Part 6

Death rates from all causes by weeks in certain large cities of the United States during the winter of 1918–19. (_Pearl._) ]

CHART VII.

Concerning geographical position, he did find some slight relationship with linear distance from the city of Boston, where the epidemic was supposed first to have begun in this country:

“This result means that the greater the linear distance of a city from Boston the less explosive did the outbreak of epidemic mortality in that city tend to be. This is in accord with the general epidemiological rule that the force of an epidemic tends to diminish as it spreads from its primary or initial focus. It must be noted, however, that the correlation coefficient in this case is not large. It is barely past the value where it may safely be regarded as statistically significant. This fact may probably be taken to mean that influenza does not follow the epidemiological law referred to with anything like such precision as do some other epidemic diseases, notably poliomyelitis.”

These factors having been found to be of little value in his attempt to explain the varying curves in the 39 different cities, Pearl next correlated the explosiveness of the epidemic mortality with deaths from all causes, deaths from pulmonary tuberculosis, from organic heart disease, from acute nephritis and Bright’s disease, from influenza, from pneumonia (all forms), from typhoid fever, from cancer and from measles, in the various cities.

“The outstanding fact which strikes one at once from this table is the high order of the correlation which exists between the explosiveness of the outbreak of epidemic mortality in these communities and the normal death rate from certain causes of death in the same communities. In the first four lines of the table the correlation coefficients range from about 6 to more than 10 times the probable errors. There can be no question as to the statistical significance of coefficients of such magnitude.

“The highest correlation coefficient of all is that on the first line of the table, for the correlation of epidemicity index with death rate from all causes. The existence of this high correlation at once indicates that an essential factor in determining the degree of explosiveness of the outbreak of epidemic influenza in a particular city was the normal mortality conditions prevailing in that city. In the group of communities here dealt with, those cities which had a relatively high normal death rate had also a relatively severe and explosive mortality from the influenza epidemic. Similarly, cities which normally have a low death rate had a relatively low, and not sharply explosive, increase in mortality during the epidemic.

“It will also be noted that the correlation in the next three lines of the table, namely those of pulmonary tuberculosis, so-called, organic diseases of the heart, and chronic nephritis and Bright’s disease, are of the same order of magnitude as that between the death rate from all causes and the explosiveness of the epidemic outbreak of influenza.”

Pearl suggests that this correlation might arise because of differences in the constitution of populations in the different cities, or, that it was a factor of geographical position, such as the distance from the Atlantic seaboard; but that even after correction of the results for age distribution and geographical position, the net correlations were actually higher than were the gross uncorrected correlations.

“We may conclude that the most significant factor yet discovered 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 relative normal liability of the inhabitants of the several cities to die of one or another of the three great causes of death which primarily result from a functional breakdown of one of the three fundamental organ systems of the animal body, the lungs, the heart and the kidneys.”

Winslow and Rogers studied the relation of the pneumonia death rate from 1901 to 1916 to the influenza death rate of 1918 in 40 large cities of the United States and found a distinct correlation. The cities which have been characterized by a high pneumonia rate in the past are precisely the cities which suffered most severally in the 1918 outbreak. This is not due especially to virulent types of pneumonia organisms in certain sections of the country because they found this same high correlation between total death rates and influenza death rates, in the same cities.

They believe that these high correlations may be the result of weaknesses in the population due to high incidences of organic diseases and tuberculosis in earlier years, or more probably that the correlation is an indirect one, due to the relation between each of the factors studied and one or more underlying conditions affecting both, such as age distribution of the population, race distribution, or social and economic conditions in the various cities studied. Or, finally, it may be that the high rate from tuberculosis and organic disease in 1916 was due to these latter factors, while the high incidence of influenza was due chiefly to proximity to the original focus of infection. None of these explanations are considered entirely satisfactory.

It is important to call attention to the fact that the American observers quoted have been studying the death rate from influenza as it is revealed in the increase of death rate from all causes, whereas Leichtenstern and Wutzdorff, and Greenwood, in his studies in the Royal Air Force have concerned themselves with _morbidity_. The comparison of morbidity and mortality cannot be easily made as we will show when discussing these two subjects, so we cannot conclude that the work of Pearl and of Winslow and Rogers is at variance with the other work quoted. The mortality curves form another characteristic of the local spread of influenza in a community.

It is characteristic of influenza that the curve of deaths does not fall as rapidly as does the curve for influenza cases. Thus in morbidity curves we may expect to find a symmetrical curve for a primary epidemic, but the mortality is rarely if ever symmetrical, the curve rising rapidly and falling very much more slowly.

