THE THIRD (1926) Milroy Lecture ON EXPERIMENTAL EPIDEMIOLOGY.

645 THIRD*(1926)

THE

Milroy Lecture ON

EXPERIMENTAL EPIDEMIOLOGY. Delivered before the Royal College of Physicians of London on

March 9th

BY W. W. C.

TOPLEY, M.D.CAMB., F.R.C.P. LOND.,

PROFESSOR OF BACTERIOLOGY, UNIVERSITY OF MANCHESTER.

IN the previous lectures it has been shown that it possible to initiate an experimental epidemic among mice, and to maintain its activity for an apparently unlimited period of time. Is it possible to modify the course of events in the opposite direction, to alter our experimental conditions in such a way that an epidemic fails to develop after we have brought together the parasites and their hosts, or to intervene

is

a wave of mortality has commenced its rise bring about its premature fall ?

after

and

THE EFFECT OF DISPERSAL. In the experiments on the initiation and maintenance of epidemics it was found that the numerical relation between infected and susceptible mice, and the rate of immigration of the latter, exerted a decisive influence on the mortality-rate. It was natural to inquire whether any form of redistribution of the population at risk, carried out after the initial admixture of infected and susceptible animals, would prevent the occurrence of a wave of mortality, or bring to an early close an epidemic that had already developed a high death-rate. Two experiments on these lines have been completed. In each case the infection studied has been mouse-typhoid, and the redistribution has consisted in the dispersal of the population at risk into smaller groups housed in separate cages. In the first experiment of this kind,which is recorded in Chart 12, the dispersal was carried out during the pre-epidemic phase On the day when the experiment commenced, 305 normal mice were added to 24 survivors from an earlier epidemic. During the following week there were 15 deaths. Of these, 11 occurred among the normal mice and four among the presumably infected animals. Seven of the dead mice were eaten, seven were examined post mortem with negative results, while one of the survivors, which died on the sixth day, was found to have succumbed to typical mousetyphoid. On the eighth day redistribution was carried out; 294 of the added mice were then alive. Of these, 100 were placed in one large cage ; 94 were dispersed into four groups, three of 25 and one of 19 mice ; 100 were dispersed into ten groups, each of ten mice. The mortality was observed for the following 68 days. On the sixty-ninth day the survivors from each of the two dispersed herds were re-aggregated in a single cage, and the course of events was watched for another 93 days. Chart 12, which gives the daily percentage of survivors for each of the three herds, carries the record to the 120th day of the experiment. In the undispersed herd the epidemic developed rapidly, and had run its course by the sixty-eighth day, when the survivors numbered 5 per cent. of the original population. The herd, which was dispersed into four groups, derived no significant advantage from this dispersal : 9-6 per cent. remained alive on the sixty-eighth day. The herd which was dispersed into ten small groups suffered a far lower mortality ; on the sixty-eighth day the survivors from this herd amounted to 55 per cent. of the mice originally at risk. In the period which followed re-aggregation this herd suffered further loss. It will be seen from the chart that the curve recording * The first and second lectures appeared in THE LANCET of March 6th and 13th respectively.

