General Manner of Spread in Individual Localities.

Having discussed the mode of propagation of influenza among individuals we will follow the disease as it attacks one person after another in a community and study the epidemiologic picture, drawn no longer with the individual as a unit, but with the community as the unit.

We must here distinguish between a primary epidemic, the first wave of a progressing pandemic, and the secondary type in which may be grouped those large or small recurrences which light up for a period of one to three or more years after the primary wave.

Primary type of epidemic.—One of the first important statistical studies on this subject was that of P. Friedrich who charted the influenza morbidity in Munich between the months of December, 1889, and February, 1890. Similar observations have been made by Parsons, Raats, Linroth, and H. Schmid, following the 1889 epidemic.

Between the occurrence of the first known case of influenza and the time of the first very definite increase in influenza incidence in a community, which interval may be termed the invasion period, there is as a rule two weeks. During this period, of course, more and more cases are occurring, but remain usually sufficiently isolated to attract no public notice. From this point the epidemic develops very rapidly and reaches its peak, usually within two or at most three weeks. In another two or three weeks the incidence has fallen away nearly to normal. The epidemic period comprises from four to six weeks, or, including the invasion period, an entire duration of six to eight weeks. This is the picture produced in a community by a primary uncomplicated epidemic of influenza. Greenwood well describes the salient features of a primary epidemic as “first a rapid and quasi-symmetrical evolution, and second, a frequency closely concentrated around the maximum.” In other words the duration is short, the rise to a peak rapid, and the subsequent fall equally rapid. He showed that in the July and August, 1918 epidemic in Great Britain nearly 80 per cent. of the total incidence in the localities studied was grouped within three weeks time. His curve corresponds so well with that of the Munich epidemic that he is able to superimpose them (Chart I). The rapid rise to a peak, almost explosive in character, more characteristic of this disease than of any other, is to be explained by the high degree of invasiveness of the organism, by the short period of incubation, by the fact that many of the sick continue at their work, thus spreading the disease, and by the non-immunity of large masses of people, together with the fact that the transmission of a respiratory infection is accomplished much more easily than is any other type of infection.

The author holds that the infrequency of immunity is a most important factor in the production of this type of outbreak. The mode of transmission of influenza is the same as that of other respiratory diseases. The infectivity is probably no greater than that of measles, although that indeed is relatively great. The means of transmission are presumably the same in each. Were we able to develop an immunity for influenza of as high degree and permanence as we possess against measles, pandemics of influenza would disappear. We wish to emphasize that the primary type of curve is a phenomenon not peculiar to influenza, but that under certain circumstances it may be found in other infectious diseases, and that it would be found more frequently in the other diseases if the immunity developed against them was of as short duration as it appears to be against influenza. If, for example, measles were to break out in a large group of individuals, none of whom had had the disease, the type of curve would be the same. We will produce evidence supporting our theory under another subject. Of course, other factors such as short incubation period and unusual opportunities for spread through mildly ill individuals play a not unimportant role.

CHART I.

The curves of incidence of influenza in Munich, and of deaths in London during the 1889 and subsequent epidemics. (Greenwood.)

Secondary type of epidemic.—There is a decided difference between the curve of a primary wave as it appears in the onward rush of a new pandemic and that of a secondary wave occurring at a greater or less interval following the primary spread. A secondary epidemic affects, according to Greenwood, a relatively small proportion of the population, is slower in reaching its maximum, and thereafter declines slowly and irregularly, more slowly than it increases. The distribution of the curve is less symmetrical and there is less concentration around the maximum. A secondary epidemic may be characterized by a much higher fatality than a primary one.

We believe that the configuration of a secondary type of wave is due chiefly although not entirely to a certain degree of residual immunity in a large number of individuals remaining from the first spread. There is a striking similarity between Chart I and Chart XXVIII, the latter showing the measles incidence in epidemics among rural or chiefly non-immune troops in the United States army. Chart XXIX shows a similar epidemic among urban or chiefly immune individuals. Here the curves correspond more to those of a secondary type of influenza epidemic. Thus we see that, in the absence of immunity, other infectious diseases may produce the primary type of curve, and that this curve is not a feature of influenza alone.

A striking difference between the two types of waves of influenza is the uniformity and relative constancy of the primary type as contrasted to the great variation in the secondary type. The story of the first spread of influenza in one community is usually similar to that of its spread in any other community. Certain exceptions will be alluded to later. But in the case of recurrent epidemics we may find them more severe or much milder; we may find that they attack a large number of individuals or a very few; we may even find an entire absence of recurrent epidemics in certain communities. The primary curves are relatively uniform; the secondary curves are variable.

Between 1889 and 1894 in England there were four epidemics. The first was primary, symmetrical, and lasted between December and February, 1889–90. The second was asymmetrical and much more fatal in the localities studied by Greenwood. It occurred in the spring and summer of 1891. There was a third epidemic in the autumn and winter of 1891–92 and a fourth occurred from November, 1893 to January, 1894. The third epidemic, according to Greenwood, showed some tendency to revert to the primary type in respect to symmetry, while the fatality rate partook of the character of a secondary epidemic.

