The Shattuck Lecture*
Epidemiology and Etiology of Influenza

By Allan J. McLaughlin, M.D., Washington, D.C.,
Assistant Surgeon-General, U.S.Public Health Service

Dictionary of Selected Influenza-Related Medical Terms
(opens in separate window, so it can be read along with the text)

The paucity of facts, especially of etiological facts, might stamp influenza as a unwise selection as a subject for this lecture. It is true that doctors prefer dogma to conjecture, and definite procedure to advice on general principles, but few dogmatic statements can be made concerning etiology, prophylaxis or treatment of influenza.

While it cannot be proven there is every reason for believing that influenza is a disease of great antiquity. From a period of four centuries before the Christian era accounts are found of a plague or epidemic which may have been influenza. The description of these earlier epidemics is too vague to warrant any positive conclusion.

From the sixth to the tenth century numerous epidemics are recorded with history of cough and catarrhal symptoms which are more suggestive of influenza. Hirsch1 in his tabulation of influenza epidemics or pandemics excludes those prior to 1173 as too indefinite and uncertain to serve any useful purpose in his compilation.

There is not only uncertainty as to the nature of the disease but obviously the tabulation must be incomplete, as many epidemics occurred of which we have no record. Thus, many of the intervals shown in Hirsch's table may not have been quiescent periods but simply evidenced failure to record epidemics. No safe deductions as to cycles or periodicity can be made in these years from the 12th to the 18tyh century. Noah Webster2 describes the first American epidemic in 1647 and compiles a table of epidemics from 1174 to 1797. Webster coupled the epidemics in each instance with a volcanic eruption, earthquake or some unusual climatic condition.

Webster made many interesting comments on the epidemiology of influenza. He noted that the epidemics were sometimes limited to the American continent and that contrary to usual custom in certain pandemic years, the disease beginning in America spread to the entire world. In 1698, 1757, 1761 and 1781, "it spread over the American hemisphere one year prior to its pervading the other hemisphere." In regard to incomplete records, Webster says: "I regret my want of materials to complete


* Delivered before the Massachusetts Medical Society, June 8, 1920.

--1--

a view of this subject. No regular register has been kept in America, of the seasons, diseases and phenomena, from the first settlement, and whether any notices of all the catarrhs in this country are in existence, I do no know. I have found no accounts of any, between 1655 and 1698--nor between the latter year and 1733. One of these instances, that in 1698, came to my knowledge by accident, as I have mentioned under that year, in the foregoing history. From the uniform appearance of this epidemic as often as once in ten or twelve years, in other periods, we have ground to believe, it has always occurred in nearly the same periods."

As to the date of the first authentic epidemic of influenza there is much difference of opinion. Hirsch places it as 1173; Webster 1174; Zerviani 1239; Gluge 1323; Schmeich, Haeser and others 1837; Thompson, Zulzer and Seifert 1510. Since 1510 detailed descriptions are more often available and little doubt exists as to the identity of the first real pandemic in 1580 and the pandemic in the Western Hemisphere 1647, described by Webster. In the 18th century besides rather widespread epidemics in 1709-12, there were decided pandemics in 1729, 1732, 1742, 17757-8, 1761-2, 1767, 1781-2, 1788-90 and 1799.

In the 19th century because of a greater availability of detailed information we are on firmer ground and we begin to find clear pictures of pandemics very similar to those known to the present generation.

In 1824-25-26 influenza was widely epidemic in the Western Hemisphere, and in 1827 was generally diffused in Siberia and Eastern Russia.

According to Hirsch, after three years quiescence in 1830-31 one of the really great pandemics spread over the entire world. Its course was chronologically China, Philippines, Polynesia, Borneo, Sumatra, Russia, Baltic Provinces, Poland, Germany, Austria, Finland, denmark, Belgium, France, Sweden, England, Scotland, Switzerland, Italy, Spain, and, in January 1832, North America.

After a quiescent period of one year this pandemic was repeated. Beginning in Russia in January 1833, it followed a course from east to west almost identical with that of 1830-31.

Three years later, in the autumn of 1836, a pandemic spread over Australia, South Africa and the East Indies--and beginning in Russia in December spread all over Europe and was reported in Mexico in July 1837. In 1841-2 influenza was epidemic in Germany, Austria-Hungary, Ireland, belgium, England and France. in 1843 there was a widespread epidemic in the United States and in 1844 Germany, France, and Switzerland and Russia were stricken in the order named. This seemed to be a reversal of the usual geographic progression from east to west.

Apparently, according to Hirsch, there was pandemic prevalence in 1846-8, 1850-1, 1857-8, 1874-5. The same writer states that pandemics occurred exclusively in the Western Hemisphere in 1843 and 1873.

It is true that we have more accurate knowledge of influenza during the 19th century than at any previous period, nevertheless, for practical purposes the epidemiologist must not begin his study earlier than the great pandemic of 1889-90. It is not possible to compare recent epidemics with those occurring prior to 1889 except in the most general way.

In every phase of the subject the student finds not definite precise information but rather confusion and conflict in statement and in laboratory findings as well.

It is discouraging to the average reader and especially so to the health officer who needs and seeks instruments and agencies for combating the disease. However discouraging our present lack of knowledge, the magnitude of the problem and the certainty of its return warrant the closest study of such facts as are known, and a redoubled effort in research to clear up the etiologic points now in dispute.

I have selected the subject of influenza chiefly because of its tremendous potentiality for disaster on a large scale and the certainty that future visitations may be expected. It seems wise to review such facts and observations as we possess in the hope that an exposé of our utter helplessness may stimulate further research and give us weapons with which to fight this scourge.

As a demonstration of what did happen and of what may be expected to happen again we need consider only the appalling disaster of 1918.

Never in the history of influenza has such a death toll been exacted. It is probable that in the whole history of the world no parallel will be found for the tremendous catastrophe

--2--

of 1918, if we consider the short space of time and the wide area of distribution in which the results were manifested. Statistics can never be accurate in such times of stress and world-wide estimates are notoriously inaccurate, but the data indicate that in four months a half million lives were sacrificed in the United States and that in the entire world this particular pandemic was responsible for not less than six million deaths. [Estimates in 2005 increase the world-wide death toll to 20-50 million; far exceeding the butcher's bill for all of the First World War.--HyperWar]

At best our statistical data are far from ideal. We would prefer to have morbidity to mortality statistics, but they are so incomplete that they are practically useless. Many objections can be made against the use of mortality statistics but the fact remains that with proper corrections they afford us the best indices available for large areas.

The use of influenza mortality statistics alone is misleading because of the fact that the majority of deaths are not recorded as influenza. The combination of influenza and pneumonia deaths makes a much better index of the extent of damage done by influenza, provided we express this in terms of excess death rates, that is, the rate over and above the normal expectancy of mortality for the given period.

Was the Influenza Pandemic of September-October, 1918, Due to the Same Cause as Previous Epidemics?

That the influenza of 1918 is the same disease as described by authors in ancient and medieval times is incapable of proof although it is probable from the meager and indefinite description that many of the outbreaks so described were influenza. As to the identity of the 1918 outbreak with that of 1889-90 there is more definite evidence and the similarity, epidemiologically and clinically, is so striking that we must conclude that they were due to the same cause.

The epidemiologic similarity in the two pandemics is well shown by Frost,4 and he considers that in the history of influenza we have cycles in which great pandemics alternate with periods of relative quiescence, the length of the interval between pandemics being usually a matter of decades.

Usually preceding a real world-wide pandemic there are prepandemic increases in prevalence, amounting to considerable epidemics which have often passed unnoticed except in retrospect. The significance of these pre-pandemic waves is not clear although there is a natural tendency to connect them with the great pandemic rise which follows.

The special characteristics of the great pandemics are rapid spread, wide area of distribution and definite geographical progression along the most widely used routes of travel and trade. Such characteristics undoubtedly marked many of the historical epidemics tabulated by Hirsch and others, but are especially noteworthy in the pandemics of 1889-90 and 1918. These striking characteristics are lacking in the epidemics occurring in the interpandemic periods.

After a gr eat pandemic such as '89-90 or 1918 there has been apparently such a thorough seeding of the population with the microbic cause that for years after outbreaks may occur anywhere whose spread is limited, and if the distribution is wide this seems more the development from many foci rather than an orderly geographic progression from a single source.

These post-pandemic outbreaks in succeeding years become more local and sporadic and bearing progressively less resemblance to the mode, rapidity and scope of spread of the real pandemic outbreaks.

On clinical grounds the similarity of the two great pandemics is no less probable. Kinsella5 very positively states:

"Without doubt, the epidemic just past is of essentially the same nature as the epidemic of 1889. Then, as in the epidemic under discussion, clinicians recognized two phases or features in the disease: first, the 'influenza' proper, which lasted from three to seven days, and the complications which are regarded as secondary and not part of the original disease, though directly due to the damages that it caused. The descriptions of the disease, in England by Robertson and Elkins, in Germany by Strumpell, and in France by Duflocq, make these points clear."

Conceding the identity of the great pandemics 1889-90 and September-October 1918, there is still doubt expressed by some as to their relationship to the relatively mild inter-pandemic outbreaks.

