adult age. On the other hand, where only a smaller propor-
tion of the natives become infected in childhood, many of the
remainder may be attacked for the first time at later ages.
This must be especially the case where there are seasons, such
as a sharp winter or a hot dry season, during which infection
ceases. I notice, for instance, that the law of Daniels is not
much insisted upon by Italian workers.
From these considerations an important deduction, exactly
opposite to the common opinion, may be drawn. Where there
is much malaria, so many of the native adults will be immune
that few of them should show signs of infection ; and, conversely,
where the adults are frequently attacked, the amount of malaria
should be comparatively slight. In other words, if the malaria
index amongst adults (parasites, fever, splenomegaly) is high, the
malaria rate of the locality is probably low ; and where the adult
hidex is low, the rate is probably either high or low. If, there-
fore, a considerable number of adults are coming to hospital
with fever or enlarged spleen, we should believe, not that the
malaria rate of the population is usually high, but that it is
usually low — comparatively, of course. If malarial immunity
were complete, and if every one were infected in childhood, no
one would be ill afterwards. For example, if many soldiers in
a regiment are attacked, we may infer that they have not been
recruited from very malarious districts.
Unfortunately, however, we are not sure {a) how far immunity
derived from a single infection protects against subsequent infec-
tions, and {b) how far immunity against one species of parasite
protects against other species. Thus many partially-immune
adults may suffer from short attacks of fever due to re-infection,
especially by parasites of a species different to the one to which
they have become habituated.
Practically, therefore, for the purpose of the measurement of
240 MALARIA IN THE COMMUNITY [Sect.
malaria, it is best to consider only the malaria index among
children. Among adults the malaria index cannot yield very
definite results : if high, it indicates that the malaria rate is
comparatively low ; if low, that the malaria rate is either very
high or very low.
The study of large numbers of children is required to
establish the age at which the malaria index is highest. Most
of the figures suggest that that age is between one and three
years. Similar studies are required to determine the age of
maximum average spleen, and of maximum enlarged spleen.
(lo). Determination of the inoculation rate. — According to
section 26, the malaria rate of a locality is the percentage of the
people who, at the moment of enquiry, contain plasmodia. But
as many of these people must have been inoculated weeks,
months or years previously, the malaria rate does not directly
inform us as to what is going on at that moment — it instructs
us regarding the past rather than the present. We may wish to
know how many people are actually being inoculated during
the week, month or year of observation ; and if we can ascer-
tain this we can obtain a measure of the actual nialariousness
of the locality at the moment when we are investigating the
malaria in it.
This could be ascertained experimentally if we could dis-
tribute throughout the population a large number of healthy,
non-immune immigrants, and could then observe accurately
how many of them become infected within a given period : or
if we could observe a large number of non-immune immigrants,
and learn how long, on the average, they remain healthy. Such
studies can sometimes be made with adults ; but, unfortunately,
Nature herself is always carrying out the experiment for us by
providing a constant influx of non-immune babies, and we have
only to ascertain the proportion of them which have become
infected within a week, month or year after birth.
In his original article [1895], Daniels stated that in 43
infants who had died under one month after birth pigmentation
3i] THE INOCULATION RATE 241
of the spleen could not be found ; that it occurred in I2"5% of
16 infants of one to six months of age, in iiy of 14 infants of
six to twelve months, and in 50% of 1 1 infants of one to two
years of age. From these figures we gather that in British
Guiana at that time the chances were small that infection should
occur within one month, and increased by about 2^ roughly for
every month after birth — so that the chances were that about
50% of all newcomers would become infected in two years.
Meanwhile, however, immunity would tend to reduce the cases.
Of course, with such small actuals, the error of random
sampling is very large.
R. Koch in his third Report [1900] states that in nine
localities in Java, the parasites were found in 6y, or 11^, out
of 619 infants of under one year examined ; and in 47, or 8"i^,
out of 574 children of over one year of age.
The infantile malaria index can easily be obtained by the
parasite test or spleen test, or both combined, and, with large
actuals, will give a good direct measure of the inoculation rate.
But it will be still more useful if compared with rates calculated
according to sections 27 and 28.