Creighton writes: “That which chiefly distinguishes the influenza of the end of the nineteenth century from all other invasions of the disease is the revival of the epidemic in three successive seasons, the first recurrence having been more fatal than the original outbreak, and the second recurrence more fatal (in London at least) than the first. The closest scrutiny of the old records, including the series of weekly bills of mortality issued by the parish clerks of London for nearly two hundred years, discovers no such recurrences of influenza on the great scale in successive seasons.”

Greenwood, who has studied this subject in great detail in England, discusses Creighton’s remarks as follows: “He would be a bold man who challenged the accuracy of Creighton upon a point of historical scholarship, and I have only to suggest that there are faint indications of increased mortality in years following primary epidemics of influenza prior to the nineteenth century. Thus 1675 was a year of primary epidemic influenza, fully described in Sydenham’s Observationes Medicae.

“The nature of the succeeding constitutions is not clear, but the deaths ‘within the bills’ for 1676 were considerably more numerous than in 1675, although smallpox, fever and ‘griping of the guts’ were noticeably less fatal.

“In the English Responsoria (1, 54) the epidemic constitution of 1679 is described as a recurrence of that of 1675—that is, as having the features of primary epidemic influenza. In the five following years intermittents prevailed, and in one (1684) the mortality much exceeded that of 1679, although the deaths from smallpox were fewer. Again, a hundred years later, in 1782, there was a famous summer epidemic of influenza in London which gave rise to much discussion. The London mortalities in 1782 and 1783 were, however, almost equal, when the smallpox deaths (which were nearly three times as numerous in 1783 as in 1782) are subtracted from the total mortality of each year.

“Whether these vague indications are sufficient to permit of our thinking that the epidemic constitution of 1889–94 was not entirely unprecedented is disputable. But the contrast of the latter period with the preceding single epidemic of 1847–48 is striking; that was a primary epidemic without important sequelae.

“We have now to consider whether our experience this year is concordant with that of the early nineties, a reversion to the earlier type, or a new phenomenon.”

After comparing the 1889 curves with those for the July, 1918, outbreak in England, Greenwood concludes: “I believe that the evidence just presented establishes a substantial identity between the summer outbreak of 1918 and the primary wave of 1889–90. We do not need to appeal to any new factor arising out of the war to account for it.

“I next consider the secondary epidemic which we are now experiencing (October, 1918). Evidently our knowledge of the events in 1891 would lead us to feel no surprise at the emergence of a secondary wave, although we could not be sure that the precedent of 1847 would not be followed.

“The summer epidemic of 1918 in the Royal Air Force included nearly 80 per cent. of the total incidence within the three weeks containing the maximum, and the Munich epidemic included just over 80 per cent. within the same limits. Now if the current epidemic has reached its maximum, not more than 65 per cent. of the incidence will probably be so concentrated, and the duration will therefore be longer than in the summer; if, as suggested by the ratio of the last two ordinates, the maximum is not yet attained, then the quota of the three first weeks is likely to be still smaller and the complete duration still longer.

“The diagram of factory sickness leads to the same inference, which is that, from the standpoint of prevalence, the present is a typical secondary epidemic, congruent with that of 1891.

“It appears, then, that the origin of the summer epidemic must be explained upon such epidemiological principles as will account for the primary wave of 1889–90, that the current outbreak is in pari materia with that of 1891, its excessive mortality being mainly due to the accident of season, aided by the special circumstances of overcrowding and fuel shortage which are due to the war. In a word, this is not essentially a war epidemic.”

Wutzdorff found that in some towns, particularly in North Germany, the 1891–1892 wave was almost as extensive as that of 1889–90 had been in other places, but that in general the morbidity in Germany was much lower. He bases these conclusions on a study of the extent of crowding in the hospitals in the two years, on statistics of government physicians, etc.

In Europe the recurrent epidemics of 1891 increased as a rule very gradually, developed slowly, reached their high point frequently after many weeks, and as gradually decreased. The epidemic duration in the winter of 1891–92 lasted four or five months. The morbidity in spite of the longer duration was decidedly less. This is very different from the explosive appearance of 1889 when the peak was reached in fourteen days and the whole epidemic had been completed in six to eight weeks. There were some exceptions to this rule, as in Yorkshire, England, where the epidemic broke out suddenly between the 11th and 13th of April, 1891, had reached its peak after ten days, and for another twenty days declined. Especially interesting was Sheffield, where the first spread began gradually and ran a slow course, while the second epidemic of 1891 began explosively, lasted a short time and declined rapidly, but showed a significantly greater mortality than that of 1889.

The experiences in various communities in the United States have been not unlike those described for European cities. Abbott in describing the successive epidemics in Massachusetts remarked that the 1889–90 spread manifested itself by a sudden rise in the mortality from influenza and pneumonia, beginning about December 20th and culminating in the middle week of January, thereafter falling off quite suddenly in February to about the usual rate for these diseases. The second epidemic two years later began with a more gradual rise in October and November and then increased sharply in December, continued for nearly three weeks at its maximum in January, and declined nearly as sharply as in the previous epidemic two years before.