the daily 1x values continues its downward course; indeed this curve, from the tenth day of the experiment onwards, approximates closely to a straight line. When the experiment terminated, on the 161st day, the survivors from the undispersed herd numbered 2 per cent., from the herd divided into four groups during the period of dispersal they numbered 2-1 per cent., from the herd dispersed into ten small groups they numbered 24 per cent. These figures give the total mortality. The specific mortality, which excludes those deaths in which no evidence of enteric infection was found post-mortem, was 87 per cent. in the undispersed herd, 83-7 per cent. in the four-group herd, and 18 per cent. in the ten-group herd. Thus, while dispersion into relatively large groups during the pre-epidemic phase had little influence on the spread of enteric infection among the mice at risk, dispersion into many small groups had a marked inhibitory action. Re-aggregation was followed by a further spread of infection among this herd, but on the 161st day after the experiment started they still showed a relatively low mortality from this disease. Before discussing the significance of these results, we may consider another experiment, in which dispersal was carried out at a later stage of the epidemic process. This epidemic, which is recorded in Chart 13, was initiated by adding 420 normal mice to 60 others, among which enteric infection was actively spreading. This mixed population was allowed to remain in a large cage for seven days, during which time 45 deaths from mouse-typhoid occurred among the 420 added mice, showing that an epidemic was well established. On the eighth day 360 survivors from the 420 mice were divided into four herds, each of 90 animals. Two of these, recorded in the chart as Pand C 1, were each kept as a large group in a single cage. The two others, referred to as P 2 and C 2, were each dispersed into 18 small groups, each of five mice. The survivors from each of these dispersed herds were re-aggregated in a single cage on the forty-second day, and the experiment was continued until the eighty-fourth day. The mice constituting the herds P 1 and P 2 were injected with a bacteriophage active against B. aertrycke, but this fact does not concern us here, and it will be seen from the chart that their fate did not differ from that of the mice of the control herds Cand C 2, which received no treatment. Confining our attention to the effect of dispersal and re-aggregation, it is clear that distribution into many small groups during the epidemic phase has no such marked effect as was observed when dispersal was carried out before any wave of mortality had been developed. During the first three weeks there is no significant difference between the course of events in the four herds. The undispersed herds suffered, during this period, mortalities of 72-2 and 67.8 per cent., and the two dispersed herds mortalities of 64-4 and 62-2 per cent. From this point onwards, however, the Ix curves diverge. The two curves recording the proportion of survivors in the dispersed herds show that change in direction which we have already noted in describing other epidemics of mouse-typhoid, while the two curves recording the fate of the undispersed herds continue On their downward course for some time longer. the forty-second day, when the mice in the dispersed herds were re-aggregated, the percentage of survivors.in these two herds amounted to 25-5 and 24-4 per cent., and in the two undispersed herds In this case the reto 11-1I and 7-8 per cent. aggregation of the dispersed herds was not followed by any marked rise in the death-rate, though a few deaths occurred between the forty-second and eighty-fourth days. The percentages of survivors on the eighty-fourth day were 17-8 and 16-7 per cent. among the dispersed herds, and 8-9 and 2-2 per cent. In all cases the among the undispersed herds. mortality was almost entirely specific. Considering these two experiments together, it appears that dispersal carried out during the earlier

646

I

phases of

more effective it would be, since the chance of the inclusion in each group of one or more highly infected mice will obviously increase as infection spreads among

an epidemic tends to reduce the total mortality. This effect is marked during the pre-I epidemic phase, but much slighter when the waveI

of

mortality has commenced

to rise.

It

probable that dispersal will become effective as the wave approaches its

CHART 12.

seems

less

crest. Once it has commenced to subside it is unlikely that dispersal will have any appreciable influence on its further course. It is not difficult to suggest a possible mechanism for the effect of dispersal on the total mortality. Whether the important factors in the production of fatal infection be the size of the first dose of bacteria received, the rapidity with which successive doses follow each other, the relative

virulence

of the bacteria, or the relative resistance of the receiving host, we have seen that the effectiveness of these factors must be expressed in terms of chance. By dispersing the population at risk we shall redistribute these chances in a random way, and the sphere of

the undivided herd. We may note that observation of shows that the death of a single mouse among a small centres acting dangerous infection will be limited. During group from enteric infection is not necessarily followed the period of dispersal some by any further deaths among its companions within that group, while examination of the faeces shows that mice may be exposed to increased risk, since they one or more mice in any small group may be infected may come into more with B. aertrycke without any deaths occurring in frequent contact with that group during the period of dispersal. That highly infective com- re-aggregation should be followed by a further spread panions, but even though of infection is to be expected, and it is not surprising these contract a fatal that this should be more marked in the case in which infection they will not dispersal has been carried out earlier, and the re-aggrebecome centres of redis- gated population is more numerous, since, in this case, tribution of th3 parasite many of the mice may have been exposed to little risk to large numbers of their of infection during the period of dispersal. We may call attention to the analogy between these companions. It would be expected, o n results and the observations which have been made this hypothesis, on the effect of the dispersal which follows school that the earlier closure in rural areas. The report by Power 56 on an dispersal was outbreak of diphtheria at Pirbright in 1883 affords carried out an excellent example. In this connexion we may refer to certain experiments which we have carried out in con-1 I nexion with mouse-typhoid and pasteurellosis,’ but which cannot be described in detail here, We found that it was not possible to propagate 1

action ot tnose mice wmcn are as

ii

an

I

the

Bj;

