London's Pulse: Medical Officer of Health reports 1848-1972
View report page
Hints from the Health Department. Leaflet from the archive of the Society of Medical Officers of Health. Credit: Wellcome Collection, London
London County Council 1925
[Report of the Medical Officer of Health for London County Council]
This page requires JavaScript
36
Since influenza is prevalent normally during the winter months it might be
thought that the assumption that an epidemic will recur in a year's time from any
given prevalence would be as accurate as that given by the above method, but
while such a forecast would serve well in the case of epidemics falling about the 9th
week of the year, it would be hopelessly at fault in the case of epidemics several
months later. Thus in the 7 instances where an epidemic has occurred after the
13th week and before the 35th week, the forecast would be nearly 15 weeks too
late on the average, while the average error on the hypothesis of balanced incidence
for the same instances would be less than two weeks.
It may be asked, why the 35th week ? In regard to this it should be borne
in mind that while it might be expected that the maximum temperature would
be recorded at the time of the summer solstice, actually the maximum lags
considerably upon this time. The lag, therefore, of 9 or 10 weeks, in the case of
epidemics which take several weeks to develop, is not inconsistent with dependence
upon seasonal conditions.
Dr. Brownlee (Lancet, 8 Nov., 1919. Also Min. of Health, Influenza Report
1920, p. 575) has published a periodogram analysis of influenza mortality. In a discussion
in the Lancet* of the results reached by him, it was pointed out that in certain
epidemics the amplitude of the "wave" as measured by the mortality was so great
as to over-shadow and practically to eliminate smaller prevalences and thus the
periodogram shows a periodicity where none existed.
The results obtained, on an assumption of Dr. Brownlee's 33 weeks' periodicity and by the suggested method of balanced incidences, are compared in the following table, in which are shown the errors in forecast of influenza prevalences in the years 1890-1924 inclusive, (I.) on hypothesis of balanced incidence about the 35th week of the year, and (II.) on hypothesis of periodicity of 33 weeks.
| Method of forecast. | Forecast too early by | Forecast correct to within 4 weeks. | Forecast too late by | ||
|---|---|---|---|---|---|
| 14 weeks or more. | 5—13 weeks. | 5—13 weeks. | 14 weeks or more. | ||
| I | — | 6 | 20 | 6 | 1 |
| II | 2 | 6 | 8 | 9 | 8 |
It may be of interest to note that the last prevalence of influenza in London
reached a maximum in the 6th week of the year (1925). This is 29 weeks before the
35th week and, therefore, the next prevalence should occur 29 weeks thereafter or
about the 11th week of next year (1926). An "interloper" may arrive about the end
of this year, but in London there have not been more than 3 such interloping epidemics
in the past 36 years.
In conclusion, although it appears from the London figures that forecasts
as long ahead as two years can be made, the number of such instances is too small
to justify any reliance upon them. These may, therefore, be dismissed and emphasis
rather laid upon so much of the hypothesis as appears to be fully established, as
well by the London and Copenhagen records as by others which have been examined,
namely,—
"When an epidemic of influenza reaches a maximum mortality in the
half-year between the 9th and 35th weeks (i.) it will recur in the succeeding
half-year, (ii.) the two epidemics will, as to the time relation of their
mortality maxima, tend to balance about the commencement of the
35th week of the year.
The table on page 37 shows that there are several instances in which the
hypothesis fails, but with the reservation which thus attaches to any particular
forecast it may be noted that a prevalence of influenza is indicated about the end
of 1926.
* The periodicity of Influenza. B. E. Spear, F.S.S., " The Lancet," 23th March, 1920.