The catastrophe of September-October 1918 is so sharply differentiated from ordinary epidemics of influenza that some have suggested that it is due to an entirely different cause. They have not gone so far as to assume that

--3--

this etiologic factor was not responsible for some previous epidemics of so-called influenza but are slow to believe that the etiologic factor responsible for the comparatively mild outbreaks is the same as that which caused the pandemic of September-October 1918. Kinsella says:

"Perhaps the simplest and most obvious inference that can be drawn from a consideration of the epidemic of influenza that has just passed through this country is that it was something unusual, something that had not been seen for many years, and something that has departed leaving few representative cases that can ever be regarded as typical instances of the disease. Whatever the cause of the disease may be, it is clear that this agent is one to which the body is not accustomed. In fact, it would almost seem necessary to postulate at the outset that this agent is not commonly present in the body under normal circumstances, because it is difficult to conceive that any bacterium or virus that is even partially adapted to the environment of the human body should suddenly become unadapted and assume such a high degree of invasiveness as the agent of influenza possesses. Moreover, if the cause of influenza is some bacterium or virus related to a variety of bacterium or virus commonly present in normal individuals, then it would seem necessary to postulate that the cause of influenza is a very highly individualized variety."

Other writers have found difficulty in believing that the organism causing influenza in interpandemic periods, such as the mild epidemics of March, April, 1918, could so completely change as to become the cause of the frightful mortality five months later. These observers have been prone to ascribe the pandemic to virulent strains introduced from abroad.

Soper6 inclines to the view that the organism or virus was introduced from abroad. He states that the disease was reputed to be epidemic in Spain in the early spring but that it was known to be present in the United States (Fort Oglethorpe) in March 1918. In the latter part of March the disease appeared in the A.E.F., the French and British armies and the civil population. Many patients with influenza arrived in the United States upon ships from Europe in June, July and August, 1918. He cites many ships carrying influenza to our ports and concludes:

"The patients from the vessels were sent ashore and soon mingled with the civilian populations. There were thus scattered rather widely along the Atlantic seaboard sparks from which the pandemic not improbably arose."

Winslow and Rogers7 suggest the same origin. From information furnished by Dr. T.E. Reeks, Connecticut State Department of Health, the following statement is made:

"Influenza first appeared as an epidemic in Connecticut in New London, in the eastern part of the State, on or about September 1, 1918, when several cases of the disease were reported by the naval hospital at New London. These cases came primarily from the experimental Station and Fort Trumbull, where vessels from foreign ports had discharged patients."

On the other hand, we must concede the possibility of an increased virulence of the strains of organisms responsible for the relatively mild prepandemic prevalence of March-April 1918. Frost lays stress on the significant rise in general prevalence and the many definite local outbreaks in the spring of 1918 and concludes:

"The rise in mortality from this group of etiologically heterogeneous diseases in the spring of 1918 is so sudden, so marked, and so general throughout the United States as to point very clearly to the operation of a single definite and specific cause, something largely independent of meteorologic and other local conditions. The observed occurrence of local epidemics of influenza at that time in widely scattered localities, the intimate association established at Camp Funston between the epidemic of influenza and pneumonia, and the subsequent development of the influenza pandemic, all indicate that the increased pneumonia mortality of March and April 1918, was the consequence of a beginning and largely unnoticed epidemic of influenza, and beginning, in this country, of the great pandemic which developed in the autumn."

General Epidemiologic Characteristics of Influenza

Prevalence Since 1889. For intensive study there is available for certain localities fairly complete data from 1889 to date. For Massachusetts we have the deaths from influenza and pneumonia by months from 1887 to date. This covers what may be termed one completed cycle or the period from one great pandemic to another,

--4--

and it affords a basis of comparison between these two great pandemics.

Chart I shows the excess annual death rates per 100,000 for influenza and pneumonia by months for Massachusetts from 1887 to 1919. After a quiescent period of several decades the pandemic beginning December, 1889, is manifest on the chart in three successive waves with highest point as follows: January, 1890; May, 1891, and January, 1892. That of January, 1890, being very high, and that of May, 1891, considerably lower, and that of January, 1892, highest of all. As an aftermath of the great pandemic period 1889-1892, the chart shows a significant rise in epidemic prevalence in some month from January to April every year except 1898, 1902 and 1904, or in twelve of the fifteen years from 1890 to 1904. In 1905 and 1908 sharp rises are shown and an excessive prevalence from November, 1910, to February, 1911. From 1912 to 1914 the rates were probably as near normal as can be expected in the winter months. A noticeable rise took place March-April, 1915, with a higher peak in January, 1916, and lower peaks in March, 1917, and April, 1918, as if beginning a new cycle.

Chart II shows the annual excess death rates per 100,000 for forty-two large cities with an aggregate population of about 22 million by months from January, 1915, to August, 1918, inclusive. This chart very clearly shows the prepandemic rises in prevalence occurring each year in the winter and spring months, that culminating in the peak of April, 1918, being especially noteworthy.

There was then not only a regular prevalence in the winter and spring months every year but there was an epidemic prevalence in twenty-two out of thirty years and the longest period without epidemic prevalence was from February, 1911, to March, 1915, a period of four years. Further, beginning March, 1915, there has been an annual epidemic prevalence in the winter or spring months in 1915, 1916, 1917 and 1918, followed by the great pandemic of September-October 1918.

Periodicity

Early writers asserted that there was a definite periodicity in epidemic appearances of influenza. Brownlee8 deduced from a study of influenza mortality that epidemics occurred in 33-week cycles, and he forecasted an epidemic for January or february, 1920.

Stallybrass,9 using statistics for Liverpool since 1890, confirms Brownlee's deductions as to the 33-week cycle.

Speare10 considers mortality statistics unsuitable, because of their great fluctuation, for analysis by the periodogram method employed by Brownlee and says that the 33-week cycle has no existence in fact.

Speare divided the years into 13 periods of four weeks each and tabulated the frequency with which the observed week of maximum mortality fell within one or the other of the thirteen groups.

by this method he shows that the bulk of the maxima fall within the months of February and March. He explains Dr. Brownlee's conclusions by the fact that there is such great variation in the amplitude of the waves. A single epidemic such as October, 1918, would overshadow and practically eliminate many smaller epidemics when analyzed by the periodogram method. He concludes that the average interval is about one year.

Table I.
Annual Death Rate per 100,000 and Variation from Median Rate for Influenza and Pneumonia (All Forms) for 42 Large Cities Included in the Weekly Health Index of the U.S. Census Bureau.
  Jan. Feb. Mar. Apr. May June July Aug. Sept. Oct. Nov. Dec.
Annual Death-Rate per 100,000
1915 231 264 296 284 151 114 76 67 77 106 153 354
1916 446 266 258 213 170 102 80 77 87 112 162 246
1917 389 346 303 246 204 115 75 69 91 129 178 235
1918 315 312 402 421 187 96 79 56        
Median Rate (per 100,000) for Period 1910-1916
  279 280 290 234 177 109 82 73 82 111 163 228
Variation of Annual Rate from Median Rate
1915 -48 -16 6 50 -26 5 -6 -6 -5 -5 -10 126
1916 167 -14 -32 -21 -7 -7 -2 4 5 1 -1 18
1917 110 66 13 12 27 6 -7 -4 9 18 15 7
1918 36 32 112 187 10 -13 -3 -17        

--5--

Chart I: Massachusetts Excess Death Rates from Influenza and Pneumonia,
 by Months, 1887 to 1918
Chart I.
Massachusetts Excess Death Rates from Influenza and Pneumonia, by Months, 1887 to 1918.

Seasonal

While no claim for annual periodicity is made, a glance at Chart I reveals the remarkable predilection which influenza has for the winter and spring months.

Every significant rise on Chart I from 1887 to 1918 occurred in the months from December to May. The pandemic of September-October 1918, was the great and only exception to this rule.

Mortality Statistics According to Age and Sex

In the pandemic of 1889-90 Abbott11 states that the deaths from pneumonia and acute bronchitis for the first two weeks in 1889 and 1890 in Paris were as follows:

             First Two Weeks
Ages   1889 1890
0-4   45 111
5-19   6 22
20-39   7 127
40-59   14 249
over 60   35 290

This table shows the excess of deaths to be chiefly in the age groups from 20 years upward. He presents other tables which show that in Paris the mortality in male adults was double that of female adults.

Parsons12 shows that the deaths from influenza in London for the epidemic of 1847-8 and for the first quarter of 1890 occurred in the different age groups in the following percentages:

             Percentage
Ages   1847-8 1890
under 1   10.5 5.2
1-5   13.1 4.3
5-20   3.8 4.7
20-40   8.6 24.7
40-60   18.5 36.2
60-80   16.9 22.4
above 80   8.6 2.5

Winslow and Rogers13 analyzing statistics for Connecticut for the last four months of each of the years 1917-1918, show that the normal distribution of deaths in 1917 was 25% of all deaths occurred under 5 years; 25% occurred between 5 and 40 and the remaining 50% over 40. In 1918 the distribution was strikingly different; instead of 25% between the ages of 5 and 40, this period included 49% of all deaths and 40% of all deaths occurred in the two decades between 20 to 40, as against only 17% in 1917. Considering influenza and pneumonia alone, the age period 20 to 40 included

--6--

56% of the deaths, while only 9% occurred in ages over 49. The decade 20-29 was most severely affected, including 30% of all deaths.

The high rates of influenza mortality in certain age groups may be either due to an excessive incidence of influenza, or to a very high case fatality rate within these age groups. It is therefore necessary to study the limited morbidity survey statistics available.

Chart II.
Monthly Variation in the Death Rate from Influenza and Pneumonia from that of Corresponding Month in Median Year in 42 Large Cities Included in Weekly Health Index of U.S. Census Bureau.

Chart II: Monthly Variation in the Death Rate from Influenza and
  Pneumonia from that of Corresponding Month in Median
 Year in 42 Large Cities Included in Weekly Health Inndex
 of U.S. Census Bureau

Chart III.
Case Rate per 1000 Persons from Influenza and Pneumonia (All Forms).