647 either of these infections over more than a short period cent. were surviving on the eighty-fourth day. A by allowing infection to spread from one small group second herd (E 4) consisted of 100 normal mice which to another, each group being separated from the had been exposed to risk for 14 days. They received preceding group and kept in isolation for a few days one inoculation of vaccine on the day when the before the succeeding group was added to it. Under period of observation began, and a second inoculation such circumstances, which limit the effective action seven days later. Of this herd 52 per cent. survived. of any particular group of mice to that exerted The mice constituting the third herd (E ) were inoculated with the same vaccine on the twenty-eighth and on a few companions for a short interval of time, the infection dies out after a few successful transfer- thirty-fifth days of the period of observation. The ences have occurred. Recent unpublished experiments percentage of survivors was 43 per cent. The "fourth herd (E 1) received no treatment of any kind and served on the spread of enteric infection have yielded much evidence pointing in the same direction. It would as a control. On the eighty-fourth day 51 per cent. remained alive. It will be noted that the mortality seem that, for the development of a high epidemic potential, full play must be given to the effect of among all herds was relatively low. This series chance associations by the maintenance of the of epidemics were of the type referred to in my first population at risk at a reasonably high level. As lecture, in which the slow development of mortality So far as we have seen, the most favourable conditions appear is followed by a continuous low death-rate. to be provided by the continuous immigration of the results go, however, they are suggestive. The mice which were actively immunised before exposure susceptible hosts. to risk appear to have maintained a relatively high HERD IMMUNISATION. level of resistance over the 84 days of the period of Another method of intervention clearly merited observation. The final percentage mortality among inquiry. It had been shown by Webster, and by this herd was low as compared with that among the that it was possible to increase the resistance others, control herds, though occasional deaths from typical of mice to B. aertrycke by immunisation with killed enteric infection were occurring among them. The cultures ; and observations by Lynch 19 on the two herds, which received their first inoculation of spontaneous epidemic of mouse-typhoid, which was vaccine after 14 days, and after 42 days respectively, the starting-point of the experiments carried out at show no advantage over the controls ; indeed, the the Rockefeller Institute, suggested that vaccination herd vaccinated at the later period show the highest with killed cultures might be of value in limiting the percentage mortality of any of the groups. There epidemic spread of the disease, although this con- is a suggestion-we would not put it higher-that clusion is not in entire accord with the later observaof this type must be conferred during the tions of Pritchett.37 We have carried out a few immunity pre-epidemic phase if it is to be effectual, and that experiments to investigate the possibility of effective interference by this means, once the epidemic process herd immunisation by the inoculation of killed cultures has a certain stage, may do more harm than passed of B. aertrycke. In the first experiment of this kind We would enter a caveat here against any six groups of 30 mice were fed on three occasions with good. from analogy as regards the control by argument bread soaked in a broth culture of B. aertrycke, the vaccination of naturally occurring epidemics, without subsequent course of events being observed over a careful consideration of the differences which may be in these of 60 The of mortahty days. progress period introduced by the close admixture of all animals at groups is recorded in Chart 14. Groups A and B con- risk in our experimental cages. sisted of normal untreated mice. Groups E and F MECHANISM OF THE PROPAGATION OF AN of mice which had previously received two consisted EPIDEMIC WAVE. inoculations of a killed suspension of B. aertrycke, We have now completed our summary of the data the second inoculation having been given 21 days before the first feeding on living culture. The mortality which havebeen yielded by direct experiment, and following inoculation was negligible, and the examina- have tried to indicate the conclusions which we should tion of the blood serum of a sample of the inoculated draw from the available evidence and the degree of mice had shown the presence of agglutinins in high probability which attaches to them. It is clear that concentration. Group C contained 20 normal and the experiments so far completed can be regarded 10 vaccinated mice, and Group D 10 normal and only as preliminary studies, defining the problems 20 vaccinated animals. As the chart shows, the at issue rather than solving them. It seems to us vaccination produced a significant reduction in that we have obtained some useful indications, that mortality and, with the exception of the relatively a few fixed points have been charted from which we low mortality in Group C, the percentage mortality may proceed to further explorations ; but it would, was lower the higher the proportion of vaccinated in our view, be altogether premature to attempt, on mice. It is clear also that the protection was only the basis of the experimental findings, any generalised relative, and the occurrence of deaths from mouse- description of the complex interactions between typhoid in Groups E and F during the later stages of opposing factors, which must determine the course the experiment suggests that mice which had acquired of epidemics such as those we have studied. an immunity adequate In this connexion we must note a view which has against the initial feeding succumbed to the later spread of infection. In a recent been put forward by Webster 27 as the result of his experiment we have attempted to estimate the effect observations on mouse-typhoid. We have referred of active immunisation carried out at various periods to the mortality curve which he has obtained by of the epidemic process. We brought together in a inoculating samples of mice with B. aertrycke, and large cage 879 mice. Of these 83 belonged to a group ’, plotting percentage mortality against time ; and we among which mouse-typhoid was slowly spreading, ’, have pointed out that these results can be plotted as but in which no marked wave of mortality had yet frequency curves, showing the proportion of the total occurred ; 146 mice had received two inoculations deaths that occur in successive intervals of time. of B. aertrycke vaccine, the second of which had been Webster has plotted his " standard curve " in this way, given 14 days previously ; 650 mice were urtreated and has obtained, as he could not fail to do, a frequency normal animals. During the following 14 days curve, rising to a maximum at a point on the time 35 deaths occurred. The majority of the mice were base-line when deaths were most numerous, and eaten by their companions, but six of the deaths were sinking to zero again as deaths became less frequent, proved to be due to mouse-typhoid. The surviving and finally ceased to occur. This, of course, is simply mice, excluding those which had been employed to the graphical expression of the fact that, if we inoculate start the epidemic, were now divided into seven a large number of mice with a pathogenic bacterium herds each containing 100 mice. We are here con- in an amount which is in the neighbourhood of the cerned with four of these herds only. The course of average lethal dose, many of these mice will die during events was observed for 84 days, and is recorded in some particular time interval, which marks the limit Chart 15. One herd (E 3) consisted of mice which had of the modal duration of a lethal infection, fewer mice received two inoculations of vaccine before being will die at some earlier time of a more acute infection, exposed to any risk of infection. Of this herd 71 per or later of a more protracted attack of disease, while