_Morbidity curves in 1920 recurrences._—The curves of influenza incidence in the recurrence of 1920 have varied in different localities, but in certain communities where the record has been carefully reported the epidemic appears to be characterized by a symmetrical evolution and usually a lower death rate as compared with 1918. The curve of incidence in the State of Massachusetts in January, February and March, 1920, is symmetrical, if anything falling away more rapidly than it ascends, and the duration is at least ten weeks. The crest of the influenza wave in Massachusetts was reached on February 4th, 5th and 6th. The peak is recorded as being in the week of February 7th.

During the 1920 epidemic the author made a house-to-house canvass in six representative districts in the city of Boston covering a population of 10,000 individuals. The curve of incidence of influenza corresponds closely with the curves for the city and the state as a whole. The peak was reached in the same week, the week ending February 7th, the curve was symmetrical, and the duration of the entire epidemic was about the same. The morbidity rate for 1920, according to our influenza census, was but half of that for 1918 for the same population. The recurrent epidemic as we will show later was decidedly milder (see Chart XVIII).

In Detroit the 1920 epidemic reached its peak for morbidity on the 9th day, and that for mortality on the 16th. In 1918 the morbidity peak was not attained until the 15th day and the death peak on the 22d. The recurrent outbreak had nearly run its course within three weeks. The following comparison between the influenza incidence in 1918 and 1920 in Detroit is taken from a report by H. F. Vaughan, Commissioner of Health for that city. In it is shown a comparison of the total figures on the twenty-seventh day of each of the two epidemics:

_A Comparison of the 1918 and 1920 Epidemics of Influenza in Detroit. Statistics Made to Include Through the Twenty-seventh Day of Each Epidemic._ ═══════════════════╤════════════╤════════════╤════════════╤════════════ │ │ │ Normal │ Excess │ │Deaths from │ influenza │ influenza │ Influenza │ influenza │ and │ and │ cases │ and │ pneumonia │ pneumonia │ │ pneumonia │ deaths for │deaths above │ │ │this season │ normal ───────────────────┼────────────┼────────────┼────────────┼──────────── 1920 (Jan.–Feb.) │ 11,202│ 1,642│ 197│ 1,445 1918 (Oct.–Nov.) │ 16,423│ 1,286│ 124│ 1,162 ───────────────────┴────────────┴────────────┴────────────┴────────────

There had been fewer cases reported on the twenty-seventh day of the 1920 epidemic, but these had resulted in a greater number of deaths. On this day the recurrent epidemic had run its course, while the 1918 one was still in full swing. On the twenty-seventh day of 1918 there were 137 influenza cases reported and 49 deaths. On this day in 1920 there were but 24 cases and 34 deaths. Thus the second outbreak was of shorter duration, but was more deadly while it lasted.

Seven weeks of the 1920 epidemic in Detroit killed 0.20 per cent. of the population, two out of every one thousand people. A similar period at the beginning of the epidemic of 1918 witnessed the death of 0.17 per cent. of the population. This was a smaller number, but the epidemic at this time had not completed its course, and continued to be more or less prevalent for twenty-one weeks, resulting finally in the death of 0.28 per cent. of the population. The recurrent epidemic was more highly fatal, but, being of shorter duration, Detroit actually suffered less from it.

SPREAD IN COUNTRIES AND CONTINENTS.

The spread of influenza is usually not limited to a single community. Almost invariably it will travel on to another locality, carried thither by human intercourse, and will there build again a local epidemiologic picture more or less modified by changes in the environment and changes in the virulence of the virus itself.

_Spread, in primary waves._—Reference to the table of epidemics in history will show that in many of the epidemics and in most of the widespread epidemics and pandemics there appears to have been a definite, clearcut, direction of spread from one locality to others. In the recent literature there has appeared considerable discussion concerning the site of origin, the endemic focus of pandemic influenza. Briefly the question raised is as to whether there are single or multiple foci. We will for the time ignore this perplexing question. In either case, after the influenza virus has once attained such communicability as to produce a pandemic it does follow a direct course over countries and continents. This may be followed in resumé in our table.

The disease does not at any time spread more rapidly than the available speed of human communication between the areas affected. If influenza does appear simultaneously in two widely separated communities without having been brought there from a common source it must be that it arose spontaneously from simultaneous increase in virulence of the virus in those localities.