As before, let p be the total population, and m and m^ the
proportion of infected persons at the beginning and the end of
the enquiry respectively. Then at the beginning of the enquiry
there were nip infected persons and {\—iii)p healthy ones.
Now let rnip be the number of infected persons who recover
during the enquiry — suppose that r=\l^ and that the enquiry
lasts one month; and let I{i—m)p be the number of healthy
persons who become infected during the enquiry. Then by
section 28, equation 4,
m^p = mp + /( I — in)p — rmp
or, /( I — m)p = in^p — mp + rmp
Thus suppose that, in a village of 1,000 people, there were
250 infected persons at the beginning of the month and 300
infected persons at the end of it. Then, obviously, though
50 infected persons recovered during the month, there was
Q
242 MALARIA IN THE COMMUNITY [Sect.
nevertheless an increase of 50 infected persons at the end of
it; so that 100 healthy persons must have become infected
during the month. That is, I{\ — ni)p — \oo.
Here I={nt~^ — m-\-rin)j{i—m), and is the inoculation ratio;
that is, it gives the proportion of people, healthy or unhealthy,
inoculated during the enquiry. In the above example /= 10/75
= 0'i33. Out of the 750 originally healthy people in the
village 100 were newly infected, and out of the 250 previously
infected people 33 were reinfected {Imp gives the number of
reinfections); so that altogether 133 persons out of 1,000 were
inoculated during the month. The inoculation rate of the
village was /x 100= 13*3%; that is, the chances of a new-
comer in the village, or of a new-born child, becoming infected
during the month was as 13*3 to 100.
From equation i of section 27, we might infer roughly that
there were about 106 Anophelines to each person in this village.
In fact, on comparing that and the following section, we see
that /= b^saint.
The fractions m, m^, I and r are connected by the equation
m^ = m + I{i—m) — rm; and if we can ascertain any three of
them we can calculate the fourth — which enables us to com-
pare observed and calculated estimates. If the malaria is
static, so that nt-^ = m, then I{i—vi) = rm. We may now esti-
mate / from the infantile malaria index, w from the general
malaria index, and r from statistics, and check our results by
comparison. Suppose, for example, that r= 1/5 a month, and
that 4% of the infants are infected in a month ; then the
general infection rate of the locality should be about i6-6%.
Comparing these figures with observed estimates, we shall
probably arrive nearer to the truth than if we had relied on
calculation or observation only.
(11). Practical conclusions. — Having considered the various
methods of measurement which may theoretically be adopted,
it remains for us to select the best ones for practical use.
A. First suppose that we wish to form an Actual Estimate
31] PRACTICAL CONCLUSIONS 243
of the Malaria Rate ; that is, of the percentage of people who
contain plasmodia at the moment of observation. We may
employ {a) the parasite index, {b) the spleen index, and {c) the
combined methods.
{a) The parasite index gives exact information only as to
the lowest proportion of infected persons. The
proportion of those who contain plasmodia in
numbers too few to be found during the time
allotted for the test may be very considerable,
and has not been correctly estimated. The test
requires a skilled observer, and the expenditure of
so much time that it can be applied only to com-
paratively few people. The error of random
sampling is therefore generally very large.
{b) The spleen index (by palpation only) can be quickly
ascertained, by almost any intelligent person, for
large numbers of people — so that the error of
random sampling may often be reduced nearly
or really to zero. A certain proportion of infected
persons have no palpable enlargement of the spleen,
and possibly in others with palpable enlargement
(especially with great enlargement) the parasites
may have died out in consequence of immunity ;
but by the penultimate paragraph of (2), these two
sources of error (amounting to say 20% each) will
tend to annul each other, so that the spleeji index
should be a nearly correct measure of the true malaria
rate. Quinine, local conditions and other diseases
(especially kala-azar) may affect the spleen index,
but in most countries probably only to a small
extent. Percussion of those who show no palpable
enlargement will reveal a greater proportion of
cases ; but with this test the infected non-spleen
ratio will not be balanced by the non-infected spleen
ratio — so that percussion may give a ratio further
244 MALARIA IN THE COMMUNITY [Sect.
from the truth than that given by palpation alone.
Percussion will also demand more time and skill.