648

certainly cannot accept, in speaking of this curve as the simple epidemic curve." It is not necessary to discuss this question in any Webster himself has shown, a curve essentially in form is obtained after detail. In the numerous experiments which we have inoculating mice with a described above, in which the daily addition of normal mice to an infected herd was associated with a series poisonous metallic salt. It is clear that such a of well-marked waves of mortality, it is clear that no frequency curve will re- large number of normal mice, hitherto unexposed to semble certain of the the risk of infection, were suddenly exposed to the curves of same chance of receiving any given dose of the bacterial mortality parasite-quite certainly, they did not all receive the

very few or none will die within a very short interval or will succumb after a certain limit of time. As

similar

are which observed during the

which

we

frequency

"

dose at the same moment of time. The course of events in these experiments, and in many of those in which a single large batch of mice were added to

same

infected

companions, indicates clearly that the rise of a wellmarked wave of mortality is preceded by a wide spread of infection, which will probably have modified in many ways the original condition of the animal hosts at risk. The successful analysis of the epidemic process is, in our view, unlikely to resolve itself into any

such simple formula. We should note that Webster himself favours

complex hypothesis, in connexion with the spread of a pasteurellar infection among rabbits. a more

course of an epidemic. Webster 27 has noted this resemblance between the curves obtained in his own inoculation experiments and some of those recorded by Amoss 21 in the experimental epidemics in which mice were added irregularly to an infected population. From this resemblance he has drawn the conclusion that the form of the mortality curves in the experimental epidemics depends upon the varying degree of susceptibility within a mouse population

simultaneously exposed

to dose of mousetyphoid bacilli. We are not certain how much Webster significance attaches to this suggestion. If he implies that very occasionally, and under very exceptional circumstances, a large number of hosts may simultaneously receive the same dose of a virulent organism, and that, should such an event occur, the form of the distribution in time of the resulting deaths will be mainly determined by the variable resistance of the hosts, then we have no That criticism to make. Webster does set some limiting condition to the application of this hypothesis to natural or experimental epidemics is clear from several of his papers ; but, unless we have misread him, he regards this variation in normal resistance as the factor which determines the time-distribution of deaths in any wellmarked wave of mortality

a

fixed

during

an experimental epidemic of mouse-typhoid. In fact, he uses a term

EPIDEMIC SPREAD OF INFECTION : SOME GENERAL CONSIDERATIONS. If we look beyond our experimental results, and include in our survey the bacteriological data which have been obtained in the observation of naturally occurring epidemics, can we, without too adventurous an application of the dangerous argument from analogy, obtain any further suggestions as to the nature of the epidemic process ? The results which have been obtained along two lines of inquiry appear to us to be particularly pertinent to the points at issue. The first of these is that which has been developed in connexion with the so-called carrier problem, and which has led to a steady accumulation of knowledge concerning the distribution of certain bacterial parasites among the human hosts at risk. It is impossible to discuss in any detail the enormous mass of data which has been collected, but we may attempt a generalised summary of our present conception of the kind of equilibrium which exists between parasite and host in any area in which a particular disease is endemic, with occasional or recurrent appearances in an epidemic form.

