An Intensive Study of the Spread of Influenza in Small Groups of Closely Associated Individuals.

We have been discussing the disease under consideration chiefly from the viewpoint of the statistician. The statistician, possessing a wealth of information of a general nature, and from all parts of the world, and armed with fascinating complex mathematical instruments, is able to dissect the information at his disposal, and to reconstruct therefrom both facts and apparent facts of absorbing interest. To him we are indebted for the bulk of our knowledge of the distribution and spread of the influenza through metropolises, through countries, and through continents.

Like the aviator flying over enemy territory he acquires a breadth of vision and a general perspective which is to a great extent denied to those remaining on the ground. But also like the aviator, from the very fact of his high position he loses the ability to recognize detail. The small subdivisions in the enemy lines are slightly blurred and he can distinguish the front line trenches in which most of the action is occurring no more clearly than the reserve and support trenches. An Army depending entirely upon its aeroplane reconaissance would find itself helpless in combating the enemy. The aeroplane is useful, yes, it could not be dispensed with, but never an opportunity is lost for scouting parties to explore the enemy front lines; it is these latter who bring back information as to the local strength and distribution of the enemy, as to what particular divisions are in action, as to the strength of the entrenchments, the enemy distribution within the trenches, and finally what is most important, information as to the weak points in the enemy’s lines, places at which we may concentrate our attack with the hope of driving the enemy from its stronghold. Occasionally a raiding party will return with a prisoner. He will be examined thoroughly and may yield some valuable information. All such prisoners are not dressed alike. We recognize that some belong to one regiment and some to another. After we have extracted what information we can from the prisoner he is incarcerated, if we may extend our metaphor, in a test tube, and there he remains for future reference. We do not believe that these individual private soldiers are the cause of the war, but we do know that they are doing their share of the killing—that they are doing most of the killing.

It is characteristic of human enthusiasm and hopefulness that each raiding party prays that it may bring back with it a general officer, a field marshall, the one who is chiefly responsible for the enemy offensive. In our own little battle with our invisible host we have long since discovered that field marshalls here as elsewhere are difficult to discover by raiding parties. But the raids are and should be made just the same. They almost invariably bring back some new item of information, and it is the experience of many wars that even though the commanding general be never captured, repeated small or large attacks following preliminary reconaissance, if diligently and valiantly prosecuted under good leadership may win the war.

In studying the life and habits of the influenza virus and its army of secondary invaders, and the results thereof in small groups composed of individuals as the unit, instead of large groups with communities as a unit, we will be able to discover a certain number of additional facts, some of which may have considerable value.

In the study by the author of six selected districts in Boston a special study was made of the occurrence and manner of spread of the influenza in the household or family as a unit. The 10,000 individuals canvassed were distributed through 2,117 families. Of these two thousand odd families, 45.44 per cent. were visited with one or more cases of the disease in the 1918–19 epidemic, and 27.25 per cent. in the winter of 1920. Of these, 14.31 per cent. had cases in both epidemics. In either one or both of the two epidemics under consideration, 58.38 per cent. of all families had influenza (see Table III).

TABLE III.
Per cent. of families invaded by influenza.
Boston District No. 1918–19. 1920. 1918–19 and 1920. Total.
1 49.59 32.79 20.05 62.33
2 36.04 17.36 7.25 46.15
3 45.89 26.43 14.71 57.61
4 48.48 32.20 14.39 66.29
5 52.48 34.11 19.53 67.06
6 43.16 24.21 11.23 56.14
All Districts 45.44 27.25 14.31 58.38
Explanatory note:45.44 per cent. of all families were invaded in 1918.
27.25 per cent. of all families were invaded in 1920.
14.31 per cent. of all families were invaded in both epidemics.
58.38 per cent. of all families were invaded in one or the other or both.
41.62 per cent. of all families remained free from influenza throughout both epidemics.

In this discussion of family incidence, as in our work on the incidence among individuals, the question naturally arises as to the reliability of our information and the accuracy of our results. We have shown the close correspondence between our own results and those of Frost, done on a vastly larger number of individuals. The information for families was obtained from the same sources and from the same individuals. The thoroughness with which the inspectors did their work is indicated by the fact that in addition to the 2,117 families on which we base our results, only the records of 194 families have been discarded for various reasons. In discarding the family records we also discarded the individual records and such are, therefore, for individuals above our total of 10,000. One hundred and fifty-four of these were for families whose homes were in the districts surveyed, but who were not at home at the time of the first survey. These were omitted during the second survey, irrespective of whether individuals were at home. In this group are also included a few in which children were at home, but were unable to give reliable information. Fifteen of the 194 families gave insufficient information, and 25 refused to co-operate. The small number in this last group speaks well for the efficiency and methods of the inspectors. All families accepted for tabulation co-operated to the best of their ability, and we believe that the records are as accurate as this type of work may be made.

Dr. Niven, in the work referred to by Carnwath, made an inquiry covering 1,021 houses, with a population of 4,721. Five hundred and three households or almost exactly one-half, were invaded in either the summer 1918, or the autumn-winter 1918 epidemic. This proportion of families is quite similar to our own, but it must be pointed out that Niven was not studying the same two epidemics that we are discussing. Two hundred and sixty-six of his total households, or 26.05 per cent. were invaded in the autumn epidemic.

Previous to the present time the author has been unable to find records of investigators having used this method of studying influenza to any appreciable extent. Certainly there has been nothing done on the subject previous to the last pandemic. Since then Frost has studied, as indicated in his report, family incidence to the extent of determining the relationship to overcrowding and to economic status, and Niven has studied family incidence with special reference to immunity.

Thomas Sydenham, speaking of the epidemic of 1675, says that: “No one escaped them whatever might be his age or temperament, and they ran through whole families at once.”

According to Waldschmidt, during the epidemic of 1712, in Kiel, ten or more persons were frequently taken ill in one house.

In 1732, Huxam tells us that, “not a house was free from it, the beggar’s hut and the nobleman’s palace were alike subject to its attack, scarce a person escaping either in town or country, old and young, strong and infirm, shared the same fate.”

