View report page

34
The following note relating to the periodicity of influenza prevalences prepared,
by Mr. Spear of the public health department, a year ago, is of interest, inasmuch
as the week in which the maximum influenza mortality was to be expected in 1926
is indicated by Mr. Spear as being the 11th week, which, as will be seen from the
above table, is within three weeks of the actual time of maximum.
There appears to be a law governing the interval separating successive influenza
"waves," and a fairly accurate forecast can, as Mr. Spear shows, be made of the
time likely to elapse between an epidemic and its successor.
Note by Mr.
B. E. Spear
on Periodicity
in
Influenza.
On summing the weekly mortality in London in the past 36 years it is found
that the greatest number of deaths occurs about the 9th or 10th week of the year,
while the minimum mortality is reached about the 35th or 36th week. This latter
period seems to be in the nature of a limit up to and from which influenza " waves "
surge and rebound. For instance, in the remarkable epidemics of 1918 the first
reached its climax in the 28th week. It was followed by the great epidemic of the
autumn of that year with a maximum mortality about the 44th week. The first
epidemic it will be noticed reached its climax about 8 weeks before the period
associated with the minimum seasonal mortality, while the corresponding phase
in the succeeding epidemic was reached 8 weeks after that period.
This balancing about the 35th or 36th week of the year appears to be
characteristic of influenza prevalences and may, therefore, be made to serve as
a means of forecasting approximately the time when a prevalence may be
expected.
Since 1889 there have been 39 prevalences or epidemics of influenza in London
and of these the method would have failed to forecast 3, while it foreshadowed
a prevalence in 1896 of which the maximum cannot be located in the small
figures of London deaths, though in Copenhagen, where influenza is notifiable, cases
reached the maximum about the time expected.
In addition there were in 1900 two influenza prevalences within 14 weeks of
one another, and here the forecast falls midway between the two. The second of
these two prevalences was a relatively small one and is not shown at all in the
Copenhagen figures of cases.
The remaining prevalences follow the rule of balanced incidences about the
35th week approximately, and in 21 instances the error does not exceed 4 weeks.
(See tables on pages 36 and 37.)
The diagram on page 35 illustrates the balancing of the epidemics. The contours
show the influenza mortality in 4-weekly periods in the years indicated. The
vertical line passes through the 4-weekly period which contains the 35th week of
each contour, and it will be seen that the epidemics tend to balance about this line.
On reference to the table on page 37 it will be seen that the error in the forecast
based on the second epidemic of 1916 was greater than in any other year except
1920, being 11 weeks later than expected. The epidemic due at the end of 1891
was 7 weeks too late, while the two epidemics of 1916 balanced exactly. In the
Copenhagen section of the diagram, which is based upon cases instead of deaths, the
vertical line is advanced two weeks, as the maximum cases during a prevalence is
attained earlier than the maximum dea'hs.
It may be thought that an error of 6 weeks in a forecast of this kind is considerable,
but it must be borne in mind that the only way in which the prevalence can
be timed from past mortality returns is by the week of maximum deaths, and while
in some epidemics the maximum is sharply defined, there are others in which a
nearly level mortality is maintained for two months or even longer, and in these
instances it is difficult to locate the week in which the prevalence attained a maximum.
Further, the balance of the epidemics about the late summers of each year
suggests that some climatic condition is associated with the prevalence and in that
case no rigid observance of set times can be expected.