Winslow and Rogers who have studied the 1918 epidemic as it affected the various towns of Connecticut observed that the outbreak in a given community generally occupied a period of from six to eight weeks, and was steep and abrupt in communities which were badly hit, flatter and more gently sloping in those which escaped lightly. Also the outbreak was more severe in communities receiving the infection early than in those later affected.

Mortality curves.—Pearl has studied the epidemic constitution of influenza in forty-two of the large cities of the United States. He has plotted the annual death rate per 1,000 population from all causes in each week, from the week ended July 6, 1918, up to January 1, 1919, and observed a very distinct difference in the type of curve for deaths from all causes during the epidemic period in the various cities. These differences have been chiefly in respect to the severity and suddenness with which they were attacked. Thus Albany, Boston, Baltimore, Dayton and Philadelphia show an initial explosive outbreak of great force, while Atlanta, Indianapolis, Grand Rapids, Milwaukee and Minneapolis exhibit a much slower and milder increase of the mortality rate. In Albany and Baltimore the curve of the first epidemic outbreak rises to a peak and declines at about the same rate. In Cleveland and St. Paul, on the other hand, the rate of ascent to the peak is very rapid, while the decline is slow and long drawn out.

Some of the cities, such as Albany, show but a single well defined peak in the mortality curve. Others, such as Boston, New Orleans and San Francisco show two peaks; while still others, like Louisville, show three well marked peaks.

Usually the first was the highest and the second and third were progressively lower. Milwaukee and St. Louis, on the other hand, showed second peaks higher than the first. The usual phenomenon, however, was a large first wave followed by smaller ones.

The highest, or maximum peak rate of mortality during the epidemic varied greatly, from 31.6 per 1,000 in the case of Grand Rapids, to 158.3 per 1,000 in the case of Philadelphia.

The death rates which were of the most frequent occurrence were, generally speaking, rates below 70 per 1,000 per week.

The date of the week in which the maximum peak rate occurred was earliest in Boston and Cambridge, where it occurred October 5th, and latest in Grand Rapids, Milwaukee and St. Louis (December 14th). Thirty-one of the 40 cities studied had attained the peak rate of mortality prior to November 2d. In the case of Milwaukee and St. Louis the maximum peak was the second peak, whereas in Grand Rapids it was the first peak that was so late. Sixty-five per cent. of the 40 cities showed two distinct peaks in the mortality curve, while 15 per cent. had one peak, and 8 or 20 per cent. had three peaks.

“It appears clearly that there was a definite tendency for the two-peak cities to fall into two groups in respect of the time elapsing between first and second peaks. About a third of them had the second mortality peak around eight weeks after the first peak. The remaining two-thirds had the second peak, on the average, about thirteen weeks after the first. The three-peak curves had the second peak on an average 7.1 ± 0.3 weeks after the first, and the third peak on an average 13.1 ± 0.3 weeks after the second. The cycle in the epidemic waves would therefore appear to be nearly a multiple of seven weeks rather than the ten weeks tentatively deduced from the dates of peaks. There the process of averaging obscured the true relations.”

Duration of explosive outbreak.—The range of the duration of the first outbreak of epidemic mortality is great, varying from five weeks in Richmond, Virginia, to twenty-three weeks in Atlanta, Georgia. Twenty of the cities, one-half the total number, showed a duration of ten weeks or less, while in the other half the duration was eleven weeks or more. The mean duration of epidemic mortality in the first outbreak was 11.90 ± 0.55 weeks. The ascending limb of mortality rate was rapid in nearly all cities. The descending limb was usually slower. In 34 of the 40 cities it required four weeks or less time for the mortality rate to pass from normal to its epidemic peak. But in only half as many (17) of the cities did the rate come down from its peak to normal again in a period of four weeks or less. The mean time from normal mortality rate to peak was 3.90 ± 0.21 weeks. The mean time from peak mortality rate to normal was 8.00 ± 0.50 weeks. Thus it took about twice as many weeks for the mortality curve to come back from its peak to normal, as were required for the increase from normal to peak at the beginning of the explosion. This is on the average. The ascending limb occupied about a month and the descending limb two months.

Pearl’s curves which have been copied in this report (Charts II to VII) enable us to follow his conclusions. Pearl offers a partial explanation for the variations in the different cities. There can be no doubt but what many factors play a role in the causation of these variations, and it is to be regretted that up to the present no statistics for smaller, more homogeneous communities have as yet been reported which could be compared with Pearl’s excellent work on the large cities of the country. Were his work supplemented by records from smaller towns in which the varying factors are less numerous, in which there is less occupational variation, additional conclusions could probably be reached. The unfortunate feature is that as a rule statistics from the smaller cities and towns are less reliable.

From a detailed mathematical study of influenza in 39 of our largest cities, done chiefly by the means of multiple correlation, with the hope of being able to explain the differences in the epidemic curves of weekly mortality in the various cities, Pearl concludes as follows:

“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 between the several cities in respect to explosiveness of the outbreak of epidemic mortality in which we are interested. Significantly causal or differentiating factors must be sought elsewhere.”

CHART II.