Metzger says that the influenza was so universal in March, 1782, that in very many houses all of the inmates were attacked. On the other hand, Mertens did not believe the influenza a contagion during the same epidemic for the reason that according to his observations now only one, and again all, of the members of a family, were stricken.

In 1833, in Königsberg, according to Hufeland, parents, children, and servants were frequently smitten with the disease at the same time, so that strange help had to be obtained for the family.

Parkes taught that, “Persons in overcrowded habitations have, particularly in some epidemics, especially suffered, and several instances are on record of a large school or a barrack for soldiers being first attacked, and of the disease prevailing there for some days before it began to prevail in the town around. Sometimes, on the other hand, schools and prisons have escaped. A low, damp, ill-ventilated and unhealthy situation appears to predispose to it, and in some instances, in hospital patients, it has assumed a malignant character. In other cases again, hospital patients have escaped; for example, the old people in the Salpêtrière in 1837, when the younger attendants were attacked.”

Effect of overcrowding.—The family or household forms a social unit in which human intercourse is very close, and in which the opportunities for contact infection either direct or indirect are manifold. In addition to all of the opportunities which each individual has for contracting the disease outside of the family every case in the family exposes every other member many times during the day. One of the first questions arising in a study of the disease in the family is, therefore, whether the size of the family in and of itself exerts any predisposing influence on the total incidence in any one family. Are large families more likely to have a greater percentage of cases than small families? We have endeavored to answer this question by grouping together all families containing only one individual, all of those with two, three, four, etc., and determining the percentage of individuals contracting influenza in each of the groups. The standard for comparison is the percentage of the total 10,000 who contracted the disease in either year, or in both. 19.71 per cent. of all persons canvassed contracted influenza in 1918–19. Reference to Table IV shows that of persons living in families of one, 17.95 per cent. developed the disease; of those in families of two, 18.46 per cent.; in families of three, 19.96 per cent.; in families of four, 20.10 per cent.; and in families of from five to seven, between 22 and 23 per cent. Families of over seven all showed lower, but varying incidence of the disease. As is seen by the table, they comprise only a small number of families.

TABLE IV.
The incidence of influenza in families of different sizes.
(Influence of size of family).
No. of individuals in family. No. of such families. Total No. of individuals included in all such families. Number of these individuals who developed influenza.
1918. 1920. Total.
No. Per cent. No. Per cent. No. Per cent.
1 39 39 7 17.95 3 7.69 10 24.42
2 260 520 96 18.46 55 10.58 151 29.04
3 359 1077 215 19.96 128 11.88 343 31.85
4 396 1584 319 20.10 169 10.67 488 30.81
5 375 1875 423 22.56 203 10.83 626 33.39
6 264 1584 361 22.79 151 9.53 512 32.32
7 179 1253 279 22.27 109 8.70 388 30.96
8 103 824 156 18.93 55 6.67 211 25.61
9 57 513 85 16.57 21 4.09 106 20.66
10 28 280 40 14.14 26 9.29 66 23.57
11 15 165 10 6.06 7 4.24 17 10.30
12 4 48 0 0.0 5 10.42 5 10.42
13 2 26 5 19.23 3 11.54 8 30.77
14 1 14 0 0.0 0 0.0 0 0.0

In 1920, 9.55 per cent. of the entire canvassed population contracted the disease. The table shows that 7.69 per cent. of all individuals in families of one contracted influenza, and between 10 and 12 per cent. in families of from two to five individuals. Above the family of five the incidence rates again are lower and varying within wide limits. The last column shows the percentage of individuals by size of family contracting the disease in either or both epidemics.

The average size of all families was 4.7 individuals.

If we consider only those family groups having over 1,000 individuals as being sufficiently large to be representative, we may conclude that families of from three to seven individuals show no progressive increase in influenza incidence with increase in size of the family. But all the available evidence indicates that other things being equal, the age incidence is a very important factor. Its influence will be felt in the subject under consideration, and it will modify the results. Thus, families of one or two are almost invariably adults; families of three are very frequently made up of two adults and a child or infant, while families of from five to seven will be more likely to have a high proportion of young adults—the age period more seriously affected.

The next question arising is whether those families, large or small, which are living in crowded circumstances, are more likely to develop the disease. Arbitrary standards must be chosen as indices of crowding. We have chosen two in order that they may check each other. The first is based upon the number of individuals sleeping in a bedroom. Families are classified as follows: Maximum sleeping in a single bedroom, 1; maximum sleeping in a single bedroom, 2; maximum per bedroom, 3, 4, etc.

The second standard of crowding is based upon the ratio of the number of individuals in the family and the number of rooms occupied. One person living in one room is not crowded; two in two rooms, three in three rooms, four in four rooms, eight in eight rooms, twelve in twelve rooms, are not crowded. Two people living in one room four in two rooms, six in three rooms, twelve in six rooms, are decidedly more crowded. On the contrary, one individual in two rooms, two in four, three in six, four in eight, five in ten, etc. have an unusual amount of room.

The ratios P⁄R are then throughout, ¹⁄₁, ²⁄₁, ½. These are used as dividing lines. All families with ratios higher than ²⁄₁ are classed as very crowded. Families with ratios above ¹⁄₁ up to and including ²⁄₁ are classed as crowded. Families with ratios above ½ up to and including ¹⁄₁ are classed as roomy, and those with ratio of ½ or lower are classified as very roomy.

Classifying all families in all six districts according to these last four degrees of crowding, we find, as is shown by Table V, that there is a progressive increase in the proportion of families with one or more cases of the disease, with increase in the extent of crowding.

According to the standard first described we find as is shown in Table VI that families with three, four and five individuals sleeping in a single room show a progressive increase of incidence over those families with but one or two per bedroom. This again is shown best in the total for all families, but is borne out in a study of each district. These statistics are however of little value for the study of the effect of overcrowding, because crowded families are usually large families. With an influenza incidence of 20 per cent. we would theoretically expect every family of five or larger to have one or more cases. This would amount to 100 per cent. infected families and such a state would not only influence, but dominate the statistics regarding overcrowding.