649 It has become clear that, where these conditions So far as we are justified in combining our experiexist, the clinically recognisable cases of disease form mental data with such field observations as those we only a fraction of the infected individuals of the host have referred to, it would seem that we may regard species. Adequate bacteriological examination will the epidemic spread of any disease as involving a reveal numerous hosts who are suffering from some series of reactions between parasites and hosts which atypical disturbance of health, and from whose tend to produce quite different results. We may mucous surfaces or excreta the specific organism of suppose that the primary interaction of the pathogenic the disease can be recovered, and others, still more bacteria and the susceptible hosts will lead to fatal numerous, who are harbouring the parasite without any infections, to infections clinically recognisable as definite evidence of illness. If we examine adequate cases of disease, to atypical infections involving some sairples of cases of disease, of contacts with atypical recognisable departure from health, and to latent lesions or symptoms, of healthy contacts, and of infection or to the carrier state, in which the host healthy non-contacts, we shall generally find a may harbour the parasite over long intervals of time descending order of frequency of the presence of the without any sign of disease. The proportions which specific parasite ; but in many cases-and this is these various products of reaction bear to one another will be determined by various factors, the isolation a point of particular importance—we may obtain

equal values for the carrier-rate in healthy contacts and non-contacts within an epidemic area, so that there is no reason to suppose that contact with a clinically recognisable case of disease is the main mode of infection leading to the carrier state. The practical significance of this distribution of different states of equilibrium between the individual hosts and the parasites they harbour will, of course, depend on the relative frequency of the various classes, and it is here that more knowledge is urgently required. In some infections, such as cerebro-spinal fever, we know that the cases of disease are in an insignificant minority, that they are, to use a convenient but equivocal term, an epiphenomenon. In this case the carrier-rate among the population at risk may be carefully observed as it rises from a figure of a few units per cent. to 20 per cent. or more, before a single case of meningitis occurs, and the actual outbreak of disease may number only a few cases among many thousands at risk. The real epidemic, as Arkwright has pointed out, is an epidemic prevalence of meningococci in the naso-pharynges of the population at risk. No purely clinical description of the course of events could ever have been other than a revelation so partial as to obscure all essential truth. We can be fairly sure that poliomyelitis and encephalitis lethargica behave much like cerebrospinal fever in this respect, because the bacteriological findings in the latter disease have taught us how to interpret the clinical observations in the two former. In the case of poliomyelitis, indeed, the carrier state has been shown to exist. It is probable that diphtheria, and perhaps scarlet fever, show a higher ratio of cases to carriers, though this ratio is certainly much less than unity. It is possible that such a disease as small-pox is at the other end of the scale, and that here the carriers are in an insignificant minority, though none will deny the frequency of the atypical case. I would hazard the personal guess-it is nothing more, and I have found few supporters among those best qualified to offer a clinical opinion-that this relation only obtains in an area in which small-pox is introduced after a considerable period of complete freedom from this disease, and that when we are faced with an annually increasing prevalence, as we now are in this country, we may feel fairly confident that the carrier is there, although he receives no professional recognition. It would appear that the more we have learned of the bacteriology of any particular infective disease, the more probable is it that the distribution of the parasite is far wider than the distribution of the disease. The second line of inquhy has been more recently developed, but has already yielded evidence of the greatest significance. The introduction of the Schick test has enabled us, in the case of diphtheria, to add to our study of the distribution of the parasite a study of the distribution of resistance among the hosts at risk. Further, and even more significant than this, we can now observe the gradual immunisat on of an infected herd, and correlate the spread of immunity with the spread of the parasite. Such a study as that which Dudley 53 has carried out, on the natural immunisation of the inmates of an institution during the progress of an epidemic of diphtheria, affords an admirable example of the most fruitful kind of field observation based on bacteriological data,

and measurement of which is the main

object of our experimental inquiry. We may suppose, by an induction the imperfection of which we fully realise, that those hosts who contract a non-fatal infection will thereafter possess an altered resistance to the specific parasite, and that this alteration will, in general, be in the direction of an increased resistance of varying duration. Using a chemical analogy of doubtful validity, we may regard the reaction as in some degree autocatalytic, since some of the products of the primary reaction will, by their mere presence, tend to accelerate the shift of equilibrium in the direction of the accumulation of infected hosts. But one line of reaction will produce a somewhat different result. If a non-fatal infection leads to immunisation, there may be a gradual accumulation of resistant individuals who may be relatively unresponsive to further doses of the bacterial parasite, and a consequent elimination of those susceptible hosts who form an essential reagent. It would be easy on these grounds to account for the predominant importance of a steady immigration of susceptibles, which has emerged so clearly from our experimental studies. If such a working hypothesis should be established on a firmer basis by further research, it would appear probable that the relative amounts of those products of an epidemic, which are clinically recognisable as disease or death, will depend upon the relative rate of those reactions which lead to immunisation on the one hand and to disease on the other. If the conditions favour immunisation there may be a widespread epidemic with few cases and still fewer deaths ; if they favour more severe infections there may be more disease, and many of the cases may end fatally. If we could determine the nature of the factors which control the rate of these various reactions, we might reasonably hope to interfere successfully in the epidemic spread of disease, by breaking some essential link in the sequence of events on which this spread depends.