Chart III: Case Rate per 1000 Persons from Influenza and Pneumonia (All Forms)

Chart IV.
Influenza Morbidity Incidence of Cases per 1,000 Males and Females of Each Specified Age, In All Localities Canvassed.

Chart IV: Influenza Morbidity Incidence of Cases per 1,000 Males and Females of
 Each Specified Age, In All Localities Canvassed

Chart V.
Influenza--Case Fatality (per cent) in Males and Females of Each Specified Age Group In All Localities Canvassed.

Chart V: Influenza--Case Fatality (per cent) in Males and Females of Each
 Specified Age Group In All Localities Canvassed

--7--

Chart VI.
Chart VI: Pneumonia Incidence per Thousand Males and Females in
 Specified Age Groups, Ratio of Deaths to Pneumonia Cases in Males and Females
 of Specified Age Groups

Morbidity Statistics As Showing General Attack Rate and Attack Rate According to Sex and Age

Frost14 analyzed the results of a morbidity survey of 10 cities from 25,000 to 600,000 and in several smaller cities and rural areas in Maryland. His assistants canvassed at least 5,000 persons in the smaller cities and over 5% of the population of large cities. His work has great value as a check on the deductions made from mortality statistics alone, and confirms the approximate accuracy of many of these deductions.

Frost found that the attack rate varied from 150 to 405 per 1,000 population and the average for all communities was 280 per thousand.

He found the attack rate was highest in the age group 5 to 9 (Chart III), declining in each successive higher age group, except 25 to 34, which exceeded the rates for the ages 15 to 24. With few exceptions he found the attack rate at all ages higher in females than males (Chart IV), He says:

"The most striking excess of incidence in females occurs between the ages of 25 to 40, the difference between the sexes being relatively slight in periods above and below these limits. These facts indicate that females over the age of 15, especially between the ages of 15 to 45, were either more susceptible to infection or more generally and more intimately exposed than males of corresponding age."

Case Fatality Rate

Frost found that the ratio of deaths to total cases of influenza varied form 3.1% in New London to 0.8% in San Antonio, the variations showing no consistent relation to incidence rates. He found that the case fatality rates were higher in the Northern Atlantic and Pacific coasts and lowest in Central and Southern cities.

In regard to sex and age he found remarkable differences. Chart V shows the case fatality under 15 was somewhat higher in females and over 60 very much higher in females; while form 15 to 60 there was a much higher case fatality rate in males. The lower case fatality

--8--

rate in females from 15 to 60 seems to be explained in part at least by the lesser incidence of pneumonia in this group (Chart VI).

Winslow and rogers, from the reports of the New Haven Visiting Nurses' Association, conclude that the case fatality rate of 4.3 per 100 cases (736 cases--32 deaths) is too high since undoubtedly many light cases failed to receive nursing care. Winslow and rogers also analyzed information supplied by Reeks from a survey in New Britain, Connecticut, and says these figures would indicate a morbidity rate of 234 per 1,000 and a fatality rate of 3.9 deaths per 100 cases. Here again they note that light cases were incompletely reported and that this rate is probably too high. They conclude that the attack rate in Connecticut was from 200 to 400 per 1,000 and the case fatality rate from two to four deaths per 100 cases, the higher morbidity and lower fatality rate being more likely to be correct (Chart VII).

Chart VII.
Influenza Incidence of Cases per 1,000 Persons
Influenza-Pneumonia: Death Rate per 1,000 Persons
Influenza Fatality per 1000 Cases

Chart VII: Influenza Incidence of Cases per 1,000 Persons;
 Influenza-Pneumonia: Death Rate per 1,000 Persons;
 Influenza Fatality per 1000 Cases

Race

Winslow and Rogers made one striking observation in regard to race. They found that native Americans of Irish, English, or German descent had a relatively low mortality rate, while Slavic or Latin peoples had a very high mortality rate. They admit that the age distribution and economic environment account for some of the excess, but believe that these factors cannot explain the enormous differences which are shown by the Italian population.

Density of Population

Table II shows the total annual excess death rate per 100,000 influenza and pneumonia, and

--9--

Chart VIII.
Excess Death Rate Epidemic Period, September, 1918-June 28, 1919.

Chart VIII: Excess Death Rate Epidemic Period, September, 1918-June 28, 1919

Chart IX.
Values for Index of Explosiveness.

Chart IX: Values for Index of Explosiveness

--10--

Chart X.
Relative Figures--Death Rates.

Chart X: Relative Figures--Death Rates

Chart XI.
Relative Figures--Death Rates.

Chart XI: Values for Index of Explosiveness

--11--

the density of population in persons per acre in 40 cities. Chart VII shows graphically the lack of correlation between the curves of these two factors.

The group of cities with low density of population included some of the highest rates (New Orleans, Kansas City and Worcester). Conversely some cities with a high density of population had relatively low excess death rates (Milwaukee, Chicago and st. Louis).

Table II
The Density of Population in Persons per Acre with Excess Deaths, Influenza and Pneumonia, all Forms, per 100,000, for Epidemic Period September 8, 1918, to June 28, 1919
City Density
Persons
per Acre
Excess
Death Rate
1918-1919
Los Angeles 2.4 505
New Orleans 3.0 761
Birmingham 5.7 497
Fall River 6.0 602
Oakland 6.4 528
Worcester 7.0 618
St. Paul 7.4 386
Kansas City 8.0 689
Omaha 8.3 562
Albany 8.9 631
Cincinnati 9.1 531
Washington 9.6 680
Nashville 10.1 662
Richmond 10.8 520
Toledo 10.9 329
Indianapolis 11.0 354
Minneapolis 11.3 299
Atlanta 14.4 330
Memphis 12.1 470
Dayton 12.6 532
New Haven 13.1 629
Syracuse 13.3 580
Lowell 13.6 548
Columbus 15.2 409
Louisville 16.6 530
San Francisco 17.6 724
Rochester 18.6 433
Buffalo 19.0 567
St. Louis 19.4 406
Cleveland 20.1 546
Chicago 20.3 404
Philadelphia 21.0 814
Providence 22.4 534
Pittsburgh 22.8 891
Milwaukee 26.9 337
Boston 27.4 685
Newark 27.5 538
Cambridge 28.2 532
New York 29.5 544
Baltimore 30.6 629

Density of population is not always a reliable index of economic and sanitary conditions, although usually the most densely crowded cities have the worst housing conditions. Frost15 attempted to find if any relation existed between overcrowding and poverty and a high attack rate. He found from a study of Little rock, Arkansas, that the attack rate increased as the number of rooms per person decreased. He says that these limited statistics suggest that domestic environment is a factor of some importance in the influenza attack rate, the tendency being toward a higher morbidity under the complex of conditions associated with relative poverty.

Variation in Explosiveness and Excess Death Rates in Forty Large Cities

A consideration of the epidemic beginning September, 1918, brings out the very remarkable variation in the various cities in degree of explosiveness of the epidemic and the total excess death rate for the period.

Explosiveness

Great variation in explosiveness in individual cities was apparent in the 1918 epidemic.

Table III shows the explosiveness of the epidemic

Table III
Index of Explosiveness of Epidemics
City 1918-19 1920
Albany 285 33
Atlanta 62 28
Baltimore 347 58
Birmingham 112 24
Boston 264 35
Buffalo 197 27
Cambridge 162 35
Chicago 115 63
Cincinnati 85 19
Cleveland 107 49
Columbus 87 84
Dayton 268 54
Fall River 270 28
Indianapolis 74 67
Kansas City 80 84
Los Angeles 86 32
Louisville 189 29
Lowell 221 36
Memphis 201 92
Milwaukee 18 64
Minneapolis 33 103
Nashville 416 76
Newark 128 58
New Haven 101 38
New Orleans 305 22
New York 127 57
Oakland 189 27
Omaha 227 38
Philadelphia 451 31
Pittsburgh 168 110
Providence 124 50
Richmond 314 29
Rochester 102 27
St. Louis 29 80
St. Paul 44 49
San Francisco 264 27
Syracuse 202 53
Toledo 132 43
Washington 266 104
Worcester 227 24
All cities considered as a single population group 78 33
Note.--The index of explosiveness has been computed for each city as follows: the death rate from influenza and pneumonia (all forms) in the peak week (i.e., week in which the maximum mortality occurred) has been divided by the number of weeks from the beginning of the epidemic up to and including the peak week. It must be remembered that the result is only approximate.

--12--

in both the 1918 and 1920 epidemics in forty cities, using as an index the death rate for the peak week divided by the number of weeks from the beginning of the epidemic up to and including the peak week. This gives an arbitrary figure which will serve as an index perhaps quite as well as that secured by more complicated methods.

Chart IX shows the great variation in explosiveness in graphic form. The cities are arranged according to the degree of explosiveness. In the 1918 epidemic eleven cites chowed a very high degree of explosiveness (451 to 264) measured by this index, viz., Philadelphia, Nashville, Baltimore, Richmond, New Orleans, Albany, Fall River, Dayton, Washington, Boston and San Francisco. Next with high rates (227 to 189) come Worcester, Omaha, Lowell, Syracuse, Memphis, Buffalo, Louisville, and Oakland. Next comes a group with moderate explosiveness (168 to 101), Pittsburgh, Cambridge, Toledo, Newark, New York, Providence, Chicago, Birmingham, Cleveland, Rochester and New Haven.

The last ten cities may be said to have had a low index of explosiveness (87 to 18). These cities were Columbus, Los Angeles, Cincinnati, Kansas City, Indianapolis, Atlanta, st. Paul, Minneapolis, St. Louis and Milwaukee.