Influenza was prevalent in Turkestan, Western Asia, in May of 1889. It spread first to Tomsk in Siberia and did not appear in Petrograd until the end of October. By the middle of November it had reached Berlin and Paris, and one month later it was epidemic in New York and Boston. Four months had been required for the disease to reach Petrograd from Bokhara in Turkestan, while within two months thereafter it had traveled from Russia to the United States. In both cases the rapidity of spread corresponded to the rapidity of the means of communication of the locality; the caravan in Turkestan and the transatlantic liner to America. North America was widely infected in January of 1890. So, also, Honolulu, Mexico, Hong Kong, Japan. Ceylon first experienced the epidemic early in February, India at the end of the month, Borneo and Australia on the first of March, Mandalay towards the first of May, China and Iceland in July, Central Africa in August and Abyssinia in November of 1890.

It should be noted that influenza was reported to have been prevalent in Greenland at about the same time that it was in Bokhara. There appears to have been no relationship between these two outbreaks.

The spread of the pandemic may be followed also by recording the period of greatest mortality in the various cities. This period at Stockholm followed that at Petrograd by three weeks, and that of Berlin by another week. The period for Paris was a week later than for Berlin, that for London another week later, and that for Dublin three weeks later than that for London. The week of highest mortality in Dublin was later than that for New York or Boston.

The earlier epidemics progressed more slowly. That of 1762 prevailed in Germany in February, in London in April, in France in July, and in America in October. In 1782 it attacked London in May, Exeter two weeks later and Edinburgh early in June. In 1830–1832 the spread from Moscow and Petrograd through Germany required no less than eight months to cover the latter country.

In 1872 the time required for spread from Leipzig to Amsterdam was eighteen days, the same time that was required for a merchant in the latter town to reach Leipzig.

There are many instances on record in which influenza has passed by small towns in its onward course to attack a larger city and only at some later date has the small town, not on the main line of communication, been affected. Not only is the speed of transportation between two communities of importance, but also the volume of the transportation undoubtedly plays a part in the rapidity of development in a second locality. When the disease is carried by a vessel the first places to be attacked are the seaports and the coast towns, be the land a continent or an island. From there it spreads inland either rapidly or slowly according to the transportation facilities. Formerly the question was raised whether influenza spread in continuous lines or radiated in circles. Naturally it follows the direct lines of communication, most of which are radially distributed around large centers.

Leichtenstern calls attention to the fact that in the 1898 epidemic, as in the previous one, the general direction of spread was from East to West across Europe. This was also true of the epidemics of 1729, 1732, 1742, 1781, 1788, 1799, 1833, and 1889.

There have been in Europe two general routes followed by pandemics, a Northern one through Russia and following the lines of travel into Germany and through the countries of Europe; and a Southern path coming from Asia, through Constantinople, and entering Europe from the South, particularly Italy. With the latter, after reaching Europe, the spread is northerly; with the former it is southerly, and usually Spain was the country last infected.

In the United States as well, pandemic influenza usually has spread from East to West, entering the country at or near New York or Boston, and spreading West and South. This was true in the autumn epidemic of 1918.

_Spread in recurrences._—As a rule the manner of spread of a secondary epidemic following the primary pandemic wave is quite different. At a longer or shorter interval following the first spread the disease breaks out anew in one locality or another, sometimes simultaneously in widely separated districts. Sometimes we can distinguish a direction of spread in the relatively small community affected, it frequently being observed that the disease will start up in a large city which has experienced the illness during the first pandemic, and from there will spread to small nearby localities which may have remained free until that time. Again, any clearcut direction of spread may be entirely lacking. It is rare indeed that an epidemic following another by a short interval will follow a definite line over an entire country or continent. Such an example is, however, to be found in the epidemic of 1833, which traveled over Europe from Russia, spreading to the west and the south and following practically the identical path that it had taken in 1830. Even so it was not as widespread, for while the epidemic of 1830 had covered the entire earth, America appears to have escaped the second epidemic.

These disseminated and independent outbreaks are believed to arise from endemic foci in which the virus has been deposited during the progress of its first spread and in which the germ has survived until it has acquired once again exalted virulence.

Usually these endemic outbreaks show in their local configuration, a secondary type of wave. That this is not always the case we have already indicated. The epidemic of 1732–1733 was a recurrence of that of 1729–1730. The epidemic of 1782 had as its source the epidemic of the years 1780–1781. The epidemic of 1788 recurred until 1800, and was quite possibly associated with those of 1802, 1803 and 1805–1806. That of 1830 recurred in 1831–1832. Next we have in 1833 the true pandemic originating in Russia. Recurrences of the epidemic of 1836–1837 were found in 1838 and in 1841. Those spreads which occurred in 1847 and 1848 found successors in the year 1851. In 1890 the influenza outbreaks were as a rule single or isolated and occurred in only a few places of Europe, particularly in Lisbon, Nürnberg, Paris, Copenhagen, Edinburgh, Riga, London, etc. It is reported that there was an unusually severe local outbreak in Japan in August, 1890. In 1891 no general direction of spread was manifested, yet in heavily populated areas, or states rich in lines of communication, especially those of Europe and North America, one could frequently trace some definite direction followed by the disease within these relatively small territories.

