years after notification dominates the percentage surviving later on. For this reason an ... For this review nominal fatality is recorded for the three contrasting.
FATALITY AND
SURVIVORSHIP
IN
PULMONARY
TUBERCULOSIS; A. Study
of
Notified Cases
By MURRAY
in
Jldinburgh
S.fACKER, CM.D.,
:
1936-44^)
D.P.H>.
Departme7it of Public Health and Social Medicine, Edinburgh University In recent years, many reports have been published describing major trends in regard to pulmonary tuberculosis in Scotland. In most cases these reports are based upon mortality or notification data, and almost invariably attention is drawn to the disturbing situation, particularly ?during and after the last war. Daniels (1949), in a comprehensive review of tuberculosis mortality
Europe during and after the second world war, pointed out that Scottish tuberculosis trends merit special study for several important reasons, one of which is the fact that the war-time rise was not followed by a drop in the post-war period. He emphasised the fact that this
in
unfavourable trend was visible to a much greater degree in the large burghs than in the counties. McDougall (1950) in a later report underlined the same unfavourable situation in Scotland. Edinburgh, while not entirely typical of the large burghs in respect
pulmonary tuberculosis mortality, experienced these unsatisfactory war-time and post-war changes. The mortality rate rose during the After dropping to war, reaching a peak of 71 per 100,000 in 1943. 51 in 1945, the rate increased again during the next two years, and of
then declined
the years 1948, 1949 and 1950. The was arrested when the war started. A marked war-time rise occurred and a rate of 142 per 100,000 was reached by 1943. After dropping off slightly toward the end of the war, the rate continued to rise during the post-war years. Aside from actual numbers of notifications increased minor fluctuations, continuously during and after the war. The situation described in terms of mortality and notifications to some extent in
pre-war fall in the notification rate
generally indicates the disturbing war-time and post-war trends of pulmonary tuberculosis in Edinburgh. However, death rates and notification rates considered separately only give the crude outline : they do not give an indication of a number of important aspects such as the lethality of the disease in general or in particular age groups, the prognosis of those actually discovered, the effect of treatment and the like. Many more precise measures have been sought by various workers in studies designed to fill in some of these details. Recent Investigations In recent years, various investigators have used measures of fatality ?and survival in order to throw further light upon the lethality of the
disease itself.
VOL. LVIII. NO.
IO
489
2 K
MURRAY
490
S. ACKER
Stocks (1950), in his analysis of tuberculosis statistics in England and Wales, related deaths to the corresponding notifications of persons among whom the deaths were calculated to have occurred. He showed that the ratio of fatality, calculated according to his method, remained constant at about 50 per cent, from 1922 to 1939. It decreased to 40 per cent, by 1946, and fell to 38 per cent, by 1948. According to Stocks, this decrease was more probably due to the increased notification of non-fatal cases rather than to a fall in the fatality of the disease
generally. Thompson (1943)
carried forward an earlier investigation of the survival of a group of sputum-positive adult cases diagnosed in Durham County between 1928 and 1930. He found that 60 per cent, were alive at the first anniversary following diagnosis, 43-2 per cent, at the second, and only 26-8 per cent, at the fifth anniversary. He also observed that among his group of patients the outlook for those between
twenty and thirty-nine years of age was better than for cases in the older and younger age groups. The survival of cases notified from the City of Aberdeen and recorded by Fraser (1947), was more favourable experience of the Durham County group. Of more than bacteriologically proven cases notified from 1934 to 1943, 70-0 per
Aberdeenshire,
as
than the 1200
alive at the first anniversary, 61-6 per cent, at the second and by the fifth anniversary 44-9 per cent, were still alive. Fraser also found a lower probability of survival among patients over forty in both males and in females and a more favourable chance of survival for males than for females in the 10-19 year age group. All other age-sex differences were not significant. By comparing the patients classified by stage of disease, he found the survival of those who were in stage 3 at the time of diagnosis to be considerably less favourable than those in stages 1 and 2 at diagnosis. Stocks (1944), in a statistical study of wider scope, analysed the rates of dying from pulmonary tuberculosis in England and Wales cent, were
during the inter-war period 1922-39, and calculating further from his data it is found that 77-8 per cent, were alive one year following notification, 68-8 per cent, after two years, and 55-5 per cent, after five years.