(c) The combined methods (on the same persons) will yield a
greater malaria index than each method by itself —
often a much greater one than that given by the
parasite index by itself, and a considerably greater
one than that given by the spleen index. If, how-
ever, Laveran, Thayer, Mannaberg and others are
right in their opinion that some enlargement of
the spleen can be detected by percussion in almost
all infected persons, then the combined parasite
index and spleen index (by palpation alone) can
yield little more information than that given by
a careful spleen rate ascertained by percussion.
The combined method will, however, require much
more time, and will give no exact information
regarding the non-infected spleen ratio.
Subject to correction, then, I conclude that, as a general
rule, the spleen index (obtained by palpation only) is by far
the best method of measurement. It requires no great skill ;
it can often be applied to almost all the people in a place ;
and it should yield by itself a very correct measure of the
actual malaria rate. As a rule, local conditions, race and
other diseases are likely to cause a percentage of error far
smaller than that due to the insufficient random sampling
which must almost always attend the laborious estimation of
the parasite index (the advocates of which frequently overlook
this important point). But of course, in the presence of other
widespread causes of splenomegaly (kala-azar or some unknown
cause), the spleen index by itself will not be so reliable.
Practically, therefore, it comes to this, that we must avoid
two principal sources of error, (a) the possibility that the
local splenomegaly may be due to other causes than malaria,
and ih) the great error of insufficient random sampling. For
the former purpose we must first satisfy ourselves, by the study
3i] PRACTICAL CONCLUSIONS 245
of hospital records (especially of autopsies), by consultation
with local physicians, and by the blood examination of selected
cases, that the local splenomegaly is really due to malaria. For
the second purpose, if the first point is established, we should
rely upon a large spleen census. The time required for a
laborious study of the parasite index will generally be much
better spent in extending the spleen census so as to avoid
large error of sampling ; and the combined method will only
occasionally be required for some special purpose, such as
ascertaining the infected non - spleen ratio, or the infantile
malaria index.
We must, of course, always consider the season, the time
allotted for the work, the amount of assistance available, and
the population and area to be examined. For large populations
the parasite index becomes almost impossible. Not the least
advantage of the spleen index is that it can be applied equally
well to a population of any size.
B. TJie actual infantile malaria rate. — This is especially
valuable by (10) for determining the inoculation rate. It can
often be obtained among infants on estates and factories, in
soldiers' families, and in many villages. As, however, infants
up to one or two years of age are not always very numerous
among small populations, and the information required is
important, it is generally advisable and possible to use com-
bined methods for examining them. The ages should be
obtained as accurately as possible for each month for two
years, and the infants examined as regards spleen, parasites,
fever, anaemia and history of illness.
The information obtained should be compared with the
formula at the end of (10). If we assume, as argued above,
that the general spleen index practically coincides with the
general malaria rate, and denote the spleen ratio by S and
the infantile ratio by /, then the formula becomes
S^^S-^l{\-S)-rS
246 MALARIA IN THE COMMUNITY [Sect.
If 5^ = 5, r=i/S a month, and / is a monthly infantile rate,
then we should have /=5/5(i-5).
It is, of course, best to examine, if possible, all the infants
and to compare their numbers at different ages with the
numbers given by, or calculated from, the last census of the
population.
C. Comparative estimates from time to time or place to
place. — For these we adopt the same methods according to
the same principles. But we can now often utilise the valuable
information given by the attendances at hospitals and dispen-
saries (5). That is, we obtain the ratio of attendances for
malaria to attendances for all causes, and compare them from
time to time or place to place.
Remember that, for comparative estimates, ratios of indices
should, ceteris paribus, be equivalent to ratios of actual rates —
which makes the task much easier.
The conduct of the malaria census will be referred to in
connection with the whole campaign (section 40).
32. The Mortality and Cost of Malaria. — A little considera-
tion will convince the reader of the difficulty of this part
of the subject. The term malarial death-rate should mean
the percentage of persons who die of malaria. Such rates
can easily be obtained in the case of plague, cholera and
other diseases in which the cause of death can be generally
assigned with certainty ; but malaria is a benign and protracted
disease which is often complicated and terminated by other
maladies, such as pneumonia, infantile diarrhoea, dysentery,
ankylostomiasis, and so on ; and it is often, perhaps usually,
impossible to say whether death has not been due as much
to one of these complications as to the original infection.