Geographically the distribution of the eleven highest cities was wide and included cities in New England, Middle Atlantic, Gulf and Pacific coast cities, with only Nashville and Daton in the Middle West. The distribution in the next group of cities with high indices including New England, New York State, Oakland, California, and two cities in the Ohio or Mississippi Basin--Louisville and Memphis. The moderate group had Pittsburgh, Cleveland, Toledo and Chicago in the Middle West, Cambridge, New Haven, Providence, Newark, Rochester and New York in the East with Birmingham in the South. The ten cities with a low explosive index were distributed as follows: Los Angeles on the Pacific Coast, Atlanta in the South and Columbus, Cincinnati, Indianapolis and St. Louis in the Middle West with farther north St. Paul, Minneapolis and Milwaukee.

In general the 1920 index of explosiveness did not show such extremes of variation. There were, however, sharp rises above the average for several cities. The index of explosiveness was relatively high in 1920 in Nashville, Memphis, Washington, Pittsburgh, Columbus, Kansas City and Minneapolis. Of these Nashville, Washington and Memphis had a high explosive index in both epidemics. Pittsburgh was moderate in 1918 and high in 1920. Columbus, Kansas City and Minneapolis had a low index in 1918 and a high index in 1920.

What relation did the explosiveness bear to the actual excess death rate for the epidemic period? Of the eleven cities with highest indices for explosiveness, eight were also among the eleven cities with the highest actual excess death rates for the epidemic period. These cities were Philadelphia, Nashville, Baltimore, New Orleans, Albany, Washington, Boston and San francisco. Thirteen cities had an actual excess death rate for the 1918 epidemic period of 600 or over, ten of these also had a very high explosive index. The three exceptions were Pittsburgh with a moderate explosive index and the highest mortality, Kansas City and New Haven with low explosive index and a very high total mortality.

Ten cities had a low explosive index of less than 100, viz., Columbus, Los Angeles, Cincinnati, Kansas City, Indianapolis, Atlanta, St. Paul, Minneapolis, St. Louis, Milwaukee.

All of these had relatively low excess mortality rates for the epidemic period except Los Angeles and Cincinnati with moderate rates and Kansas City with a very high excess mortality rate.

In spite of the notable exceptions there seems to have been some correlation between the degree of explosiveness and the amount of mortality for the whole epidemic period.

What are the reasons for the great variation in explosiveness shown by the forty large cities?

Pearl16 in an excellent monograph analyzes geographical position, density of population, differences in distribution of age groups, and great increase or decrease in population in relation to explosiveness. In regard to these factors, which have not been without effect in other diseases, he says:

"The general conclusion to which we come from an examination of the correlation data assembled to this point is that these four general demographic factors,--density of population, geographical position, age distribution of population and rate of recent growth in population have practically nothing to do, either severally or collectively with bringing about those differences

--13--

between the several cities in respect to explosiveness of the outbreak of epidemic mortality in which we are interested."

Pearl believes that the most significant factor in causing the variation to be the relative normal liability of the inhabitants to die of one of the three great causes of death,--disease of the lungs, heart, or kidneys. He says:

"Such an analysis, by the method of multiple correlation, appears to demonstrate that an important factor so far found in causing the observed wide variation amongst these 39 American cities in respect of the explosiveness of the outbreak of epidemic influenza mortality in the autumn of 1918 was the magnitude of the normal death rates observed in the same communities, particularly those death rates from pulmonary tuberculosis, diseases of the heart and of the kidneys."

Table IV shows the fifteen cities with the highest explosive index in September-October 1918,--and the combined death rate for pneumonia, tuberculosis, heart disease, and nephritis--eleven of the fifteen cities have a very high combined death rate for these four diseases (average for 1911 to 1917 inclusive). The exceptions are Dayton, Lowell, Syracuse, and Omaha. Dayton, Lowell and syracuse had average rates for pneumonia and heart diseases, but Omaha had low rates for all four diseases.

Table IV.
City Explosive
Index
Combined Death
Rate per 100,000
Pneumonia, Tuberculosis,
Heart Disease
and Nephritis
Philadelphia 451   726  
Nashville 416   723  
Baltimore 347   802  
Richmond 314   790  
New Orleans 305   919  
Albany 285   835  
Fall River 270   635  
Dayton 268   569  
Washington 266   765  
Boston 264   684  
San Francisco 264   688  
Worcester 227   641  
Omaha 227   469  
Lowell 221   550  
Syracuse 202   545  

Of the four principal factors in our total mortality, pneumonia and heart disease are the ones most likely to affect explosiveness deduced from mortality statistics. If the combined death rate per 100,000 (average for 1911 to 1917 inclusive) for pneumonia and heart disease was more than 350 the explosiveness was marked.

Table V.
City Combined Death
Rate per 100,000
Pneumonia and
Heart Disease
Explosive
Index
Boston   412   264
Worcester   412   227
Pittsburgh   403   168
Baltimore   393   347
New Haven   384   101
New Orleans   381   305
Nashville   280   416
Washington   276   266
Fall River   375   270
Richmond   374   314
Philadelphia   367   451
San Francisco   362   264

The twelve cities with the highest combined rates for pneumonia and heart disease (Table V) showed a very high explosive index in 1918. There are two exceptions,--Pittsburgh and New Haven; of these Pittsburgh had the highest explosive index for the 1920 epidemic, but New Haven showed a low degree of explosiveness in both epidemic periods.

Variation in Severity Measures by Excess Death Rates

What are the reasons for the very great variation in the severity of the epidemic as measured by mortality in certain cities? Why did Pittsburgh have a very high rate in all three epidemic periods and why did cities like Milwaukee, St. Paul and Minneapolis have low rates in all periods?

Table VI shows in the first column what may be considered a rough index of the total damage in the forty large cities from influenza and pneumonia combined. The figure given for each city is the sum of the excess death rates for the three epidemic periods from January, 1917, to March, 1920. In the other columns are given the general death rate per 1,000 and the death rate per 100,000 for pneumonia, tuberculosis, heart disease and nephritis.

Considered as a whole there seems to be some correlation between the loss from influenza and the death rates from certain diseases. Considering the cities in three groups this correlation is even more striking.

  Total Excess
Death Rate
Group 1 above 800
Group 2 600 to 800
Group 3 below 600

--14--

Table VI.
City Total Excess
Death Rate
per 100,000 for
3 Epidemic
Periods, Jan.,
1917, to June,
1920
General Death
Rate per 1,000
Average for
1914-1917,
Inclusive
Death Rate per 100,000 Average
for 1911 to 1917, Inclusive
Pneu-
monia
Tuber-
culosis
Heart
Disease
Neph-
ritis
Albany, N.Y.   728     19.4   157.5 254.5 231.3 191.2
Atlanta, Ga.   533     15.7   167.3 148.0 112.6 176.4
Baltimore, Md.   810     18.1   210.3 233.3 182.9 175.3
Birmingham, Ala.   855     16.8   171.0 217.5 115.5 105.2
Boston, Mass.   822     16.4   203.3 173.9 208.5 98.0
Buffalo, N.Y.   722     15.6   158.2 156.6 165.3 118.6
Cambridge, Mass.   659     13.4   155.5 192.8 170.9 76.0
Chicago, Ill.   524     14.4   188.0 162.22 150.4 110.2
Cincinnati, Ohio   650     16.1   142.4 242.5 200.5 157.2
Cleveland, Ohio   740     14.1   139.9 144.1 117.8 92.7
Columbus, Ohio   598     14.9   130.0 116.8 151.4 85.6
Dayton, Ohio   690     14.6   132.7 157.5 176.4 102.3
Fall River, Mass.   673     16.6   219.6 157.4 155.1 103.0
Indianapolis, Ind.   546     15.6   133.9 199.4 177.3 106.3
Kansas City, Mo.   1043     14.9   137.6 163.7 138.6 121.3
Los Angeles, Calif.   559     12.5   90.8 232.4 161.4 104.3
Louisville, Ky.   701     15.7   145.9 198.5 158.0 150.8
Lowell, Mass.   736     16.1   161.4 139.3 156.2 92.7
Memphis, Tenn.   684     20.4   170.7 270.6 136.2 174.4
Milwaukee, Wis.   537     12.2   133.2 100.8 97.9 74.1
Minneapolis, Minn.   437     11.9   109.8 140.6 107.4 93.6
Nashville, Tenn.   933     17.3   172.4 229.3 207.9 112.9
Newark, N.J.   702     14.2   156.5 178.9 140.3 135.5
New Haven, Conn.   765     16.5   210.2 140.9 173.4 130.0
New Orleans, La.   854     20.0   159.2 283.9 220.7 255.1
New York, N.Y.   708     13.9   191.2 190.5 152.8 135.0
Oakland, Calif.   666     11.0   97.8 126.8 189.3 87.5
Omaha, Neb.   743     13.3   155.3 111.2 111.5 90.5
Philadelphia, Pa.   1087     16.2   166.2 195.1 191.5 173.0
Pittsburgh, Pa.   1305     16.6   275.1 135.4 127.7 86.8
Providence, R.I.   680     15.2   168.7 158.1 159.1 130.5
Richmond, Va.   608     19.2   179.4 220.9 194.4 195.1
Rochester, N.Y.   534     14.5   135.3 118.0 183.3 133.4
St. Louis, Mo.   629     14.7   158.1 154.8 137.7 167.7
St. Paul, Minn.   510     11.0   92.8 132.6 105.8 80.0
San Francisco, Calif.   874     15.5   128.3 198.6 234.1 126.6
Syracuse, N.Y.   787     14.6   141.9 114.7 178.5 110.0
Toledo, Ohio   431     16.6   123.3 189.2 158.9 91.7
Washington, D.C.   885     17.6   152.8 221.2 225.3 165.3
Worcester, Mass.   754     164.   187.5 147.4 224.3 82.3

Group 1:--the cities having the highest excess death rates combined for the three epidemic periods (January, 1917, to March, 1920) also had very high rates of pneumonia, tuberculosis, heart disease and nephritis and a high general death rate. There was one exception--Kansas City.