TABLE V.
Effect of crowding on development of influenza in families.
(A higher proportion of crowded households than roomy are invaded).
(Standard used: ratio of number individuals to number rooms).
Living conditions. No. of such families. Proportion of these families visited by influenza.
In 1918–19. In 1920. In both epidemics (Recurrent). Total families invaded.
No. Per cent. No. Per cent. No. Per cent. No. Per cent.
District I.
V. Cr. 53 30 56.61 15 28.31 12 22.64 31 58.49
Cr. 195 107 54.87 59 30.26 43 22.05 123 63.08
R. 79 36 45.57 24 30.38 18 22.78 42 53.16
V. R. 16 7 43.75 1 6.6 0 0.0 8 50.00
District II.
V. Cr. 4 1 25.00 1 25.00 1 25.00 1 25.00
Cr. 137 70 51.09 31 22.63 2 8.76 89 64.96
R. 208 70 33.65 39 18.75 7 8.17 92 44.23
V. R. 103 20 19.42 7 6.80 2 1.94 25 24.27
District III.
V. Cr. 13 9 69.23 2 15.38 1 7.69 10 76.92
Cr. 213 99 46.48 65 30.52 40 18.78 124 58.22
R. 143 62 43.36 35 24.48 15 10.49 82 57.34
V. R. 21 8 27.59 2 6.89 2 6.89 8 38.09
District IV.
V. Cr. 0 0 0 0 0
Cr. 27 18 66.67 8 29.63 5 18.52 21 77.77
R. 137 72 52.55 50 36.49 21 15.33 101 73.72
V. R. 95 38 40.00 27 28.42 12 12.63 53 55.79
District V.
V. Cr. 6 2 33.33 4 66.67 2 33.33 4 66.67
Cr. 110 67 60.91 37 33.64 25 22.73 79 71.82
R. 209 104 49.76 70 33.49 38 18.18 146 69.86
V. R. 14 3 21.42 3 21.42 0 6 42.84
District VI.
V. Cr. 0 0 0 0 0
Cr. 2 1 50.00 0 0.0 0 0.0 1 50.00
R. 92 57 61.96 23 25.00 14 15.22 66 71.74
V. R. 189 65 34.39 46 24.34 19 10.05 92 48.68
Living conditions.No. of families.No. 1918.Per cent. 1918.No. 1920.Per cent. 1920.No. both.Per cent. both.Total.Per cent.
Very crowded804353.752531.251822.505062.50
Crowded69337253.6820129.0012618.1844764.50
Roomy86539445.5524428.2112514.4551359.31
Very Roomy44314332.288719.64368.1319443.79
AllTotal1918Per cent.1920Per cent.BothPer cent.TotalPer cent.
208195245.7555726.7730514.66120457.86
TABLE VI.
Effect of crowding.
(Standard used: maximum number sleeping in one bed room.)
Maximum No. sleeping per room. No. of such families. Proportion of these families with cases of influenza.
In 1918–19. In 1920. In both epidemics. Total families invaded.
No. Per cent. No. Per cent. No. Per cent. No. Per cent.
District I.
1 16 6 37.50 4 25.00 3 18.75 7 93.75
2 93 52 55.91 31 33.33 20 21.51 63 67.74
3 145 65 44.83 47 32.41 27 18.62 85 58.62
4 79 43 54.43 25 31.65 17 21.52 51 64.56
5 24 11 45.83 11 45.83 6 25.00 16 66.67
6 10 3 30.00 3 30.00 1 10.00 5 50.00
District II.
1 90 15 16.67 7 7.77 2 2.22 20 22.22
2 211 68 32.23 36 17.06 14 6.64 90 42.65
3 115 59 51.30 23 20.00 10 8.69 72 66.61
4 33 20 60.60 11 33.33 6 18.18 25 75.76
5 3 1 33.33 2 66.67 1 33.33 2 66.67
6 0 0 0 0 0
District III.
1 26 10 38.46 3 11.54 2 7.69 11 42.31
2 179 73 40.78 47 26.26 23 12.85 97 54.19
3 145 72 49.66 37 25.52 23 15.86 86 59.31
4 39 20 51.28 15 38.46 8 20.51 27 69.23
5 8 5 62.50 2 25.00 1 12.50 6 75.00
6 0 0 0 0 0
District IV.
1 53 15 28.30 15 28.30 6 11.32 24 45.28
2 165 80 48.48 56 33.94 22 13.33 114 69.09
3 42 29 69.05 15 35.71 10 23.81 34 80.95
4 5 4 80.00 0 0.0 0 0.0 4 80.00
5 0 0 0 0.0 0 0
6 0 0 0 0 0
District V.
1 23 8 34.77 6 26.08 1 4.35 13 56.52
2 156 70 44.37 48 30.77 24 15.38 94 60.26
3 130 81 62.31 44 33.84 27 20.77 98 75.38
4 27 18 66.66 14 51.85 12 44.44 20 74.07
5 6 3 50.00 4 66.67 3 50.00 4 66.67
6 1 0 0.00 0 0.00 0 0.00 0 0.00
District VI.
1 120 42 35.00 24 20.00 10 8.33 56 46.67
2 146 77 52.74 34 23.29 22 15.07 89 60.96
3 10 5 50.00 5 50.00 1 10.00 6 60.00
4 0 0 0 0 0
5 0 0 0 0 0
6 0 0 0 0 0
Total
1 328 96 29.27 59 17.99 24 7.32 131 39.94
2 450 420 44.21 252 26.53 125 13.16 547 57.57
3 587 311 52.98 171 29.13 98 16.69 381 64.91
4 183 105 57.38 65 35.52 43 23.50 127 69.39
5 41 20 48.78 19 46.34 11 26.83 28 68.29
6 11 3 27.27 3 27.27 1 9.09 5 45.45

An objection will be raised, and justly so, that we have up to this point been studying influenza in families irrespective of how many cases there are in each family. Until now the family with one case was classified exactly the same as the family with eight cases. In the following classification we have taken first all families with a maximum of one sleeping in one room, and sub-divided these into families with no influenza, those with one case, two cases, etc. We have likewise classified families with maxima from two to six per bedroom. For the sake of brevity we will consider only the last column of Table VII, influenza incidence among the individuals of the various classes of families for both epidemics. Study of the table will show a correspondence in the other columns. Solitary cases were more numerous in families with but one or two per bedroom (27 per cent.) and less frequent in families with three, four and five per bedroom, (23 per cent., 18 per cent., and 20 per cent., respectively). The families of six per bedroom form such a small group that here again they should not be considered. Multiple cases become progressively more numerous as the number of individuals per bedroom increases (14 per cent. in families of one per bedroom, 29 per cent. in two per bedroom, 41 per cent. in three, 51 to 52 per cent. in four, and 45 per cent. in five). Fifty-eight per cent. of families with a maximum of one per bedroom, 43 per cent. with two per bedroom, 35 per cent. with three, 31 per cent. with four and 35 per cent. with five had no influenza at all.