PRACTICAL APPLICATION OF PRESENT KNOWLEDGE. It seems desirable to make some attempt to consider the efficacy of our present methods of dealing with infective disease, and of possible developments of them, in the light of the knowledge which has so far Such consideration is more than been acquired. a little premature, and I would offer it in the spirit of a famous contributor to the theory of the ills of populations, who says :By pursuing this plan, however, I am aware that I have opened a door to many objections, and, probably, to much severity of criticism ; but I console myself with the reflection that even the errors into which I have fallen, by affording "

handle to argument, and an additional excitement to examination, may be subservient to the important end, of bringing a subject so nearly connected with the happiness of society into more general notice." It is natural first to consider those administrative a

which may be employed to lessen the chances of contact between infected and susceptible hosts, and we may divide such measures into those which are specific, in the sense of dealing with particular diseases, and those which merely aim at a general reduction in opportunities for the transmission of infection. Under the first heading we must obviously consider the probable efficacy of the isolation of cases of

measures

N2

650 The experimental and bacterio- on the incidence of human and animal disease has been disposal do not suggest that this established beyond question. Perhaps one of the most method has much to recommend it as a measure convincing examples of its influence has been provided designed to reduce the frequency of disease and by the studies which were carried out during the late particudeath, due to an infection which is endemic among the war on the incidence of cerebro-spinal fever, population at risk. Without pressing the analogy, we larly the observations made byy Glover. 57 It seems possible, however, that we may expect too might suggest that such a procedure is rather like the continuous removal of an unpleasant product of a much in future from improvements along these lines, chemical reaction, in the hope that the reaction will though they should clearly be exploited to the limits thereby be brought to an end. A more hopeful of their possibilities. I would suggest that it might procedure would be to determine the conditions of be helpful to adopt a bacteriological definition of the reaction, as a preliminary to altering them. The effective overcrowding as " any spatial distribution opinion of experienced administrators, on the actual of a population which favours the trequent transference results of the policy of isolation, bears out the of a particular pathogenic parasite from host to host." expectation based on conclusions from bacteriological On this definition overcrowding, as it is concerned in observation and from experiment. the spread of infection in typhoid, influenza, malaria, The isolation hospital may, perhaps, be defended or typhus, would depend on entirely different factors. on the grounds of the benefits which it confers on the I think that the absence of some such definition has sick persons for whose treatment it provides. As a led to a little confusion in certain statistical inquiries. means of lessening the general incidence of infective Some measure of overcrowding must be adopted in disease it receives little support on scientific grounds, calculating its probable influence on the spread of and apparently finds few friends among those who any given disease, and it is usual for the statistician It is, to employ some such measure as the number of houses are best fitted to judge of its practical effects. of course, reasonable to argue that the isolation of per acre, the number of families per house, or the a person sick of an infectious disease must of necessity number of persons per room. I would suggest that reduce the risk run by a few persons who would other- such a measure may entirely neglect the effective wise be brought into contact with the patient. This overcrowding to which a population is subjected. The point at issue is whether the At the moment, the enemy which threatens us is none can deny. spread of infection from diagnosed cases to susceptible respiratory disease. Influenza and its allies determine hosts forms such a significant fraction of the total the reigning epidemic constitution. It seems very spread of infection, which is occurring in the affected possible that it is our trains, and tubes, and ’buses, area, that the suppression of this one avenue of spread and trams which ensure a highly successful admixture will appreciably lower the incidence of disease. of hosts and ample opportunities for transference of One point may be here stressed. It seems to us to parasites. The more we expand our residential suburbs, be no sufficient answer to a criticism of any particular keeping our workshops in a few large centres of administrative method to suggest that it probably industry, the more shall we accentuate this particular In dealing with the acute respiratory does some good, and that while any doubt remains difficulty. the public .are entitled to such protection as the diseases of mankind it does not look as though the enforcement of that particular measure may possibly reduction of effective overcrowding were likely to afford. The hard facts of the case are that public prove the strong suit in our hand. We are faced health administration is expensive in effort, time, and with another illustration of the wisdom of Robert money, and that the total expenditure which can be Watt’s aphorism : " We may, it seems, under permisdevoted to the prevention of disease is limited by the sion of Divine Providence, deprive death of some of available resources. Elementary economy demands his apparently most efficient means ; but deprived of that we should apply our efforts in those directions these, new means are discovered, or the old improved." In the control of epidemics among animals we are in which we can hope for the best return. If the isolation of the sick offers little hope of the not limited by the considerations of social expediency, effective control of infectious disease, can we expect and more may be hoped for from a strict control over better results from the detection and isolation of the movements of healthy and infected stock. It is carriers ? The answer here must be given by expe- probable, however, that the principles which limit the rience. In certain diseases, and in small outbreaks in utility of such methods are as applicable to animal as which the entire population at risk is susceptible to to human diseases. We would express a personal unusually complete control, as, for example, in out- doubt as to the benefit which is likely to result breaks of diphtheria in institutions, much may be done from the policy of slaughtering cattle which form in this way. In a few diseases, such as typhoid fever, part of a herd, among which a case of foot-and-mouth in which the total incidence may be relatively small, disease has been diagnosed. In a country which had and in which a, few chronic carriers may be the only been free from this infection over a long period of important sources of infection in a large district, the time, such a policy might well be adopted in an detec tion and control of these individuals is obviously attempt to stamp out the new infection before it indicated. In most diseases, and in all cases in which obtained a foothold. But, once that foothold has we are concerned with an endemic infection widely been obtained and recurrent outbreaks have shown distributed among a large uncontrolled population, that the infection has become widely generalised, the detection and isolation of carriers has been shown it seems possible that slaughter may have effects somewhat different from those intended. We have to be impracticable. Turning to those administrative measures which seen that the continuous immigration of susceptibles bring about a general reduction in the opportunities appears to be the most important single factor in for the transference of infection from host to host, maintaining the level of epidemic prevalence of an theoretical considerations suggest, and practical infective disease. If we may assume that slaughter experience affirms, that any measure of this kind has, in fact, failed completely to stamp out infectionis to be welcomed with enthusiasm. Any improve- and the history of the past few years suggests that this ment in sanitation, which ensures a clean water, or is the case-then it would seem that the elimination milk, or food supply, which reduces the frequency of of all those animals which havebeen exposed to risk, insect vectors of infection, or which reduces close and their replacement by uninfected susceptibles contact between individuals, may be expected to lead trom other areas, may have little to recommend it to a reduction in the incidence of those diseases, with as a means of reducing the incidence of the disease. the transference of which that particular measure of In the old days of sailing ships and press-gang, the This expectation has been experienced captain held strong views on the unwisdom sanitation interferes. fulfilled whenever an effective measure of reducing the of introducing among his crew any large number of general opportunities for the transference of a particular men, however healthy, who were "not salted to the infection has been established. ship," for he knew from bitter experience that the This brings us directly to the vexed question of over- mingling of samples of humanity who had for long crowding. The dominating influence of this factor been exposed to different risks of infection might have