A. Netter made the following observation at that time: “La Grippe a fait des explosions simultanées ou successives, et on n’a pu en aucune façon subordonner ces différents foyers comme cela avait été possible en 1889–90. Il parait y avoir eu des reveils de l’épidémie sur divers points.”

Leichtenstern describes the subsequent spread of the disease: “The transfer of the disease by ships which played such an important role in the first epidemic appeared to be insignificant in 1891, in spite of the fact that influenza was present in many of the English colonies. The third real epidemic spread of influenza was a true pandemic which began in the autumn (October) of 1891 and lasted through the whole winter until the spring of 1892. It involved all of Europe and North America and spread to all other lands, but here again the geographic distribution followed no rule. There was no spread of influenza from a central point, no continuous spread following lines of communication, and there was no longer an early predominance in the cities lying on the lines of communication or in the larger cities and commercial centers, as had been the case in the first epidemic. In England in 1891 the first outbreaks occurred frequently in country districts. The epidemic raged nearly four months in the northern part before it finally reached London in May. The same was true of Australia.

“One peculiarity of the recurrent epidemic lay in the much more contagious character of the disease and the remarkably greater mortality. In Sheffield the mortality in the recurrent epidemic was greater than in the pandemic, even though the epidemic picture was that of a primary wave.”

By way of summary of our knowledge of the primary and secondary spread in general up to the epidemic of 1918, we may enumerate the more important characteristics:

1. Occurrence of true pandemics at wide intervals, primarily intervals of several decades.

2. Indefinite knowledge and conflicting evidence regarding site and manner of origin.

3. Apparent transmission chiefly or entirely through human intercourse.

4. Rapid spread over all countries, the rapidity roughly paralleling the speed of human travel.

5. Rapid evolution of the disease in the communities where outbreaks occur, with nearly equally rapid subsidence after several weeks’ duration.

6. Apparent lack of dependance on differences of wind or weather, seasons or climate.

7. Generally low mortality in contrast to enormous morbidity. Variation in the incidence of disastrous secondary infections.

8. Tendency to successive recurrences at short intervals.

SECTION II.

INFLUENZA EPIDEMICS SINCE 1893.

In this section of our report we will describe with as great accuracy as our sources of information will permit, and in as great detail as space will allow the events which have led up to the epidemics of 1918–20 and the various phases of the epidemics themselves. Points of similarity with previous epidemics will be made obvious; the differences, when of significance, will be described and studied in detail.

OCCURRENCE SINCE 1893.

Attempts even today to determine when and where influenza has prevailed in the world since the great pandemic of the last century are met with great difficulties. There are several reasons for this, chief among which is the absence of definite characteristics by which the disease may be recognized. The isolated solitary case baffles positive diagnosis. Nearly every year there are reports in the literature of small outbreaks in institutions or communities in which the clinical picture is that of epidemic influenza. As a rule the conclusion has been in these cases that because the bacteriologic findings did not show a predominance of Pfeiffer’s bacillus the epidemic was not true influenza. This is particularly true in the outbreaks in which the streptococcus predominated. Today our views concerning the bacteriology have changed distinctly, and I believe it is safe to say that the predominance of a streptococcus in a local epidemic in no way rules out influenza, and that the only criteria by which we may judge are the clinical picture and the evidence of high infectivity, together with the epidemiologic characteristics of the local outbreak.

_Period 1893–1918._—A review of the medical literature between 1889 and 1918 gives one a certain impression which may be summarized as follows: Between 1890 and 1900 the disease was in general more highly prevalent in most localities than at any time during the preceding thirty years. At no time during this decade did the annual death rate from influenza in England and Wales fall to anywhere near the figures that had prevailed consistently between 1860 and 1889. Between 1900 and 1915 there was a gradual diminution, but still not to the extent that had prevailed previous to 1889. Since 1915 there appears to have been a gradual increase. During the entire period there has been difficulty in distinguishing between the disease in question and other respiratory tract infections, particularly coryza, sore throat, tonsillitis, and bronchitis. Many of the local epidemics which appear probably to have been true influenza have had associated with them a high incidence of sore throats. We describe this as sore throat, rather than tonsillitis, because the clinician remarks that although the throat is sore there is little if any demonstrable inflammation of the tonsils.