Nominal Fatality, Survival after Notification and Year-to-Year Survivorship as Measures of Severity In view of the
importance
of
pulmonary tuberculosis, particularly
among young adults, and of the large number of people who
are
notified annually, information concerning the advance of the disease itself and its lethality becomes a matter of vital importance. For many acute infectious diseases with short incubation periods and durations of illness, a fairly accurate measure of fatality can be obtained by relating deaths to the number of cases in a given series. In the case
FATALITY AND SURVIVORSHIP IN PULMONARY TUBERCULOSIS 491
of pulmonary tuberculosis, in which the incubation period is long and variable and the duration ot illness is usually prolonged, the relationship between deaths and notifications within the same period of time is clearly not true fatality, since the recorded deaths do not correspond to the recorded notifications. Nevertheless, if trends in both notification and mortality rates are comparatively continuous, the ratio of deaths to notifications can provide some measure of severity in terms of nominal fatality. The validity of this index is debatable, but, since it is widely used, it has been recorded here. Since, however, this nominal index is derived from unrelated groups rates
of deaths and notifications, it becomes necessary to attempt to arrive at some measure of the severity of the disease in the individual. Such a measure relating death or survival among actual notified cases is more estimate of
lethality, and, further, can indicate more of advance of the disease by ascertaining the period of survival subsequent to notification. Without precise knowledge of the actual date of onset in established cases of pulmonary tuberculosis (which is difficult, if not impossible, to obtain because of the frequent absence of any symptoms in the earlier stages), it is not possible to establish the actual period between onset and death. However, if the lengths and distributions of the intervals precise closely
as
the
an
rate
between onset and notification do not alter too suddenly or sharply, a relative measure of survival can be obtained by utilising the date of notification as the starting point. There is a certain disadvantage, however, in considering only the percentage surviving for any period of years from notification, namely, that possible changes in any one year are hidden by the cumulative experience over a number of years. This is particularly true in a disease like pulmonary tuberculosis where the mortality in the early years after notification dominates the percentage surviving later on. For this reason an additional study of year-to-year survivorship will clarify possible differences between the dangerous early years and later experience. Sources For this review nominal
of
Data
fatality
and
Method
is recorded for the three
contrasting
Nominal fatality has been obtained by calculating the ratio of each year's deaths to the average of the notifications in the same year and the previous year.* The data on deaths and notifications are taken from the published
periods before, during and after the
figures
in
the Annual
Reports
war.
of the
Edinburgh
Public
Health
Department.
The survival after notification of all cases of pulmonary tuberculosis notified in Edinburgh from 1936 to 1944 is calculated from data derived *
Various combinations of fractions of notifications in the same year and in For this reason years were calculated and all gave very similar results. seemed desirable to use the simplest combination.
preceding
MURRAY S. ACKER
492
frcm the official register of notified cases. The period of survival in case has been taken as the interval between the date of notification and the date of death. The total group of cases notified during the each
nine-year period under review includes a proportion who were notified after death, but the percentages surviving are calculated upon only those patients who were notified while still alive. The observations have been carried forward until 31st December 1950, and therefore the percentages surviving at the 8th, 9th and 10th anniversaries relate notifications during 1936-43, 1936-42 and 1936-41 respectively.. survival experience within specified age-sex groups is also examined. In addition, age and sex comparisons are made of year-to-year survivorship, this being the ratio of the number of patients alive at the end of any one year to the number alive at the beginning of that
only
to
The
Finally, data on survival after notification during the immediate pre-war and the mid-war years are compared. Throughout this study the statistical significance of any percentages which are compared has been tested. The difference between any two percentages of survivors is considered to be highly significant if this year.
difference is three or more times its standard error. A difference which is between 2\ and 3 times its standard error is considered to be probably significant; and if a difference is between 2 and 2\ times its standard error, then it is considered to be
only suggestive.
Nominal Fatality In Tables I and II are shown the nominal fatality ratios for the In Table I total ratios are war and post-war periods. presented and in Table II these are divided into ratios for broad age groups. From Table I it can be seen that the average ratio of nominal fatality in Edinburgh decreased from 61-2 per cent, during 1936-39 to 58-8 per cent, during the mid-war years 1942-44. A further decrease to 47*8 per cent, took place after the war and by 1949 a point was pre-war,
reached about one-third below the ratio for 1936. At the same time, the actual numbers of deaths, except for minor fluctuations, did not change markedly; yet the numbers of notifications increased substantially in each successive period. Differences in
fatality among the various age-sex groups are shown During each of the three periods under review, and in both sexes, the ratio of nominal fatality increases in each successively older age group. The successive decline in nominal fatality during the pre-war, war and post-war periods, noted in Table I, did not in Table II.
to the same extent in all age-sex groups. This decrease is seen have been marked both for males and for females aged 15-34 an