But here again, the fact that crowded families are usually large families interferes with drawing any conclusions. A family with four per bed room would generally be larger than one with two per bed room.

Frost observed that, considering the ratio of incidence in total white populations irrespective of housing as 100, and after adjusting all groups to a uniform sex and age distribution, the ratio where there were more than 1.5 rooms per person was 77, from 1 to 1.5 rooms per person the ratio was 94, and for individuals averaging less than one room per person it was 117. The attack rate showed a consistent increase as the number of rooms per person decreased.

Woolley observed, “Housing, if one includes in the term overcrowding, has surely been an important factor in spreading the epidemic. Whether it has had any appreciable effect upon the incidence of complications is a question. The epidemic has certainly gone faster and was over sooner because of the crowding; the hospital was filled sooner than it should have been as a result of the rapidity of spread of the disease, and overcrowding of the hospital occurred when with a less rapid spread it would not have occurred; but whether the number of fatalities or the number of pneumonias was greater than they should have been with less crowded conditions may be doubted.”

TABLE VII.
Relationship between crowding and number of cases in the family.
(Influenza appeared more frequently in crowded households and such families more frequently had multiple cases.)
Families with maximum per bed room of one.
(58.23 per cent. of these had no influenza.)
Cases developing in family. Total such families. Invaded in 1918–19. Invaded in 1920. Invaded in both epidemics. Total families invaded. Two or more cases.
No. Per cent. No. Per cent. No. Per cent. No. Per cent. Per cent.
1 85 55 17.68 37 11.89 7 2.25 311 27.33
2 32 26 8.36 16 5.14 10 3.22 10.28 } 14.44
3 9 8 2.57 2 0.64 1 0.32 2.88
4 4 3 0.96 4 1.28 3 0.96 1.28
5 0 0
6 0
7 0
8 0
2 per bed room.
(43.35 per cent. of these had no influenza.)
1 254 169 18.27 112 12.11 27 2.92 925 27.46
2 135 112 12.11 64 6.92 41 4.43 14.59 } 29.17
3 79 65 7.03 38 4.11 24 2.59 8.54
4 40 35 3.78 16 1.73 11 1.18 4.32
5 11 9 0.97 8 0.81 6 0.64 1.18
6 3 3 0.32 0 0.0 0 0.0 0.32
7 2 2 0.22 1 0.11 1 0.11 0.22
8 0 0
3 per bed room.
(35.34 per cent. of these had no influenza.)
1 136 104 17.84 50 8.58 18 3.08 583 23.33
2 103 77 13.21 55 9.43 29 4.97 17.67 } 41.33
3 59 51 8.75 29 4.97 21 3.60 10.12
4 43 40 6.86 16 2.76 13 2.23 7.37
5 22 22 3.77 9 1.54 9 1.54 3.77
6 12 12 2.06 5 0.86 5 0.86 2.06
7 2 2 0.34 0 0 0.34
8 0
4 per bed room.
(30.79 per cent. of these had no influenza.)
1 31 24 13.64 10 5.68 3 1.70 176 17.61
2 22 19 10.80 13 7.39 10 5.68 12.50 } 51.60
3 37 32 18.18 25 14.20 20 11.36 20.92
4 14 11 6.25 6 3.41 3 1.70 9.09
5 9 9 5.12 4 2.27 4 2.27 5.12
6 4 3 1.70 2 1.19 1 0.59 2.27
7 3 3 1.70 0 0 1.70
8 0
5 per bed room.
(35 per cent. had none.)
1 8 6 15.00 4 10.00 2 5.00 40 20.00
2 6 5 12.50 4 10.00 3 7.25 15.00 } 45.00
3 3 2 5.00 2 5.00 1 2.50 7.50
4 2 1 2.50 2 5.00 1 2.50 5.00
5 4 2 5.00 3 7.50 1 2.50 10.00
6 2 2 5.00 2 5.00 2 5.00 5.00
7 0
8 1 0 0.0 1 2.50 0 0.0 2.50
6 per bed room.
(50 per cent. had none.)
1 2 1 10.00 1 10.00 0 0.0 10 20.00
2 1 1 10.00 1 10.00 1 10.00 10.00 } 30.00
3 2 1 10.00 1 10.00 0 0.0 20.00
4 0
5 0
6 0
7 0
8 0

The housing methods in the cantonments and even in the tent camps resulted in a degree of congestion and close physical contact among individuals that was attained in no civil communities with the possible exception of some institutions. In cantonments the number of men in individual rooms ranged from 30 to 100 and even under the best circumstances there was very evident close crowding. An individual in any of these large rooms contracting a contagious disease had opportunities to spread it by contact and by droplet infection not only to one or two others, as in the case of the average family, but to a large group of the men in the same room. A vicious circle was thus formed which tended to propagate the disease throughout any camp with utmost rapidity. Brewer has compared the influenza incidence rate in the principal white organizations at Camp Humphreys with the floor space allowed each man in the respective organizations, and concludes that, “It is not proper or just to attribute the differences shown, alone to the amount of floor space allowed each organization, but it certainly points very strongly to the fact that the incidence of the disease varied with the density of the population, although not with mathematical regularity.” Brewer cites regiments which although housed alike showed definite variation in the influenza incidence. This merely shows that other factors also play a part. Thus, in one instance, the difference in the two regiments was in length of service. Brewer also found that among the white troops the incidence of pneumonia appears to vary with the density of the population.

V. C. Vaughan has reported on the relationship between incidence in tents and in barracks at Camp Custer. From this one observation it would appear that the incidence is little changed under the two conditions.