Table VII
City General
Death Rate
per 100,000
Total Excess
Death Rate,
Influenza,
Pneumonia
for 3 Epidemic
Periods--
Jan. 1917, to
March, 1920
Death Rate per 100,000,
1911-1917, Inc.
Pneu-
monia
  Pneumonia,
Tuberculosis,
Heart Disease
and Nephritis
 
Pittsburgh       16.6             1305       275.0       625.0      
Philadelphia   16.2     1087   166.2   725.8  
Kansas City   14.9     1043   137.6   561.2  
Nashville   17.3     933   172.4   722.5  
Washington   17.6     885   152.8   764.6  
San Francisco   15.5     874   128.3   687.6  
Birmingham   16.8     855   152.8   764.6  
New Orleans   20.0     854   159.2   918.9  
Boston   16.4     822   203.3   683.7  
Baltimore   18.1     810   210.3   801.8  
Average for
registration cities
        717   158.6   593.6  

Table VII shows the rates for these cities compared with the average in all registration cities.

Table VII
City General
Death Rate
per 100,000
Total Excess
Death Rate,
Influenza,
Pneumonia
for 3 Epidemic
Periods--
Jan. 1917, to
March, 1920
Death Rate per 100,000,
1911-1917, Inc.
Pneu-
monia
  Pneumonia,
Tuberculosis,
Heart Disease
and Nephritis
 
Syracuse       14.6             787       141.9       545.1      
New Haven   16.5     765   210.2   654.5  
Worcester   16.4     754   187.5   641.5  
Omaha   13.3     743   155.3   468.5  
Cleveland   14.1     740   139.9   494.5  
Lowell   16.1     736   161.4   549.6  
Albany   19.4     7728   157.5   834.5  
Buffalo   15.6     722   158.2   598.7  
New York   13.9     708   191.2   669.5  
Newark   14.2     702   156.5   611.2  
Louisville   15.7     701   145.9   633.2  
Dayton   14.6     690   132.7   568.9  
Memphis   20.4     684   170.7   751.9  
Providence   15.2     680   168.7   616.4  
Fall River   16.6     673   219.6   635.1  
Oakland   11.0     666   97.8   501.4  
Cambridge   13.4     659   155.5   595.2  
Cincinnati   16.1     650   142.4   742.6  
St. Louis   14.7     629   158.1   618.3  
Richmond   19.2     608   179.4   789.8  

--15--

Group 2:--Had rather severe visitations of influenza with moderate and in some instances high rates for pneumonia, and pneumonia, tuberculosis, heart disease and nephritis combined.

Group 3:--Had very low rates for influenza with low general death rates and low rates for pneumonia, and pneumonia, tuberculosis, heart disease and nephritis combined. There were some exceptions to this rule. The general death rates in Toledo and Indianapolis are a little higher than in the other cities of the group. Chicago and Atlanta had a pneumonia rate above the average--but all the others were far above the average for pneumonia.

Table IX
City General
Death Rate
per 100,000
Total Excess
Death Rate,
Influenza,
Pneumonia
for 3 Epidemic
Periods--
Jan. 1917, to
March, 1920
Death Rate per 100,000,
1911-1917, Inc.
Pneu-
monia
  Pneumonia,
Tuberculosis,
Heart Disease
and Nephritis
 
Columbus       14.9             598       130.0       483.8      
Los Angeles   12.5     559   90.8   588.9  
Indianapolis   15.6     246   133.9   616.9  
Milwaukee   12.2     537   133.2   406.0  
Rochester   14.5     534   135.3   570.0  
Atlanta   15.7     533   167.0   604.0  
Chicago   14.4     524   188.0   610.8  
St. Paul   11.0     510   132.6   411.0  
Toledo   16.6     451   123.3   563.1  
Minneapolis   11.9     437   109.8   451.4  

Table X shows that ten of the twelve cities with the highest rates for pneumonia and heart disease had also excessively high rates for influenza and pneumonia during the three epidemic periods from January 1917, to March, 1920. The exceptions were Fall River and Richmond and their rates were not far below the average.

Table X
City Pneumonia and
Heart Disease
Total Excess
Death Rate, 3
Epidemic Periods
Jan., 1917, to
June, 1920
Boston       412             822      
Worcester   412     822  
Pittsburgh   403     1305  
Baltimore   393     810  
New Haven   384     765  
New Orleans   381     854  
Nashville   380     933  
Washington   376     885  
Fall River   375     673  
Richmond   374     608  
Philadelphia   367     1087  
San Francisco   362     874  

Immunity

Does an attack of influenza confer immunity and if so what is its duration?

This question is of the greatest interest and importance to the health officer but unfortunately, as in other phases of the subject of influenza, there is much conflicting testimony.

There has been a tendency to ascribe the relatively low mortality in the age groups over 40 to an immunity conferred by the epidemic beginning 1889-90 and lasting about three decades. Such an immunity would be expected to protect those above the age of 30. As a matter of fact the significant fall in the death rate is not apparent until after 40--the rate between 30 and 40 being almost as high as that from 20 to 30.

If such an immunity was actually conferred from 1889 to 1892 it is inconceivable that it would operate to protect those from ten to twenty years of age at that time and afford no protection to the group from one to ten years of age.

Arnold17 made a study of persons attacked in Leicester, England, in the three epidemic waves--summer, 1918; autumn, 1918, and early in 1919, the peak week of these waves being the weeks ending July 20, November 2, and March 1. He concluded the considerable immunity to the autumn wave was exhibited by those who had suffered in the midsummer wave but that only slight immunity was evident in the February, March wave.

MacEwen18 found that the percentage of those attacked in autumn in schools, colleges and among London police was much less among those who had had influenza in the summer wave. For example, of 1224 pupils in Finchley Council schools only 13% of those alleging an attack in the summer had influenza in the autumn, while 35% of those with no record of previous influenza were attacked.

Scoccia19 states that all of the eighty nurses in the Spezia Hospital were attacked during May and June--late in September the epidemic recurred and not one of these 80 nurses was attacked.

Hamilton and Leonard20 found that in a girls' school by strict quarantine the first outbreak in November, 1918, was confined to three cottage units. In January the disease returned but it spared those attacked in November.

Vaughan,21 judging from experience in army

--16--

camps, states that the mild influenza of April, 1918, gave a marked degree of immunity against the epidemic of September, 1918.

Dr. Henry F. Vaughan , Commissioner of Health, Detroit, Michigan, estimates that in the period September, 1918, through March, 1919, that 125.000 cases occurred in Detroit or that 14% of the population was attacked.

If the 1920 epidemic was unaffected by immunity conferred in 1918-19, then 14% of the cases would probably occur among those previously attacked. Actually on 10% of the 1920 cases gave a history of a previous attack. From this Vaughan concludes that there was an effect in 1920 from the immunizing influence of the epidemic of 1918-19.

Frost found from a canvass in Baltimore of 4078 persons known to have escaped influenza in 1918 and 1059 known to have had influenza in 1918 that there was little difference in the percentage attacked in the two groups--and concludes form this that the 1918 attack did not confer immunity against attack in 1920 and that the immunity following an attack of influenza must be of short duration.

Judging from the results of his surveys in Baltimore, Frost believes that there is a transient immunity of a few months duration at least.

In analyzing the Baltimore epidemic he made a first canvass of 33,776 people between November 20 and December 11, 1918, and a second canvass of the same population in January to determine the extent of the recrudescence reported in December. Among 32,600 people included in this canvass, 724 cases of influenza were found to have occurred since the previous survey. Of this number, 121 cases were reported as second attacks, but on investigation through the attending physicians or by a medical officer, the clinical diagnosis of both attacks as influenza was confirmed in only 26 cases, or 0.37 per cent of the total, and even in these cases the diagnosis is necessarily uncertain. Considering that 23 per cent of the population had had influenza prior to December 11, the proportion of second attacks should have been much greater if no immunity had been acquired.

If the prepandemic rises shown on Chart II were due to the same disease as that responsible for the great pandemic rise of September, 1918, then we would expect that the rise in September, 1918, would be modified according to the number of people attacked in April, 1918--and the degree of immunity conferred by such an attack.

What effect, if any, did the previous epidemics and especially the epidemic of March-April, 1918, have upon the prevalence of influenza in September-October, 1918? Table XI shows annual excess death rates, influenza and pneumonia, for certain cities for the following periods:

A careful study of this table shows certain striking features but no regularity in the relation between the four periods covered in the matter of prevalence.

  1. Certain cities were hard hit in all periods, notably Pittsburgh, Philadelphia, Kansas City and Nashville.

  2. Certain cities escaped with low rates in all periods--Minneapolis, St. Paul and Chicago.

  3. Certain cities with high rates in the great pandemic had low rates in the other epidemic periods--Boston, Worcester, New Haven, Fall River and New Orleans.

  4. Certain cities with low rates in the great pandemic and high rates in one of the other periods--Birmingham, Milwaukee, Columbus, St. Louis and Indianapolis.

    The table does not show that the prepandemic waves had any definite or consistent influence on the prevalence in the great pandemic.