Even with troops and prisoners the cause of death in such
cases is often ascribed to one or the other cause on the
evidence of the most predominant symptoms at the end.
Even if we admit the correctness of the diagnosis as regards
32j MORTALITY OF MALARIA 247
deaths in hospital, still the hospital case mortality is no correct
guide to the total case mortality, because only the more acute
cases among the general population come to hospital at all.
Probably the great majority of malaria deaths occur among
the children of the poor, and in the tropics these are seldom
brought to hospital, or even attended by medical men. The
fact that in most tropical countries the cause of death is not
necessarily certified by medical men adds still more to our
difficulties. Indeed the so-called malaria death-rates given in
many statistics are not even worth looking at.
The parasites of malaria cause death in two ways : (a) by
direct action, and (d) by enfeebling the constitution.
Deaths by direct action probably occur only in " pernicious
attacks." Celli (section 20 (6) ) records that 330 pernicious
attacks occurred in 33,507 cases — about 1%, and of these
probably only a fraction were fatal.
W. H. Deaderick collects many useful facts in his text-
book [1909], and gives records of case mortality amongst them.
Out of 5,109,001 cases collected from twenty-five writers,
148,055 or 2-89% died (I judge that these were mostly severe
cases treated in hospital). Out of 27,039 cases of perm'a'ous
malaria, recorded by thirty observers, 7,205, or 2&6'y^ died.
Out of 6,037 cases of blackwater fever (one of the most
pernicious forms), 1,268, or 21% died.
In India for 1907 we have the following figures (combining
intermittent and remittent fevers) : —
Strength.
Admissions.
Deaths.
Malarial
mortality%.
Case
mortality%.
European troops .
69,332
10,662
14
•020
•13
Native troops
126,392
28,432
84
•066
•30
Prisoners
93.264
17,841
88
•094
•51
The highest malarial mortality here given, that of the
prisoners, may be put at about i/iooo; but we can see at
once that this is sure to be too low an estimate for a large
general population. Troops and prisoners live in cleanly sur-
roundings, are well housed, well fed, well treated in hospital,
248 MALARIA IN THE COMMUNITY [Sect.
and (still more important) are all adults. Moreover, in my
experience, medical men rarely attribute death to malaria
alone unless there are no very marked complications. The
rate amongst a large, poor, general population, including
children, is therefore likely to be several times greater. But
we have no figures to indicate the difference.
As just mentioned, the deaths attributed in public statistics
to malaria are generally worthless in the tropics, as they are
not even certified to by medical men. They are often much
too high, because the perplexed registrar has no other cause
to give when dysentery, cholera, fits or debility fail him. For
example, in Mauritius the deaths attributed to malaria con-
stituted 14 per mille, or about half of the total mortality. In
1899, however, the medical certification of deaths was enforced
in two districts of the island — with the result that the mortality
ascribed to malaria fell in them at once from 45^ of the total
deaths to only 25^. From this we calculate that quite 20^ of
the general mortality was wrongly ascribed to malaria. But
even with this correction many sources of error remain.
Is there no better criterion of the true malarial death-rate.?
In most civilised countries to - day the total death - rate is
correctly reported : we may therefore obtain useful information
by comparing the total mortality in various areas which differ
in possessing or not possessing malaria, but which are alike
in other respects. Take, for example, Mauritius (malarious)
and the neighbouring Seychelles Islands (not malarious).
n- •,• f Population
Mauritius ^ ^ ^ ,
I Death-rates
Seychelles] P^P"';^^^^"
I Death-rates
Difference
For the seven years 1897 to 1903 the mean death-rate in
the Seychelles was only lyi per inille against 347 in Mauritius.
If the whole of this great difference is to be ascribed to malaria
1904
1905
1906
Means
387,395
386,128
383,206
385,576
32'2
40 "6
40'o
37-6
20,418
20,767
20,976
20,720
i6-i
15-0
i6-5
15-9
i6-i
25-6
23-5
217
32] MORTALITY 249
alone, we must suppose that the true malarial death-rate in
Mauritius may average something like 20 per mille per annum
or more — that is, a mortality exceeding the total mortality
from all causes in the Seychelles and in most British towns.