“During September and October, 1918, a study was made on the relationship, if any, of influenza to methods of living. Of the command, 3,633 were in tents. The morbidity per thousand in these was 129. There were in barracks 36,055. The morbidity per thousand among those was 275. At first glance the lower morbidity of those in tents is striking, but going further into the matter it was found that the entire morbidity of the Quartermaster Corps was very low. Of the Depot Brigade 2,881 were in tents, with a morbidity of 128 per thousand, while 3,824 were in barracks, with a morbidity of 134 per thousand.”

Howard and Love offer three reasons why during the last four months of 1918 the deaths from influenza and pneumonia in the Army in the United States ran at a rate nearly three times as high as that among our troops in France: First, that the troops in the United States were recent recruits and therefore more susceptible to disease; second, that probably many of the troops in France who had seen much longer service had had the disease in mild form in the early spring; and, third, that the method of housing was entirely different in France. There the men were spread over a wide territory and whenever in rest area they were billeted in houses rather than crowded into barracks. Furthermore, they were living much more in the open. It was found that in commands of the Service of Supply, where troops were housed in barracks with a large number of men to a single room, the epidemic ran much the same course with high mortality, as it did in the cantonments in the United States. The percentage of infection and the fatalities from influenza and pneumonia in France were much greater among troops of the S. O. S. than among troops at the front.

Domestic cleanliness.—We have studied the relationship between influenza incidence and the cleanliness of the household by the same method used in studying overcrowding. In Table VIII we have classified according to cleanliness and according to the number of cases developing in each family. We have had four subdivisions, “very clean,” “clean,” “dirty,” and “very dirty.” There is greater opportunity for erroneous results in this table than in the one preceding because the standards of cleanliness are difficult to define. As a matter of fact we are guided entirely by the inspector’s own impression of each household, as she examined it during her visits. The following is an excerpt from the instructions given each inspector on this subject:

“A few words on this subject may describe much. State of cleanliness of the individual, slovenly condition, dust and dirt, foulness of air noticed on first entering, condition of children, of kitchen sink, etc., should be noticed, and good or bad features recorded. In the poorer districts not a few families will be found in which the cleanliness, considering the surroundings, is quite laudable. Of particular importance are amount of daylight, ventilation, care of bathroom and toilet, garbage, whether windows are kept open at night.”

On the basis of these returns we have classified the families as indicated, but each inspector was governed to a certain extent by the average cleanliness of her district, and it is difficult to compare the cleanest tenement with any of the districts of well-to-do individuals. We will therefore probably find it more profitable and more nearly accurate to combine the groups and classify them only as “clean” and “dirty.”

TABLE VIII.
Relationship between cleanliness and number of cases in family.
(Clean families were invaded less frequently and had solitary cases more often than did dirty households.)
Very clean.
(47.62 per cent. had none.)
Cases in families. Total families. ’18. Per cent. ’20. Per cent. Both. Per cent. Total. Per cent.
1 124 72 15.65 50 10.87 8 1.74 460 26.96
2 53 41 8.91 27 5.87 15 3.25 11.52 } 25.42
3 37 33 7.17 13 2.82 9 1.95 8.04
4 18 16 3.48 8 1.74 6 1.30 3.91
5 4 3 0.65 2 0.43 1 0.21 0.87
6 3 3 0.65 0 0.0 0 0.0 0.65
7 2 2 0.43 1 0.21 1 0.21 0.43
8 0 0 0 0
Clean.
(41.52 per cent. had none.)
Cases. Families. ’18. Per cent. ’20. Per cent. Both. Per cent. Total. Per cent.
1 301 212 18.45 120 10.44 31 2.70 1149 26.19
2 177 143 12.45 91 7.92 57 4.96 15.40 } 32.29
3 101 83 7.22 52 4.53 34 2.96 8.79
4 52 47 4.09 20 1.74 15 1.26 4.53
5 30 29 2.52 17 1.48 16 1.22 2.61
6 8 7 0.61 3 0.26 2 0.17 0.70
7 3 3 0.26 0 0.0 0 0.0 0.26
8 0 0 0 0
Dirty.
(36.89 per cent. had none.)
Cases in families. Total families. ’18. Per cent. ’20. Per cent. Both. Per cent. Total. Per cent.
1 79 59 17.40 36 10.62 16 4.72 339 23.30
2 58 48 14.16 29 8.55 19 5.61 17.11 } 39.81
3 37 31 9.14 22 6.49 17 5.01 10.91
4 26 22 6.49 12 3.54 8 2.36 7.67
5 6 5 1.79 4 1.18 3 0.94 1.77
6 7 7 2.06 4 1.18 4 1.18 2.06
7 0 0 0 0
8 1 0 0.0 1 0.29 0 0.0 0.29
Very dirty.
(39.26 per cent. had none.)
1 22 16 14.95 8 7.47 2 1.85 107 20.56
2 11 8 7.47 6 5.10 3 2.80 10.28 } 40.18
3 14 12 11.21 10 9.35 7 6.54 13.08
4 7 5 4.67 4 3.73 2 1.85 6.54
5 6 6 5.10 1 0.93 0 0.0 5.61
6 3 3 2.80 2 1.85 2 1.85 2.80
7 2 2 1.85 0 0.0 0 0.0 1.87
8 0 0 0 0

But even without combining in this way, the table shows us that for both years 27 per cent. of the very clean families, 26 per cent. of the clean, 23 per cent. of the dirty and 21 per cent. of the very dirty, had but one case, while 25 per cent. of the very clean, 32 per cent. of the clean, 40 per cent. of the dirty, and 40 per cent. of the very dirty, had multiple cases.

The cleaner the family the less is the likelihood of multiple cases.

It is rather difficult to find concrete examples of the influence of domestic habits and environment in the 1918 pandemic. The remarkably high incidence among the natives of India and among the American Indians might by some be attributed to unfavorable environment. Lynch and Cumming obtained records from a large number of institutions and from business concerns having their own records, and discovered that the influenza incidence was higher in those institutions where dish washing was done manually than in those in which mechanical washing was performed. They appear to conclude that the difference in the two methods of washing dishes was the cause for the greater incidence in influenza, thus bearing out their theory of the propagation of influenza chiefly through eating utensils. On the contrary it is possible that the presence of the mechanical washer is an indication of advanced methods, greater care in the kitchen, and better hygiene probably not only in the kitchen and dining room, but throughout the institution.