    We must bear in mind, however, that mortality statistics alone do not accurately indicate prevalence unless we have knowledge of the ratio between cases and deaths also. Thus the prevalence in a mild epidemic might appear low because of a low case fatality rate, yet the reverse might be an actual fact. For this reason it is difficult to estimate the immunizing influence of an epidemic by mortality rates alone. A very mild epidemic with very few deaths might be very potent in immunizing large numbers of the population provided that an attack of the disease produced immunity. We cannot conclude, therefore, that a low mortality rate in April, 1918, necessarily means a low prevalence.

    On the other hand, it is natural to conclude that the high mortality rates shown by Birmingham, Pittsburgh, Philadelphia, Nashville and Kansas City had apparently no effect on the pandemic prevalence although Birmingham

--17--

had a relatively low prevalence in the pandemic of September-October, 1918.

These indications suggest one of two things: either that in the great pandemic we were dealing with a new and entirely different disease, or that the immunity conferred by an attack, if any, was of a very fleeting character--a matter of less than six months.

Table XI
Actual Excess Death Rates per 100,000 Influenza and Pneumonia (All Forms) for Each Epidemic Period.
City Jan., 1917
to
Sept., 1918
Sept., 1918
to
June, 1919
January
to
March, 1920
Albany, N.Y.           33                     631                     64          
Atlanta, Ga.   80     330     123  
Baltimore, Md.   87     629     94  
Birmingham, Ala.   238     497     120  
Boston, Mass.   53     685     84  
Buffalo, N.Y.   67     567     88  
Cambridge, Mass.   62     532     65  
Chicago, Ill.   25     404     95  
Cincinnati, Ohio   52     531     67  
Cleveland, Ohio   107     546     87  
Columbus, Ohio   65     409     124  
Dayton, Ohio   58     532     100  
Fall, River, Mass.   40     602     31  
Grand Rapids, Mich.   38     -----     -----  
Indianapolis, Ind.   67     354     125  
Jersey City, N.J.   70     -----     -----  
Kansas City, Mo.   132     689     222  
Los Angeles, Cal.   7     505     47  
Louisville, Ky.   113     530     58  
Lowell, Mass.   97     548     91  
Memphis, Tenn.   75     470     139  
Milwaukee, Wis.   98     337     102  
Minneapolis, Minn.   47     299     91  
Nashville, Tenn.   135     662     136  
Newark, N.J.   72     538     92  
New Haven, Conn.   45     629     91  
New Orleans, La.   35     761     57  
New York, N.Y.   60     544     104  
Oakland, Cal.   30     528     108  
Omaha, Neb.   72     562     109  
Philadelphia, Pa.   175     814     97  
Pittsburgh, Pa.   238     891     176  
Providence, R.I.   58     534     88  
Richmond, Va.   40     520     48  
Rochester, N.Y.   53     433     48  
St. Louis, Mo.   103     406     120  
St. Paul, Minn.   37     386     87  
San Francisco, Cal.   33     724     117  
Syracuse, N.Y.   60     580     147  
Toledo, Ohio   65     329     57  
Washington, D.C.   80     689     116  
Worcester, Mass.   42     618     94  
Average   74.96     544.37     97.72  

If an attack of influenza conferred immunity for a period of years we would expect an epidemic in April to modify a later epidemic in September of the same year.

Thus cities which are severely stricken in March-April, 1918, might be expected to escape with relatively low mortality in September-October, 1918.

On Chart X are plotted the excess death rates, influenza and pneumonia, all forms, in the epidemic periods, January, 1917, to September, 1918, and September, 1918, to June, 1919, for the forty large cities. The excess death rates were plotted as "relative figures" or variations form the mean, which is placed as 100 on the chart. The cities are arranged in the order of the severity of their death rate during the first of the two epidemic periods in order to bring out any correlation between the rates in the two epidemics.

The five cities which suffered most severely in the period January, 1917, to September, 1918, were Birmingham, Pittsburgh, Philadelphia, Nashville and Kansas City.

In the epidemic period September, 1918, to June, 1919, Birmingham escaped with a rate below the average--but Pittsburgh and Philadelphia were again hit hardest and Nashville and Kansas City were near the top of the list.

Louisville and Cleveland, too, were above the average for both epidemic periods. St. Louis and Milwaukee were more orthodox, being above the average for the first epidemic period and were well below the average in the second.

Baltimore, Washington and Atlanta were near the average for the first period. In the second period Baltimore suffered severely, Washington more severely, while Atlanta escaped with a low rate.

Many other examples of the contradictory results might be cited. Syracuse and Rochester both had low rates in the first period and Syracuse had an apparently compensatory high rate in the second, while Rochester again escaped with a low rate.

The best examples of cities which appear to give a compensatory high rate in the second period after a low rate in the first are New Orleans, San Francisco, and Boston. Striking examples of low rates in both epidemic periods are furnished by Toledo, Minneapolis, St. Paul, and Chicago.

Chart XI compares in the same manner the two epidemic periods--first, September, 1918, to June, 1919; second, January to March, 1920.

Here again while some cities seem to show the rates for the second epidemic to have been modified by the first, others suggest no such relation. New Orleans, Boston, New Haven, and Fall River, all with high rates in the period beginning September, 1918, have relatively low rates in 1920.

Birmingham, Memphis, St. Louis, Columbus, Indianapolis and Atlanta--with low rates in

--18--

On the other hand, Pittsburgh, Kansas City, San Francisco, Syracuse and Washington were severely stricken in both periods, while Oakland, Los Angeles, Rochester, St. Paul, and Minneapolis had low rates in both periods.

A careful study of Table XI shows many instances in which low and high rates alternate in successive epidemic periods, suggesting that immunity conferred in one epidemic period plays a part in the low rate which follows. Unfortunately for this conclusion quite a number of cities had high rates for all three periods so that charts X and XI do not show any consistent relation between a low rate in one epidemic period and a high rate in another.

It is possible that other reasons independent of immunity or lack of immunity may be found to explain why Pittsburgh, Nashville, Kansas City and Philadelphia had high rates in all periods.

Etiology

Two years after the epidemic of '89-90, Pfeiffer,22 working with purulent bronchitis and broncho-pneumonia, observed and cultivated small gram negative bacilli, which he previously had seen in the sputum of cases during the great epidemic. He demonstrated that these organisms could not be cultivated by the methods used by other workers in the epidemic of 1889 and this fact was held to explain failure to isolate during 1889. Pfeiffer felt justified in attributing the 1889 epidemic to this cause which he named the influenza bacillus.

Pfeiffer claimed that he found influenza bacilli in all fresh uncomplicated cases of influenza. He also claimed that they were found only in cases of influenza, acute or convalescent. He described what he called pseudo-influenza bacilli-larger than influenza bacillus and with a tendency to thread formation, as occurring on broncho-pneumonia complicating diphtheria.

Weichselbaum23 confirmed Pfeiffer's findings in 1892, and in 1897 Grassberger24 published similar findings. The French25 workers did not accept the claims of Pfeiffer.

Many workers have since been unable to demonstrate the existence of pseudo-influenza bacilli as a group. Since Pfeiffer's time and until the recent pandemic the great majority of workers have failed to find the influenza bacilli in cases of influenza in such high percentages as reported by Pfeiffer and have found B. influenzae in those suffering from other diseases and in health persons.

Is the B. Influenzae the Cause of Influenza?

The experience of workers in the great pandemic and since September, 1918, has resulted in very conflicting reports concerning the etiology of influenza. While Pfeiffer's claims have received some new support, many believe B. influenzae to be only a secondary invader.

Abrahams26 considers the primary infection to be due to B. influenzae and that pneumococci, streptococci, or diplostreptococci are secondary invaders, producing the fatal result. Rucker and Werner27 held a similar view. Keegan28 recovered B. influenzae from 86% of necropsies and considers it the primary cause.

Spooner, Sellards and Wyman,29 working at Camp Devens, consider that B. influenzae is established as the cause of the epidemic. The Kitasato Institute30 findings are that the influenza bacillus of Pfeiffer was the cause of the 1918 pandemic.

These writers practically all believe that the initial damage is done by the influenza bacillus which paves the way for secondary invaders.

Lucke, Wight and Kime31 report frequent finding of B. influenzae and believe that if not the primary cause it is an indication of influenza.

Duval and Harris32 believe B. influenzae is the cause of influenza and calim to have found specific immune bodies to B. influenzae in the blood of influenza patients during and after infection.

Small and Stangl33 in the 1920 epidemic in Chicago found B. influenzae in 100% of acute influenza patients. in the pneumonia cases studied they found pneumococci in 84%; hemolytic streptococci in 18.7% and B. influenzae in 75%.

Many observers finding B. influenzae in acute influenza in a very high percentage of cases believe that faulty technique may explain failure to find in all cases. Park evidently holds such a view and Keegan suggests the same explanation of some of the failures to find in high percentages in the early acute stage.

Wade and Manalang34 report an interesting variation in morphology of the Pfeiffer bacillus:

"It has been found that three different strains of an organism supposed to be Bacillus influenza will, under certain conditions, abandon

--19--

the usual bacillary form and grow as a frank fungus, morphologically of the Dicomyces type. Under other conditions they show less modification, the most striking feature then being the production of condiospores, bodies of a type not found in true bacteria. That this organism may not be the true Pfeiffer bacillus is conceivable, of course, but considering the source, morphology, ordinary cultural characteristics, and the poison production of the one strain tested, we consider this highly improbable. Further, we are confident that the cultures do not contain any contaminating organisms, as may be suggested. In short, we believe that we have been dealing solely with the true Pfeiffer bacillus."

It is interesting to note in connection with the results of Wade and Manalang cited above that Pfeiffer described what he called pseudo influenza bacilli with a tendency to thread formation.

