insured, and definite provisions for rendering the health-insurance system an aid to disease prevention. It has been proposed that the preventive force of governmental health insurance should not be limited to the financial relief during sickness, to the medical service afforded, and to the possible economic incentive to reduce sickness, but that it should be greatly increased by linking the health-insurance system to the existing public health agencies. In this sense, "sickness insurance," it is believed, would become a real health measure. It would not be merely a variety of commercial or mutual insurance or another type of public relief, but a practicable method of improving and extending the present facilities for the prevention of disease. From the viewpoint of the physician and of the public health official, the principal points which suggest themselves for the consideration of "health insurance" are as follows:

1. The sickness expectancy, i. e., the amount of sickness for which medical and surgical service must be provided.

2. Methods of providing adequate medical and surgical relief. 3. Methods of adequate prevention of sickness.

1. Sickness Expectancy.

Although in the absence of accurate statistics of morbidity in the United States it is impossible to arrive at accurate estimates of the amount of sickness occurring among wage earners, nevertheless considerable information concerning sickness expectancy may be obtained by a study of the experience of establishment sick benefit funds. Several estimates have been ventured, some of which have been based on extremely scanty material and some on more reliable data from surveys of actual sickness in industrial communities and from records of disability among employees of establishments. The wide difference in these estimates, from 6 to 9 days of sickness a year per wage earner, has served to call attention to the urgent need for accurate statistics.1

(a) Investigations concerning sickness expectancy.-In the last two years the results of several "sickness surveys" or censuses have been published and have added materially to the very scanty American morbidity experience previously existing.

By the survey or census method the number of persons found sick on a given day in an enumerated population and recorded, affords the basis for computing the sick rate per 1,000 of the censused population as a whole or in sex, age, and other groups. In 1915-1917,

'The American Association for Labor Legislation in 1911 estimated that the American wage earner loses on an average 8.5 days per year on account of sickness. The Federal Commission on Industrial Relations, inits staff report estimated from such records as were then available that the average loss of time from disabling sickness and nonindustrial accidents was about 9 days per year per wage earner. The Social Insurance Commission of California in 1917 from a study of the records of American Benefit Association that were collected by the Federal Bureau of Labor a number of years previous and of such data as were available from similar records in California, estimated that the average loss of time per year per person was 6.5 days.

579,197 persons were censused in various localities by agents of the Metropolitan Life Insurance Co.; two censuses were made of certain districts in New York City by the department of health of that city; a survey was made of Dutchess County, N. Y., by the State charities. aid association; and several surveys have been made in a number of textile villages in South Carolina by the United States Public Health Service. Without attempting to present and discuss in detail the variations in rates among persons of different sex, ages, occupations, localities, income, or other conditions, reference may be made to indicate morbidity rates and annual days of sickness per person among populations 15 years of age and over.

In the following table the experience from the above-mentioned sickness censuses is summarized. The results of the Dutchess County survey are not in a form that is comparable with the results of other surveys, and are omitted from the table.

TABLE 1.—Cases of disabling sickness and rate per 1,000 of various populations 15 years of age and over, and indicated average annual number of days of disabling sickness par

person.

1 Warren, B. S., and Sydenstrucker, Edgar: Statistics of Disability-A compilation of some of the data available in the United States Public Health Reports, Apr. 21, 1916.

2 See appendix B: Combined Sickness Experience of the Company's Surveys, 1915 to 1917, of the Metropolitan Life Insurance Co.'s publication, "Sickness Survey of Principal Cities in Pennsylvania, and West Virginia," by Lee K. Frankel, Ph. D., third vice president, and Louis I. Dublin, Ph. D., statistician. The "combined sickness experience" referred to included the results of sickness surveys made in localities in Pennsylvania, West Virginia, and North Carolina, Kansas City (Mo.), Boston, Rochester, Trenton, and Chelsea (New York City).

3 Wynne, Shirley Wilmott: Second Illness Census in the Experimental Health District. Monthly Bulle tin of the Department of Health of the city of New York, November, 1916.

Sydenstricker, Edgar, Wheeler, G. A., and Goldberger, Joseph: Disabling Sickness Among the Population of Seven Cotton Mill Villages of South Carolina, in Relation to Income. Public Health Reports, Nov. 22, 1918.

With reference to the rates in Table 1 it should be noted that the rate for Government clerks is probably for a preferred occupation. The rate approximates quite closely that for office employees afforded in the experience of the Leipsig local sickness fund during 1887-1905.1 The extremely high rate among the population of South Carolina textile villages, on the other hand, is probably due to a relatively low economic status.2

1 See Twenty-fourth Annual Report of the United States Commissioner of Labor, vol. 1, pp. 1281-1341. For a discussion of the sickness rate among persons of different family income in the population cen sused see Public Health Reports for Nov. 22, 1918. Sup. cit.

(b) The authors' investigation concerning sickness expectancy. In the investigation here described, data were collected from over 400 sickbenefit associations, covering, in the majority of instances, an experience of three years, have been collected. These data consist of records of disability due to sickness and nonindustrial accidents for which cash benefits have been paid under the various regulations of the associations, and afford this kind of sickness experience among over threequarters of a million wage earners engaged in many different industries and occupations. The collection and tabulation of the information have not been completed, but it is possible, for purposes of illustration, to present some preliminary figures for groups of wage earners who are members of one or two types of sick-benefit funds. It should be kept in mind that any conclusions suggested by these statistics ought to be regarded as tentative for the reason that more complete data covering a larger sickness experience are yet to be compiled.