Economic status.—Although in our survey information has been obtained regarding the economic status of the various families we would not stress this phase of our subject. Obviously the amount of money an individual has in his bank will not directly influence the amount of influenza he will have in his home. As nearly an accurate classification by wealth is by the separation into the districts, Districts I and III being very poor, District II poor, Districts IV and V moderate, and VI well-to-do. From Chart XXVI we see no definite relationship between influenza incidence and economic status.

Dr. Niven has had similar experiences. He remarks that the disease does not appear to have affected especially any class or section of the community. Rich and poor suffered alike. Inquiry in some towns shows that the epidemic not infrequently started in the well-to-do districts and only later involved the poorer and less prosperous areas.

We cannot state with any degree of accuracy in what section of Boston the 1920 recurrence first began. The sections studied are for relatively small portions of the city, and it is possible or probable that the original increase was in some area outside of our districts. In the districts studied the earliest increase in reported cases was from the section of the city known as Dorchester (Districts IV and V), where there was some increase in December, 1919. The latest definite increase was in the Irish district of South Boston. Geographically these two areas are quite near. The relative insusceptibility of the Irish population is probably a much more important factor in the difference.

Frost found after classifying the white population canvassed in Little Rock and San Antonio according to economic status, and adjusting the incidence rate in each group to a uniform sex and age distribution, that the ratios of incidence in each economic group to incidence in total white population did show an increase with increasing poverty. “Notwithstanding that the classification according to economic status is a very loose one, based solely on the judgment of inspectors with widely different standards, a considerably higher incidence is shown in the lower as compared to the higher economic group.”

Parsons, in 1891, discussed the influence of poverty, but believed that it is the concomitants of poverty which were the cause of the higher incidence among the poor.

“Sanitary conditions do not seem to have had any influence in determining the occurrence of influenza, and what share they have had in determining its extent or fatality cannot yet be decided. On the occasion of the last great epidemic, Dr. Peacock concluded, ‘The more common predisponents to disease, e.g., defective drainage, want of cleanliness, overcrowding, impure air, deficient clothing, innutritious or too scanty food, powerfully conduce to the prevalence and fatality of influenza.’ And Dr. Farr showed that in the last six weeks of 1847, while in the least unhealthy districts of London the annual rate of mortality was raised from a mean rate of twenty per 1,000 to thirty-eight, in the unhealthiest districts it was raised from a mean rate of twenty-seven to sixty-one.

“That overcrowding and impure air must have a powerful influence in aiding the development of the epidemic follows from what we have seen of its greater prevalence among persons associated together in a confined space; and though rich and poor have alike been sufferers from the epidemic, and even royal personages have been fatally attacked by it, it cannot be doubted that poverty must have in many cases conduced to a fatal issue in persons, who, if placed under more favorable circumstances, might have recovered, seeing that it often involves not only inferior conditions of lodgment, but also want of appropriate food, of sufficient warmth and clothing, and of ability to take the needed rest.”

Distribution of the disease through the household.—During the autumn and winter epidemic of 1918 there was considerable discussion, and particularly were there popular newspaper reports of entire families being taken ill with influenza, sometimes all on the same day. This was less true of 1920. But few of us are personally acquainted with such instances and at best they must have been relatively rare.

Among 1,236 families with influenza in either epidemic we found only 94 or 7.6 per cent. in which the entire family contracted the disease. No family consisting of over seven individuals was reported as having all the members of the family sick in either epidemic. Over two-thirds of the families with even numbers of individuals (464 out of 605) suffered the illness of less than half of the household. One quarter of all families of more than one (539 out of 2,107) had but one case per family. Over a third of all families of over two individuals (745 out of 2,006) had two or less cases per household. As a rule there were at least one and usually several individuals in each household who did not contract influenza.

That as a rule the disease did not appear explosively in a family; but that cases developed successively, is indicated by the fact that out of 577 families contracting influenza in the epidemic of 1920 the cases were all of simultaneous development in but fifteen. In thirteen of these, two individuals fell ill on the same day and no subsequent cases developed. In the other two families three individuals came down on the first day and no other cases developed. In addition there were, out of the 577 families, fourteen in which there were two or more cases developing on the first day of the invasion, but which were followed on subsequent days by later cases in the same family. Again, there were eleven families in which two or more cases occurred simultaneously at an interval of one or more days after the development of a single prior case.

We may say that as a rule in the 1920 epidemic, cases of influenza developed in families successively and not simultaneously. In only 29, or 5 per cent. of the families contracting the disease in 1920, did more than one case develop on the first day of the appearance of the disease in the family.

A certain difficulty in determining the date of onset is that we must rely upon the patient’s statement. One individual may have been sick for hours or days before a second member coming down with the disease called forth recognition of the fact that they both had it.

Unfortunately we are not able to give similar statistics for the 1918–19 epidemic. Our investigation occurred so long after the epidemic that specific dates of onset of the disease would have been entirely unreliable. The nearest date we have attempted to obtain was the month of the attack.

Dr. A. L. Mason states that 63 cases came under his observation in the epidemic of 1889 as occurring in groups in families. In but six instances were two persons attacked on the same day. The average interval between cases in the same household was four days. Sometimes a week or more elapsed. Whole families were never stricken at once.

Parsons in 1891 concluded from the results of questionaires sent to physicians that in the first spread, 1889–90, there was an interval between cases in individual households just as we have described. Among the replies to his questionaires nine described intervals of one day and under, six described intervals of two days, three of three days, three of four days, and four replies described intervals of more than four days.

Leichtenstern observed likewise: “In large families the contagious character of influenza is evidenced by the fact that the other members of the family become sick one after the other following the first case. This rule of succession is most easily seen in the early or late period of an epidemic and is less noticeable at the height, where the opportunity for all the members of the family to acquire the influenza outside the home is enormous. This latter fact explains why, when all sicken at once, the disease appears to be miasmatic in origin. There are many examples where other members of a family living with a sick individual remained unaffected. Parsons reports such cases, and this was so frequently the case that some British physicians state that it is the rule that there is but one case in a family or that the cases are widely separated in time. This was only partly true during the period of the pandemic and was very frequent in the epidemic following it. In this respect influenza acts like the common contagious diseases, diphtheria, scarlet fever, measles, etc., while the difference lies in the short incubation period and the very high contagiousness of the disease.”

