.

In so far as the mathematical accuracy of the above tabulation of the mortality experience of the companies figuring in it is concerned, there can be no question. But as to the real significance of the figures there is room for decided question, the figures on their face showing a ratio of 186 per cent for actual to expected mortality in the case of the supposedly moderate drinkers grouped under the heading of "Liberal Interpretation" as men averaging two ounces or more of alcohol per day, as compared with a ratio of only 118 per cent in the case of men grouped under the heading of "Conservative Interpretation" as men using two or more glasses of beer or one glass of whiskey or their equivalent per day. The former ratio is higher than the latter ratio by 57.6 per cent, or, in other words, would indicate a mortality more than one-half again as high, and, taking into account the generally-conceded inaccuracy of the non-abstaining applicants' statements as to their average daily consumption of alcohol, and the fact that only supposedly "moderate drinkers" would be accepted by any sound. companies, it would seem so utterly improbable as practically to be impossible that the applicants owning up to a daily average of two ounces of alcohol would have a mortality more than one-half again as high as that of the other non-abstainers who confessed to a daily average consumption of "two or more glasses of beer or one glass of whiskey or their equivalents per day." Obviously, by no means all of the users of alcohol to excess included in the tabulation would be in the class of policyholders who on their original examination had owned up to a daily average of two ounces of

alcohol, and as the total number of deaths in this class in the twenty-five years covered by the investigation was but 698, or less than 28 per year, the suspicion would naturally arise that the total number under observation in this class was too small to warrant sound conclusions.

The apparent pronounced discrepancy between the mortality rates of the two classes of non-abstainers, a difference of 57.6 per cent, furthermore, will be found by comparison with the figures of Mr. McClintock's paper presented on previous pages of this study of the subject to be considerably more than twice as high as was the difference between the entire group of non-abstainers and total abstainers in the Mutual Life's experience from 1875 to 1889, the ratio of actual to expected mortality in the case of non-abstainers being but 96 per cent as compared with a ratio of 78 per cent in the case of total abstainers, or barely 23 per cent higher. And, it will be remembered, Mr. McClintock found from the Mutual Life's experience that after the policies had been in force four years there was a difference of only about 10 per cent between the mortality experience of abstainers and non-abstainers, and he accounted for the greater part of this difference by making allowance for various irregularities in the habits of, and statements made by, the nonabstaining applicants, and also held that "the whole difference of 10 (or 8) per cent cannot be due solely to the physical effect of alcohol, because those addicted to alcohol comprise among their number a much larger proportion than the abstainers of persons naturally weak and vicious, who would on the average die earlier than others if alcoholic beverages were unknown. Some sicken because they drink, and others drink because they are infirm."

THE HITHERTO UNPUBLISHED EXPERIENCE OF THE NORTHWESTERN MUTUAL LIFE

The only other investigation made of the mortality of any great American life company, measured by its experience with its nonabstaining policyholders in the light of their avowed drinking habits at the time of their acceptance-in so far as publicly known-was one conducted several years ago by one of the largest and most conservative of all companies, the Northwestern Mutual Life In