Jordan35 found the Pfeiffer bacillus in 64% of cases examined October, 1918, to February, 1919. The other organism most commonly found was the Mathers streptococcus. He found the pneumococcus in 20%. He states:

"The observation carried out by the aerobic blood-agar plate method and recorded in this paper have not shown the predominance or constant presence of any one organism in the upper respiratory tract of influenza patients. the Pfeiffer bacillus, however, has been more conspicuous than any other organism, particularly in comparison with its relative infrequency in cases of rhinitis and tonsillitis examined during the same epidemic period."

Albert and Kelman36 find that

"the influenza bacillus produces a toxin which is fatal to mice, guinea-pigs and rabbits almost as rapidly as are broth cultures of equal dosage. This toxin is produced very rapidly and can be obtained by filtering broth cultures. It is not possible to state definitely whether it is an endotoxin or an extracellular one.

"Although the symptoms of intoxication as seen in lower animals following injections of the Pfeiffer bacillus are suggestive of the profound intoxication seen in connection with many cases of the epidemic disease influenza in the human being, these experiments do not furnish any proof that the Pfeiffer bacillus has any specific etiologic relationship to that disease. On the other hand, they suggest that a possible etiologic relationship cannot be ignored."

Huntoon and Hannum37 claim that B. influenzae produces a toxin which produces in animals congestion of the respiratory tract with hemorrhages into the alveoli.

Ferry and Houghton38 report similar findings in regard to a toxin and claim to have produced an antitoxin which protected guinea pigs.

Roos39 also reports on the toxin production of b. influenzae and reports similar results to Huntoon and Hannum in the effect of the toxin in predisposing to invasion by secondary organisms.

Another very large group believes that some unknown organism causes the initial damage by breaking down the natural protective barriers and permitting secondary invasion by the pneumococci, streptococci and B. influenzae, which they consider as a secondary invader only.

Park40 say:

"These results appear to us to throw the influenza bacilli in the cases studied as clearly into the class of secondary invaders.

"We believe that the other microörganisms, such as certain streptococci and pneumococci which are under suspicion in different localities will be found after subjection to similar severe tests not to possess the necessary identity of characteristics to allow them to remain under serious consideration as the primary agent in this epidemic, but rather like the influenza bacillus, to be reckoned among the most important of the secondary invaders.

"Our final conclusion is, therefore, that the microörganism causing this epidemic has not yet been identified."

Howard41 also believes that the influenza bacillus is a secondary invader only, which may produce a terminal broncho-penumonia.

Wollstein,42 working with sera of convalescents from influenza concludes:

"The pateints' serological reactions indicate the parasitic nature of the bacillus, but are not sufficieintly stable and clean-cut to signify that Pfeiffer's bacillus is the specific inciting agent of epidemic influennza. they do, however, indicate that the bacillus of Pfeiffer is at least a very common secondary invader of influenza, and that its presence, influences the course of the pathological process."

Sellards and Sturm43 report finding an organism

--20--

with all the characteristics of B. influenzae in a series of measles cases in both sputum and conjunctivae with the disappearance of the organism with subsidence of symptoms in three-fourths of the cases. They consider that presence of B. influenzae in two diseases shows a causal relation to neither.

The Rôle of Streptococci, Pneumococci and Other Organisms

There seems to be relative unanimity of opinion as to the rôle played by the pneumococci, streptococci and other organisms commonly found in the upper air passages. Practically all observers consider these as secondary invaders.

Many writers44 have associated streptococci with the complications of measels and have stressed the likelihood of streptococcus carriers suffering severely from those complications. This is suggestive of what may happen in influenza with both streptococci and pneumococci.

Hall, Stone and Simpson45 found pneumococci in influenza sputa in 302 cases, 273 of which were Type IV.

Blanton and Irons46 found streptococci in 451 and Type IV penumococcus in 148 cases out of a total of 749. Thirty per cent. of the streptococci were hemolytic, 70% non-hemolytic. Opie47 and his co-workers found pneumococci in 61.2% of those examined. They believe that the fatal factor was a lobar pneumonia, and that the streptococcus hemolyticus played an insignificant part in the production of pneumonia.

In thier second report these authors modify their statement in regard to the significance of streptococci:

"The sequence of events that occurs in many cases of influenza is as follows: B. influenzae descends into the bronchi; pneumococci (in this camp usually Type IV) invade the inflamed bronchi; enter the lung, and produce either lobar pneumonia or bronchopneumonia. Hemolytic streptococci may descend and infect the pneumonic lung. It is not improbable that hemolytic streptococci may invade the bronchi previously infected with B. influenzae and cause bronchopneumonia in the absence of pneumococci.

"When hemolytic streptococci invade the lung either with or without preceding pneumococcus infection, there may be no suppuration of the lung. It is probabel that death has occurred before there is opportunity for the formation of abscesses. Streptococcus hemolyticus may pass through a pneumonic lung and appear in the heart's blood although there has been no suppuration. It is not imprlbable tghat it may pass through the lung and produce empyema, the lung remaining free from actual suppuration. Lobar pneumonia appears to resist suppuration more effectively than broncho-pneuonia; but fatal streptococcus infection is common with both.

"Infection with hemolytic streptococci may spread as an epidemic through the pneumonia wards of a hostpial. A single patient with streptococcus pneumonia is a source of grave danger to every patient in the same ward. Superimposed infection with hemolytic streptococci increases the mortality of pneumonia so that it may reach from 50 to 100 per cent. of all patients with pneumonia."

Brige and Havens48 found streptococcus hemoluyticus in 60 cases ante mortem and in many post mortem cases of influenza pneumonia.

Goodpasture49 in the later stage of the epidemic found streptococcus hemolyticus in all of the 16 cases which came to necropsy.

Tunnicliff50 states that Mathers found a green producing streptococcus in 87% of 110 cases. The same writer declares that specific opsonins develop in the course of influenza, but that with onset of pneumonia these opsonins decrease. The changes in opsonic power are specific for the green producing coccus and no fluctuations being observed for streptococcus hemolyticus, B. influenzae and M. catarrhalis.

Rosenow51 from immunologic studies believes that there is a pandemic strain of green producing diplostreptococci in influenza. He produced results in guinea pigs similar to the pathologic changes in human lungs in influenza. Rosenow also claims to have found evidence in group reaction between green producing and hemolytic strains of streptococci.

Dochez, Avery and Lancefield52 claim that immunological differences have been shown to exist between strains of streptococcus hemolyticus of the human type and that four biological types have been identified by means of the reactions of agglutination and protection and that at least two ther types have been encountered and the indications are the more exist.

Jordan53 found the Mathers' coccus about

--21--

as frequently as B. influenzae and states that its association with pneumonia cases seemed to be closer than that of the Pfieffer bacillus.

Howell and Anderson54 in a complement fixation experiment conclude:

"The outstanding feature of this work on complement fixation with influenza serum is the large number of positive results with certain strains of the viridans group of streptococci isolated from cases of influenza at Camp Meade and in Chicago. The evidence indicates that such organisms probably played an important part in the morbid process even in other places. Serum from influenza patients in several different places appears to have acquired similar new properties."

Claims that a filterable virus is the primary cause of influenza havae been made by many writers. Nicolle and Lebailly,55 de la Rivière,56 da Cunha, Magalhaes, and da Fonseca,57 and Gibson, Bowman, and Connor58 have reported transmission of the disease by filtrates, and von Angerer,59 da Cunha, Magalhaes, and da Fonseca, Leschke,60 Bradford, Bashford, and Wilson,61 and Gibson, Bowman, and Connor,62 claim to have cultivated minute filterable organisms.

WIlson63 made the following claims:

  1. An organism, of definite morphological and cultural characters, has been isolated from cases of influenza.

  2. It can be demonstrated in the blood, sputum, and other exudates, and in the tissues, post mortem, by appropriate methods of staining.

  3. It belongs to the group of "filter-passers," a group of organisms which pass through bacteriological filters. It has been seen microscopically in the filtrate and has been cultivated therefrom.

  4. It has not been found in a large series of controls.

Arkwright64 took exception to these claims after an attempt to confirm them and as a result Wilson modified the claims by withdrawing the claim that a filter passing organisms had been grown in pure culture.

Other workers, notably Rosenau,65 have failed to transmit the disease by filtrates. Resenau had 100 younhg volunteers of the most susceptible age. He used suspensions of Pfeiffer's bacillus, citrated blood of influenza patients, emulsion of sputa unfiltered and filtered nucous secretions. All results were negative.

McCoy had similar results at San Francisco with enlisted men of the Yerba Buean Naval Training Station. These experiments do not definitely rule out a filterable virus and the results are particularly surprising in a disease which in the field of epidemic times appears to be very readily transmissible.

Wahl, White and Lyall66 failed to transmit the disease to man using the filtrate from a pneumonic lung directly into the subject's nasal passages. They had negative results also with emulsions of B. influenzae.

Conclusions

It is probable that influenza is a disease of great antiquity and that the cause of the world-wide pandemics and interpandmic outbreaks is the same.

With a strong predilection for the winter months we have influenza with us every year--and in retrospect we can detect in the mortality statistics outbreaks reaching epidemic proportions in 22 out of the 30 years since 1889.

In 1918-19 the attack rate varied form 15 to 40% and seemed to be highest in teh age group five to nine--declining in each successive age group except 25 to 34, which exceeded the rate for the group 15 to 24.

The incidence in 1918-19 was greater in females than males and the disparity was most noticeable in the ages from 25 to 40, indicating, according to Frost, that the females from 15 to 45 were either more susceptible or more intimately exposed to infectoin than males of corresponding age.