That West, in England, had observed the same phenomenon is indicated by the following quotation: “How is it, for instance, that one member of a household may be picked out and the others escape, though they are susceptible, as is shown by their acquiring the disease shortly after in some other way?”

Again Leichtenstern wrote: “It is noteworthy that influenza on ships usually did not occur explosively, but spread gradually, and on ships usually lasted several weeks, as on the Bellerophon, from the 27th of March to the 30th of April; on the Canada from the 11th of April to the 24th of May; on the Comus from the 10th of April to the 3d of May.

“The German Marine Report states, ‘Everywhere on the ships the disease began not suddenly but gradually.’ The frigate Schwalbe first had a large number of cases only on the 6th day after the beginning of the epidemic. There are, however, some exceptions, where the disease has begun suddenly with the greatest violence on ships as on land. Such was true of the frigate Stag which on the 3d of April, 1833, neared the influenza infected coast of Devonshire, and as it came under the land wind the epidemic suddenly broke out with great violence. Within two hours forty men took sick. Within six hours the number had increased to sixty. Within twenty-four hours 160 men were sick. As Parkes has remarked the evidence is insufficient that there had been no communication with the coast. There have been other examples of sudden outbreaks on ships, as on a Dutch frigate in the harbor of Mangkassar, where 144 men out of 340 took sick in a few days (1856); on the Canopus (1837) in the harbor of Plymouth, where on the 15th of February three-fourths of the men took sick with influenza.”

Garvie, in reporting his personal experiences with influenza in 1918 in an industrial area in England, experiences not based on statistical study, concludes that there are two types of cases, the sporadic case which occurs mainly among the wage-earning members of the family and has little tendency to affect other members of the household, and second, the type of case where a large number of individuals in the household are affected. He called this the “household wave.” If we interpret him aright he really means that there are either single or multiple cases, and that the single cases are more apt to occur in the wage-earner, the individual who is more exposed on the outside of the household. He also believes that the household wave is more severe in character than the so-called sporadic case, and is accompanied by a greater number of complications.

Armstrong, in his survey in Framingham, examined influenza convalescents. He found that of these 10 per cent. were in families in which no other cases had developed, and 87 per cent. were in families where one or more additional cases had occurred. In three per cent. information was lacking.

It is important in studying the literature on this subject to distinguish between definitely established fact and less definite description. Thus one is still left in some doubt when one reads in a London letter in the Journal of the American Medical Association for 1915 concerning the epidemic in London at that time that, “whenever it has seized an individual it has usually run through the entire household. Whole offices have succumbed.”

The first case in the family.—Chart XXVII shows clearly that in both epidemics in our experience the wage-earner was much more frequently the first case in a family than was any other occupation. The individuals whose occupations kept them at home were second. Infants, as was to be expected, were recorded as being “first case” in the smallest number of instances.

In 1889 the distribution was practically the same. Parsons found that out of 125 households the first case was a bread-winner in 96; a housekeeper in nine; a child at school in thirteen; a child not at school in two families. In the last five families the first case was in adults, occupation not given. This order is identical with our own. Neither our own observations nor those of Parsons consider the relative proportions of wage earners in the population as a whole. The results are nevertheless suggestive.

H. F. Vaughan reached comparable results for the 1920 epidemic in Detroit. During the first few weeks the age groups from 20 to 29 showed a relatively much more frequent influenza incidence than did children up to ten years. In later weeks of the epidemic there was a relative increase in the incidence among children and decrease among young adults. He concluded that the disease first attacks the young adult and from this group it extends into the home.

In the Local Government Board Report for 1891, H. H. Murphy distinguishes three groups or ways in which the disease may be brought into the family. The examples will be found to be characteristic for any epidemic and for any country:

Group A.—Cases of single exposure.

“Household 1.—Mr. Q. goes to London daily. Was ill with influenza on December 25th. No other case in this house till January 15th.

“Household 2.—Mrs. A. called on Mr. Q. on December 31st, and had a few minutes’ conversation with him. She was taken ill on January 3d. There was a Christmas family gathering at this house, and this is how the other members were affected: Mr. B., January 6th; Miss C., Mrs. D., and Master D., January 8th; Mr. J., January 10th; Mr. H., January 11th.

CHART XXVII.

“Household 3.—Miss M. went to a party January 3d. She had a few minutes’ conversation with a young lady who said she was suffering from influenza. Miss M. had a characteristic attack on the 6th of January.

“Household 4.—Mr. G. goes to London daily; taken ill January 5th. Mrs. N. visited him for a short time on January 5th, and was taken ill January 10th.”

Group B.—Where disease was brought from a distance into a previously healthy household.

“Household 8.—Mrs. R. G., living in the north of London, came here on a visit December 17th. On the 19th she was taken ill with influenza, the first case that I knew of in this neighborhood. Mr. C. G., on the 23d, servant on the 26th, Mrs. G. 31st, and Mr. G. January 9th.

“Household 9.—Mr. I. lives at his business place in London, taken ill December 20th with influenza. His family reside here. Boy C. visited his father for a few days, and came back ill on January 4th. The other members of the family were attacked as follows: Baby, 8th; Mrs. and boy, 12th; boy, 18th, girl, 22d; girl 25th.

“Household 10.—Master K. stayed a few days with some friends in London. They had been ill with influenza. Returning home on December 31st he was taken ill. Four brothers and sisters ill on the 2d January, Mr. K. on 3d, child and two servants on the 5th, Mrs. K. on the 7th.”

Group C.—Where the source of infection could not be determined or was local.

“Household 28.—Mrs. D. (who thinks she got it shopping) was taken ill 2d January, her daughter on the 5th, and Mr. D. on 6th.

“Household 29.—Mrs. L. (who thinks she got it shopping), aged 80, had influenza badly in 1847; similar symptoms, but much milder, on January 6, 1890; Miss L. was attacked on the 10th, and servant on the 17th.

Note.—A former attack did not confer immunity after forty-three years.