infective disease.

logical data at

our

651 disastrous consequences. The principle is probably of wide application in human and animal disease. It would seem that our best hope of controlling those infectious diseases, which refuse to yield to those means of general sanitation which are compatible with the

DIABETES

modern organisation of our social life, lies in devising some effective means of immunisation, and so raising the average level of herd-resistance. We have already referred to the use which has been made of the Schick test in the study of diphtheria. The results which have been obtained by active immunisation based on the information obtained by the use of this test afford an excellent example of the success which attends the application of scientific methods to the prevention No better instance -of the utilisation of of disease. such methods could be cited than is to be found in a recent paper 58 by Okell, Eagleton, and O’Brien. We have seen reason to believe that the efficacy of such methods may vary greatly according to the period of the epidemic process at which they are applied. If the best method of immunisation at our disposal will yield only a slight increase in the average herd-resistance, it may make all the difference whether such immunisation be applied early or late in the spread of infection. We cannot envisage with equanimity the immunisation of the whole of mankind to all the infectious ills which flesh is heir to. If, as seems very probable, any outbreak of infectious disease is preceded by a redistribution of the specific parasite among the population at risk, a knowledge of such a redistribution might yield valuable indications as to the appropriate measures to be adopted. It is possible that one of the duties of the public health intelligence service of the future will be to keep a watchful eye on variations in the microbial flora of mankind, by a continuous sampling of the population. Under the auspices of the Ministry of Health we are carrying out such a sampling of the Manchester population with regard to some of the more important members of the nasopharyngeal flora, and ’I correlating the results with the incidence of the various types of respiratory disease. It seems probable, however, that the earliest testing, on a field scale, of any principles which may be derived from experimental results must be carried out in connexion with animal rather than with human infections, since the opportunities for control and for the collection of special data are so much greater ; and I would end as I began by emphasising the need for still closer cooperation between the students of human and of animal disease, and for a broad biological outlook on the part of both. I should wish, on behalf of Dr. Greenwood and myself, to express our thanks to those who have helped us during the course of this work ; on the bacteriological side to Dr. G. S. Wilson, Mrs. Joyce Wilson, Dr. L. P. Lockhart, and Miss E. R. Lewis ; on the statistical side to Miss E. M. Newbold, Miss Thomas, and Messrs. Fanning and Martin ; and to Mr. W. Bale for his continuous assistance in the care of our stock of normal and infected mice. We acknowledge our great indebtedness to the Medical Research Council for their generous support throughout the past five years. Investigations of this nature are necessarily protracted and expensive, and without the Council’s support they could not have been undertaken. JEMMM)6wapA!/.—The bibliography to Lectures I., II., and III.

appeared in THE LANCET of March 6th, p. 484.