More trustworthy information, it is believed, will be afforded when certain inquiries now under way are completed and when the systematic reporting of morbidity among wage earners is begun. An effort is now being made by the United States Public Health Service to collect such statistics of disability as are at present available in the experience among employees of industrial establishments.

For presentation here the disability records of those sick-benefit associations which pay no benefits for the first three days of sickness, or for illnesses of less than four days' duration, have been selected because a similar provision has been included in the health insurance bills that have been introduced in various State legislatures. Data for 22 of these associations have so far been collected. They include approximately 150,000 members,' for the great majority of whom a three years' (1914, 1915, and 1916) experience is available, which makes possible a consideration of 463,714 years of exposure of membership. The regulations of the associations, however, are not uniform with respect to the maximum length of the period for which benefits can be paid; for this reason the statistics are presented accord

1It may be noted that the members of the 20 associations were nearly all males, the females constituting a negligible proportion, and, so far as could be ascertained, were adults of the usual wage-earning age period. They were employed in a variety of industrial plants and in various occupations; their sickness experience, however, is not large enough to permit of accurate indications of the influence of occupation. Since industrial accidents are not included, and since the members are fairly well distributed among different occupations in the groups presented in the table which follows, the occupational factor may be disregarded for the purposes of this illustration. To a considerable extent the members are a selected group; some of the associations require applicants for membership to pass a physical examination and to be under 45 years of age, and nearly all had provisions which operated to exclude casual laborers from their membership. The possible influence of administrative methods and practices upon the sick rate is more difficult to determine; the possible effect of the amount of the cash benefit, however, may be disregarded for purposes of approximation, since, for the most part, the cash benefits provided ranged between one-third and one-half of the wages.

Years of exposure of membership were ascertained from the records of the associations by securing the average memberships for each month in each year and computing the average yearly membership by dividing the total of the monthly membership by 12.

ing to groups of associations having the same or nearly the same maximum benefit period. The statistics follow:

TABLE 2.--Sickness and nonindustrial accident statistics of 22 establishment sick-benefit funds having a three days' waiting period, for 1914, 1915, and 1916: Classified according to length of benefit period.

1 By "years of exposure of membership" is meant the number of members for whom a 1 year's sickness and nonindustrial accident record was obtained. The approximate number of persons who were members of the funds can be obtained by dividing the years of exposure of membership by 3.

It will be noted that, as may be expected, the waiting period being the same for all associations considered, the average days of compensated sickness per case tends to increase according to the maximum length of the benefit period, and determines the trend of the average days of sickness per member. The importance of the length of the benefit period in determining the amount of sickness for which benefits are to be paid under a system of health insurance is thus suggested. The sickness experience covered in the foregoing statistics is too small to afford definite indications of the experience under any given benefit period except, probably, for those associations having benefit periods of 52 weeks or more. For those six associations, with 440,691 years of exposure, we have a rate of 8.8 days of sickness per year per member.

The sickness expectancy for associations having a maximum benefit period of 26 weeks is, however, of especial interest because some of the health insurance bills introduced in State legislatures contain a similar provision. Unfortunately, until the data obtained are more completely tabulated and adjustments made for varying waiting and benefit periods, our statistics are rather meager. The rate of 6 days of sickness per member per year and of 392 cases of sickness per 1,000 members per year for the group of associations having benefit periods of 23 to 26 weeks appears to be conservatively low, especially when it is compared with the indicated experience obtained in several recent "sickness censuses" in the United States, to which "in reference has been made, and with the experience of the German If the average annual case rate of 477 per 1,000 for the entire group of 22 associations included in the foregoing table be used as possibly a more accurate base, the days of sickness per member per year 4 associations with a benefit period of 23 to 26 weeks would be 7.3.

for the

sickness insurance system during the five years prior to the war. With similar waiting and benefit periods, the German experience for the years 1909-1913 showed an average of 8.4 days of compensated sickness per member per year. This was a considerable increase over the rate in 1900 and in years prior, which was about 6 or 7 days.''

While the increase was in some measure undoubtedly due to changes in the provisions of the sickness insurance law, it can be interpreted at least partly as an indication of improvements in the medical care of the sick, of the placing of a greater emphasis upon "medical inadvisability to work" rather than on actual "inability to work" as a principle in determining the return of disabled workers to employment, and of a clearer realization of insured persons as to their rights under the insurance system. It would therefore appear that all of the increase can not be attributed to malingering. Without venturing to assume that conditions affecting the health of German wage earners before the war were comparable in all respects with conditions in this country or that the German sickness rate is any guide to the sickness expectancy here, it seems reasonable to have under consideration the probability that the expectancy of sickness which is to receive cash benefits under State or other health insurance laws in the United States will be larger than that indicated by the experience of existing sick-benefit funds, especially if an adequate medical service is afforded.

Probably a conservative estimate of the total amount of sickness which will require medical service under the proposed health-insurance measures would be something between 8 and 9 days per insured person. This includes, of course, the first 3 days of sickness and sicknesses lasting less than 4 days for which medical service must be

The following table presents the German sickness insurance experience for the years 1885, 1890, 1895, 1900, and 1905-1913 (compiled for the years indicated from Statistik des Deutschen Reichs: Die Krankenversicherung):