“Household 30.—Mrs. B. (who thinks she got it out at work) taken ill 9th January, and her child on the 11th.”

In Murphy’s complete list, one of the most frequent remarks is, Mr. Blank goes to London daily. Or, Miss Blank, absent on a visit, was taken ill with influenza and returned home.

Again, in 1890, Dr. Bruce Low studied the development of influenza at East Keal, a town of 300 inhabitants. We quote in part his description:

“The following is believed to have been the commencement of the outbreak, and for these facts I am indebted to Dr. Francis Walker, Medical Officer of Health, Spilsby, R. S. D. Mrs. N., residing at East Keal Hall, went to London (Forest Hill) on a visit on November 11th. She visited Barnum’s Show on November 13th. She became ill on the night of the 14th. Her symptoms were those of a cold, attended with sore throat. No one else so far as she knew was ill in this way in the house before her. She left Forest Hill on November 16th, still feeling very unwell, and went to stay with friends at Kensington. She was too ill to return home till November 23d, at which time she was still feeling very weak. She heard from Forest Hill that, directly after she left one of the inmates of the house where she had been visiting fell ill with symptoms similar to her own. Within a few days, probably about the 27th of November (the exact date is not fixed), of her return home, her son, aged four, became unwell with what appeared to be an ordinary cold, but the child had epistaxis; he soon recovered, but during the next fortnight the four servants in the house were ill with what were said to be ‘colds,’ one of them also had epistaxis. On January 2d another son, aged six, was ill with ‘cold’ for a few days; he went out and had a relapse, which compelled him to stay in the house for another week.

“On January 3d, Mrs. N. again fell ill with ‘a bad cold,’ attended with headache, backache and epistaxis. She was in bed two days and felt miserable and prostrate for more than a week after. On January 5th, Mr. N., her husband, had headache, backache, and general soreness ‘all over.’ On January 10th, the boy, aged four, who was first attacked after his mother’s return from London, again became ill, his symptoms being the same as before. The only other remaining member of the family who had managed to escape an attack of ‘cold’ up to this date, was said to have felt ill the day the boy had his second attack; but the illness of this individual was slight, and only caused suffering for one night. Thus between the return of the mother on November 23d and January 10th all the inmates of this house, nine in number, had an attack of illness, evidently of the same nature. A boy who works in Mr. N.’s yard was taken ill with influenza about the end of November. He lives in the village. After his illness his four brothers also were ill. Dr. Walker says that ‘about the end of November’ cases of like illness were beginning to crop up in East Keal. Mrs. W., the wife of the village grocer and baker, who waited on customers in the shop and never left the shop or house, was taken ill on the afternoon of November 30th. Next morning, December 1st, her husband and six children were all attacked in the same way with what is now recognized to have been marked influenza. The only inmate of the house who escaped was a youth employed to deliver bread and groceries in a cart in the neighboring village.”

Leichtenstern relates that a physician traveling from Berlin on the 10th of December became sick in his home town, Elgesburg, on the 8th of December, but he made several visits and a few days later those people seen by him fell sick, while otherwise there were no cases of influenza in the town or its neighborhood. These cases would probably have fallen in Group C., of Murphy’s classification.

Intimacy of family contact.—We have been able to discover in a representative number of families in which influenza has occurred, not only what individuals slept in the various rooms of the household, but also what individuals slept in the same bed with influenza cases. We can, therefore, study for the 1920 epidemic three degrees of contact; contact by sleeping with a case of influenza; by sleeping in the same room but a different bed; and general contact by being in the family, but sleeping in another room. For brevity we designate these, “sleeping,” “room” and “family” contact. We have established similar information for 1918, after eliminating families in which deaths or births or other additions or losses had occurred during or subsequent to the 1918 pandemic, in which there has been a change of address, in which the cases are so widely separated that we have designated them unrelated, and finally, those families in which the information has been insufficient. With the remaining we have assumed that the distribution within the household has been the same in both epidemics. Statistics are available on 1,734 individuals who in 1918 were exposed to a prior case in the family. Of these, 462 developed influenza and 1,272 did not. 26.6 per cent. of exposed individuals in families contracted influenza, without respect to the degree of exposure.

Intimacy of contact.Number so exposed.Number infected.Per cent. infected.
“Sleeping”36016645.2
“Room”3035919.5
“Family”1,06427322.3

45.2 per cent. of individuals sleeping with cases of influenza in 1918 contracted the disease; 19.5 per cent. of those sleeping in the same room, but different beds did so; 22.3 per cent. of those living in the same family, but sleeping in other rooms contracted the disease.

Sleeping contact is more productive of influenza than are the less intimate forms.

Throughout this study the fact that there are multiple possible sources of infection both outside and often within the family complicates the picture.

The results for 1920 are similar. Here, 30.0 per cent. of all individuals sleeping with cases of influenza contracted the disease, 17.7 per cent. of room exposures contracted it, while but 11.5 per cent. of family exposures were attacked.

Four hundred and sixty-three or 29.1 per cent. of the total of 2,193 individuals exposed in 1920 had had the disease in the 1918 pandemic. Did they show by reason of any immunity a lower attack rate for the same degree of exposure than other individuals in 1920?

Type of exposure in 1920.Per cent. of exposed individuals who had had influenza in 1918–19 and who contracted it again, per cent.Per cent. of those who had not had a previous attack, and who on exposure contracted influenza, per cent.
“Sleeping”27.031.0
“Room”18.317.6
“Family”12.011.2

On the whole there is no evidence of protection afforded by a previous attack.—Individuals who had had the disease before succumbed to a second attack in the same proportion as those who had not previously had influenza.

Recurrent cases.—In certain families there were individuals who had had influenza during both the 1918 and 1920 epidemics. Were these recurrent cases the first ones to occur in the family, or did they, as a rule, follow other cases in the same household? We have records of 236 recurrent cases in which we know the order of occurrence of the various cases in the family. Out of this total number 57 were the initial cases in the household. One hundred and nineteen were the only cases occurring in the family. Therefore 176 or 74 per cent. of the total number of recurrent cases were either the first or the only cases in the family. Sixteen recurrent cases followed between other cases and 44 occurred as the last of a series of two or more in the household.