Case fatality in the 1918-198 epidemic was about 2% and was slightly higher in females under 15 and very much higher in females over 60 than in males of corresponding ages. From 15 to 60 the case fatality was much higher in males.

There was great variation on 40 large cities in explosiveness of the epidemic and in the severity as measured by the excess death rates for the entire epidemic period.

There seemed to be some correlation between explosiveness and the severity as measured by excess death rates--the greatest mortality being usually but not always in cities with a high explosive index.

There was little consistency in the explosiveness of the two epidemics, 1918-19 and 1920 upon comparing the indices in the various cities. Cities with a high explosive index in

--22--

1918-19 often had a low index in 1920. Most cities with a high explosive index for 1920 had a low index for 1918-19. Memphis, Nashville, and Washington had a high index of explosiveness in both epidemics.

there seemed to be some correlation between explosiveness and the general death rate and the rates for the four principal causes of death--pneumonia, tuberculosis, heart disease and nephritis. There seemed to be considerable correlation between the total excess death rates for the epidemic periods and the general death rate and the death rates for pneumonia, tuberculosis, heart disease and nephritis.

All the evidence points to an immunity of relatively short duration--probably for months rather than years.

the etiologic cause is unknown. There is not sufficient evidence to warrant the view that B. influenzae is anything more than a secondary invader. The claims for a filterable virus are strong but much additional work will be necessary to make certain many things which are now only possibilities.

A survey of the whole field and all available literature convinces me that while further epidemiologic studies will have great value and be of intesne interest they will not furnish a solution of the problem. We must have more intensive, comprehensive and sustained laboratory research, using the body fluids and secretions of influenza cases for material if we hope to solve the problem and secure the biologic aids which we now lack for the prophylaxis and treatment of influenza.


References

1. Hirsch: Handbook of Geographical and Historical Pathology, Vol. i.

2. Webster, Noah: A Brief History of Epidemic and Pestilential Diseases, Vol. ii, Hartford, 1799.

3. Nothnagel's Encyclopedia of Practical Medicine.

4. Frost, W.H.: Jour. A.M.A., Vol. lxxiii, p. 313, Aug. 2, 1919.

5. Kinsella, R.A.: Jour. A.M.A., Vol. lxii, p. 717, March 8, 1919.

6. Soper, George A.: Jour. A.M.A., Vol. lxxi, p. 1907, Dec. 7, 1918.

7. Winslow, C.E.A., and Rogers, J.F.: Jour. of Infec. Dis., March, 1920.

8. Brownlee, John: The Lancet, Nov., 8, 1919.

9. Stallybrass, C.O.: The Lancet, Feb. 14, 1920.

10. Spear, B.E.: The Lancet, March 13, 1920.

11. Abbott, S.W.: Report Massachusetts State Board of Health, 1890.

12. Parsons: Local Government Board Report of the Influenza Epidemic, 188990.

13. WInslow, C.E.A., and Rogers, J.F.: Jour. of Infec. Dis., March, 1920.

14. Frost, W. II.: Statistics of Influenza Morbidity, Pub. Health Reports, March 12, 1920.

15. Frost, W. II.: Statistics of Influenza Morbidity, Pub. Health Reports, March 12, 1920.

16. Pearl, Raymond: Influenza Studies, Pub. Health Reports, Aug. 8, 1919.

17. Arnold, M.B.: Forty-eighth Annual Report, Local Gov't Board, 1918-19.

18. Report of Local Gov't Board, 1918-19.

19. Scoccia, V.: Policlinico, Rome, Dec. 22, 1918, Vol. xxv, No. 51.

20. Hamilton, J.H., Leonard, A.H.: Jour. A.M.A., March 22, 1919.

21. Vaughan, Victor C.: Jour. A.M.A., Sept. 20, 1919, p. 890,

22. Pfeiffer: Zeitschrift J. H., 13, 357, 1803.

23. Weichselbaum, A.: Wien. klin. Wchnachr., 32, 459, 1892.

24. Grassberger, I.: Ztschr. f. Hyg., 25, 453, 1897.

25. Rosenthal: Arch. d. méd.l, 1, 1281, 1905.

26. Abrahams: Jour. A.M.A., Vol. lxxii, p. 454, Feb. 8, 1919.

27. Rucker and Werner: New York Med. Jour., Dec. 21, 1918, No. 25; Jour. A.M.A., Vol. lxxii, p. 1066, Jan. 18, 1919.

28. Keegan, S.J.: The Prevailing Pandemic of Influenza, Jour. A.M.A., Vol. lxxi, p. 1051, Sept. 22, 1918.

29. Spooner, L.H., Sellards, A.W., and Wyman, J.H.: Serum Treatment of Type I Pneumonia, Jour. A.M.A., Vol. lxxi, p. 1810, Oct. 19, 1918.

30. Editorial, The Etiology of Influenza, Jour. A.M.A., Vol. lxxii, p. 1001, April 5, 1919.

31. Lucke, B., Wight, F., and Kime, E.: Arch. Int. Med., Chicago, Aug. 15, 19919, 24, No. 2.

32. Duval, C.W., and Harris, W.H.: Jour. Infec. Dis., Chicago, November, 1919, 25, No. 5.

33. Small, J.C., and Stangl, F.H.: Jour. A.M.A., Apr 10, 1920, p. 1004.

34. Wade, H.W., and Manalang, C.: Jour. Exp. Med., Vol. xxxi, No. 1, Jan. 1, 1920.

35. Jordan, E.O.: Jour. Infec. Dis., Vol. xxv, No. 1, July, 1919.

36. Albert, H., and Kelman, S.R.: Jour. Infec. Dis., Vol. xxv, No. 6, December, 1919.

37. Huntoon, F.M., and Hannum, S.: Jour. of Immunology, Baltimore, July, 1919, Vol. iv, No. 4.

38. Ferry, N.S., and Houghton: Jour. A.M.A., Sept. 27, 1919.

39. Roos, C.: Jour. A.M.A., Sept. 27, 1919.

40. Park, W.H.: Am. Med. Ass'n, Vol. lxxiii, No. 5, Aug 2, 1919.

41. Howard, S.E.: Bacteriological Findings in Epidemic Influenza, Bull. Johns Hopkins Hosp., 30, 13, 1919; Jour. A.M.A., 72, 678, March 1, 1919.

42. Wollstein, Martha: Jour. Exp. Med., Vol. xxx, No. 6, Dec. 1, 1919.

43. Sellards, A.W., and Sturm, E.: Bull. Johns Hopkins Hosp., Baltimore, November, 1919, 30, No. 345.

44. Irons, E.E., and Marine, D.: Jour. A.M.A., March 9, 1918; Levy, R.J., and Alexander, H.L.: Hour. A.M.A., June 15, 1918; Cummings, J.C., Spruitt, C.B., and Atem, E.J.: Jour. A.M.A., March 8, 1919.

45. Hall, J.N., Stone, M.C., and Simpson, J.C.: Jour. A.M.A., Dec. 14, 1918.

46. Blanton, W.B., and Irons, E.E.: Jour. A.M.A., Dec. 14, 1918.

47. Opie, E.L., Freeman, A.W., Blake, F.G., Small, J.C., Rivers, T.M.: Jour. A.M.A., Jan. 11, 1919, Feb. 22, 1919.

48. Birge and Havens, Jour. A.M.A., April 12, 1919.

49. Goodpasture, E.W.: Jour. A.M.A., March 8, 1919.

50. Tunnicliff, R.: Jour. A.M.A., Nov.. 23, 1918.

51. Rosenow, E.C.: Jour. A.M.A., May 31, 1919.

52. Dochez, A.R., Avery, O.T., and Lancefield, R.C.: Jour. of Exp. Med., Vol. xxx, No. 3, Sept. 1, 1919.

53. Jordan, E.O.: Jour. of Infec. Dis., Vol. xxv, No. 1, July, 1919.

54. Howell, Katherine, and Anderson, Ruth: Jour. of Infec. Dis., Vol. xxv, No. 1, July, 1919.

55. Nicolle, C., and Lebailly, C.: Compt rend. Acad., 1918, clxvii, 607.

56. de la Rivière, R.D.: Compt. rend. Acad., 1918, clxvii, 606.

57. da Cunha, A. Magalhaes, O., and da Fonseca, O.: Brazil Med., 1918, xxxii, 376, referred to in Med. Rec., 1919, xcv, 457.

58. Gibson, H.G., Bowman, F.B., and Connor, J.L.: Brit. Med. Jour., 1918, ii, 645.

59. von Angerer: Münch. med. Woch., 1918, ii, 1280, abstracted in Bull. Inst. Pasteur, 1919, xvii, 161.

60. Leschke, E.: Berl. klin. Woch., 1919, lvi, 11, referred to in Med. Rec., 1919, xcv, 456.

61. Bradford, J.R., Bashford, E.F., and Wilson, J.A.: Brit. Med. Jour., 1919, i, 127.

62. Gibson, H.G., Bowman, R.B., and Connor, J.I., Brit. Med. Jour., 1919, ii, 331.

63. Wilson, J.A.: The Quarterly Jour. of Med., Vol. xii, No. 47, April, 1919.

64. Arkwright, J.A.: Brit. Med. Jour., Aug. 23, 1919.

65. Rosenau, M.J.: Jour. A.M.A., Aug. 2, 1919.

66. Wahl, H.R., White, G.B., and Lyall, H.W.: Jour. Infec. Dis., Chicago, November, 1919, 25, NO. 5.

--23--



Transcribed and formatted for HTML by Patrick Clancey, HyperWar Foundation