DEATH

OF

SALFORD CORONER.-The death occurred

MELLITUS:

A RECOVERABLE DISEASE ? BY G. H. PERCIVAL, M.B. EDIN., D.P.H. EDIN. & GLASG.

(From

the

Department of Therapeutics, University of Edinburgh.)

THE following cases, selected from records of a large number of diabetics who have been treated in the Royal Infirmary, Edinburgh, are reported to illustrate the possibilities of modern treatment. The diet scale used is that described by Prof. Murray Lyon,’ consisting of three types graded according to the carbohydrate-fatty-acid ratio-viz., the A diet having a ratio of 2 to 1 or over, the B diet having a 1 to 1 ratio, and the C, in which the ratio approaches in the lower caloric values, and reaches in the higher, the optimum ratio of 1 to 1-5. A further

modification is found in the D diets, where the ratio remains 1 to 1-5, but in which the protein content is increased. The A diets are always accompanied by insulin, and are employed to abolish severe clinical ketosis ; the B diets may be used with or without insulin to get rid of glycosuria ; the C diets are used alone to estimate the patient’s tolerance, and also with or without insulin to slowly advance the caloric intake. Patients may be finally kept on the higher values of C or changed over to the D scale, which affords a more suitable maintenance diet on account of its higher protein content. The various diets are selected and interchanged to suit the requirements of individual cases during varying degrees of tolerance. They are referred to under the group letter, followed by a figure indicating the number of available calories in hundreds-e.g., a B 15 diet is one yielding 1500 calories, built to give a carbohydrate ratio of 1 to 1. Insulin is withheld, except in cases possessing such a low carbohydrate tolerance that it is impossible to abolish glycosuria, or to continue an adequate caloric intake, by dietary measures alone. When required, it is injected two hours before a meal, except in very severe cases, when the interval is reduced at first to All patients requiring one or one and a half hours. insulin are taught, and those who have had preliminary hospital supervision have in most cases experienced, the symptoms of hypoglycaemia, and their remedy. The urine is tested for sugar daily by Fehling’s method by the patients themselves, and also once weekly by the sister dietitian. Gr01.tp Å.

CASE l.-A male, aged 23 years, had complained of lack of energy and tiredness on rising in the morning for about two months. This was attributed to frequent coincident attacks of trigeminal neuralgia, and two carious teeth were removed and a root abscess drained. About this time he commenced to be troubled with increasing frequency of micturition accompanied by urgency, and occurring especially after meals. The lassitude and urinary symptoms continued for three weeks. His urine contained 2-5 per cent. glucose one hour after a meal (estimated by Benedict’s method), and a trace of acetone was present. The fasting blood-sugar content was 210 mg. per cent., and a sample of urine taken at the same time contained 0’5 per cent. glucose. At that time his weight was 9 st. 12 lb., height 5 ft. 7 in. He was given a B 15 diet and in two days the urine was sugar-free. This diet was continued for a week, then he was given C 13, with an increase of 100 calories every third day till C 28 was reached. He was kept at this level for a week, then changed to the D scale and raised to D 30. During this period of dieting he lost-9 lb. weight during the first fortnight, but this was soon regained. Lassitude, neuralgia, and urinary symptoms quickly disappeared during the first month. Eight months have elapsed since he first came under observation, and he now takes an ordinary diet with some limitation of cereals, and avoids substances containing actual sugar. Glycosuria occurred on one occasion after a full ordinary meal had been taken, as the patient felt sure that he must have completely recovered. The experiment was not repeated, and there has been no reappearance of glycosuria.

at his home at Altrincham, near Manchester, on March 8th, of Mr. Arthur Holmes, coroner for the borough of Salford. Mr. Holmes was in his sixty-eighth year and was a son of the manse ; he was appointed coroner in 1896 on the death of Mr. Frederick Price. For many years it was the custom in the borough to hold inquests on licensed premises, and during a long murder case a missing witness was found to have been taken into custody, as he had been drinking to excess during the long wait. This untoward incident was largely responsible for a room being set apart in Pendleton Seven similar cases have been examined, and a town hall for sittings of the coroner’s court. It was in this court that the arsenic in beer scare was first brought to light general summary will avoid much repetition. and investigated. For some time Mr. Holmes acted as clerk Four of the cases were males and three were females, the to the Manchester Port Sanitary Authority. ages ranging from 24 to 40 years. They sought advice for