The figures are accurate, the deaths being carefully registered
both in the Seychelles and in Mauritius, and the islands have a
very similar climate. But in the mountainous Seychelles there
are few or no Anophelines as in Mauritius. The population of
the former, however, is of negro origin, and in the latter of
Indian origin and much more dense.
The following are some figures for ten other neighbouring
islands of the Indian Ocean — all not malarious. The popula-
tions were those of the census taken in 1901, and the death-
rates were the averages for the following five years. Rodrigues,
3,437. I9'3; ^i^go Garcia, 526, 27-0; Agalega, 327, 31-2; Peros
Banhos, 184, 24-0; Coetivy, 143, 36*4; Salomon, 119, 32*0; Six-
Islands, 117, 29-0; St Bratidon, ?,y, 55*3; St Jean de Nove, 75,
i6'0 ; Eagle Island, 74, 43-0. Total population, 5,134. Average
death-rate, 31-5. Owing to the smallness of the population in
the lesser islands the statistical error is large; but the mortality
in some of them, in spite of the absence of malaria, appears
to be as great as that of Mauritius. There is, however, an
important difference. In Mauritius, the Seychelles and
Rodrigues good medical attendance is available, but this,
I understand, is not the case in the other islands, where
syphilis, beri-beri, infantile tetanus, bowel complaints, ovarian
and uterine diseases abound, and deaths in child-birth are
common. Excluding these islands, therefore, the difference as
regards the mortality between Mauritius with malaria, and
Rodrigues and the Seychelles without it, remains a striking
one. The large malarious island of Reunion (negroid popula-
tion) has a mortality similar to that of Mauritius.
The entry of malaria into Mauritius in 1866 is interesting
in this connection (section 30 (21)). All deaths in the island
have been registered since 1831. In the thirty-six years from
250 MALARIA IN THE COMMUNITY [Sect.
183 1 to 1866, the average death-rate was 35*3 per mille in a
population varying from 93,000 in 1831 to 365,000 in 1866.
In the next five years the death-rates were 120*5, 567, 35'0,
22*6 and 44' i respectively; and in the forty years from 1867
to 1906 they averaged 357 — almost the same as before the
entry of malaria. Thus, apparently, while the disease caused
a great increase of mortality for two years after its entry, it
has caused no increase whatever since then ! In my report
[1908], however, I have discussed this anomaly. As Dr
Meldrum suggested, the extraordinary low mortality of 1870
was possibly due to the death from malaria of many persons
in 1867 and 1868 who would otherwise have died in 1870.
The high death-rate before the entry of malaria was probably
due to many epidemics of measles, small-pox, relapsing fever
and cholera, and to the immaturity of medical and sanitary
practice in those days. Since the entry of malaria there have
been few of such epidemics, and medical and sanitary practice
have greatly improved, while vaccination has been rendered
compulsory — so that the death-rate should have fallen to the
low level found in Rodrigues and the Seychelles. Hence we
may perhaps assume that it has not fallen because of malaria.
In other words, the disease has counterbalanced all the successes
of medical science against other maladies.
The late Dr Meldrum, Director of the Meteorological
Observatory in Mauritius, and an experienced statistician, study-
ing the monthly death-rates from all causes in Mauritius before
and after the entry of malaria, observed that they differed in
an important particular. Before the malaria entered they
remained roughly the same from month to month ; after it
entered the curve rose markedly during the malarious months.
In the Seychelles and Rodrigues the monthly rates still remain
uniform as in Mauritius before malaria entered. The following
table gives the population and number of deaths.
S2J
M(
aRTALI
TY
251
Average
population.
Januar)'.
February.
March.
April.
May.
June.
Seychelles.
1901-04
19,442
31
28
27
30
26
33
Mauritius.
1861-66
345,275
1,003
936
987
920
972
898
1870-89
353,958
790
809
1,005
1,043
1,118
1,059
1904-06
376,974
1,012
981
1,307
1,314
1,357