View report page

85
The calculation of the total reduction in the number of cases resulting from holidays is a matter
of some difficulty. It is necessary to determine approximately the contour of prevalence as it would
be if there were no holidays, and the simplest method is, of course, to assume that the cases would increase
from week to week during August at the same rate as in July, that the curve of incidence from July
to, say October, is, in fact, a straight line. The error involved in such an assumption is not determinable
from consideration of the London figures alone, and in order to form some idea of its magnitude it
was necessary to examine the records of seasonal prevalence of other towns in order to find a case where
the maximum seasonal prevalence fell at some time of the year when school holidays could not affect
the seasonal contour. New York furnishes the best contour for the purpose, and the figures relating
to that city which were kindly supplied by Dr. Bolduan to Mr. Spear are exhibited in the diagram
on p. 84—where corresponding figures for Paris supplied by Dr. Bertillon are similarly reproduced.
It will be noticed that the New York curve is decidedly asymmetric, reaching a maximum in the
course of eight months and falling to a minimum in four months, and the London contour, if allowance
be made for the effect of holidays, is also evidently asymmetric. An experimental curve was obtained
by combining the figures for New York and Paris and calculating from the resulting contour the effect
of midsummer holidays in London, and close examination showed that an assumption of linear progression
during this period does not give results which differ seriously from the more accurate method
of calculation in the series of years in question. '
The total number of cases represented by the difference between the cases actually recorded in
August and the calculated number were holidays not in operation involves, for ages 5-13, an addition
to the actual figures of about 22 per cent.
Similar contours are shown for diphtheria, but there is in this case nothing like the pronounced
seasonal recurrence found in scarlet fever, even in New York; nor do the contours show anything like
the same degree of asymmetry. The irregularity of the diphtheria contour appears to be to some
extent due to a tendency for the notifications of the two diseases to fluctuate in harmony, a point to which
reference has elsewhere been made; it may, however, be remarked here that the high
proportion of cases of diphtheria in New York and in Paris during the period February-June, absent
in the London curve, is probably a reflex of the high scarlet fever incidence in those cities at this period,
not found n London.
In the case of diphtheria the calculated addition to the actual cases in the midsummer holiday
period required to allow for the effect of the holidays was found to be about 26 per cent. ; this compares
with the corresponding figure for scarlet fever given above of 22 per cent. The difference in the
two percentages might suggest that school attendance plays a greater part in the case of diphtheria
than in scar et fever ; here, however, the loss of supervision during the holidays, to which reference
has been previously made, comes into account; for the bacteriological testing of the children's throats
on the scale adopted in any affected school or class room during term time is absent, and this circumstance
probably suffices to explain the difference between the percentages found.
The high incidence of scarlet fever in 1920 simultaneously with the relatively even higher
incidence of diphtheria put a severe strain upon the resources of the Metropolitan Asylums Board
especially during October. Uncertainty as to when the climax of the combined epidemics would be
reached added to the difficulty of allocating the available accommodation. The value of some means
of forecasting the course of the epidemics thus became apparent, and possibly gave rise to a note upon
this point by Dr. Brownlee in the Lancet for 27th November (p. 1093). The method adopted was
that of fitting a normal curve of error to the observed case-sequence in the nine large London epidemics
since 1890, and from the sequence of the weekly cases for the seven weeks known at the time of writing
his note Dr. Brownlee deduced by means of the auxiliary curve, the probable case-sequence of the
epidemic in the following eight weeks, with a fair amount of accuracy. It has, however, been shown
above that the autumnal curve for London is asymmetric, when the effect of school holidays is allowed
for; that is to say that theoretically the curve of case-sequence cannot except for short periods and
more or less accidentally follow a normal curve of error.
The forecasting of the course of an epidemic must under any circumstances be somewhat
unsatisfactory from the point of view of the authorities upon whom falls the duty of providing the
requisite hospital accommodation, for while a few hundred cases more or less than the estimate does
not disturb the equanimity of the prophet, an unexpected demand for two or three hundred beds is a
different matter.
It may well be asked, with the present high incidence of scarlet fever in view, what are the
prospects for this year ? Already, during January and February, the cases notified number over
4,000 whilst in the corresponding period of last year less than 2,500 cases occurred. In past years
the total of the cases for any given year has not depended upon the prevalence in the opening months;
the correlation has been found to be greater between the cases recorded in July or June and the year's
total than with earlier months, so that a forecast of the probable incidence for 1921, would be mere
guess-work until about July. It will be noticed from the diagram (p. 81) that in London scarlet
fever has tended to recur at intervals of about six years with epidemic force. The epidemic of 1907-8,
was followed by that of 1913-4 and this again by that of 1920-1. Upon these major "waves" the
annual seasonal fluctuations are super-imposed. At times these seasonal curves are imposed upon a
rising major "wave"; at others the major "wave" is falling or level; and it is obvious that the
contour of case-sequence for a given year is to some extent influenced by the phase of the major "wave"
in that year. On an ascending major "wave" the seasonal maximum is attained later than on a
descending "wave." The changes are shown in Sec. I of the diagram on page 87: in the epidemic
year, 1913, the maximum prevalence occurred in the middle of the last quarter, but in 1915 it appears