680
TRAN~AC~~IONS
OF THE ROYAL SOCIETY OF TROPICAL MEDICINE
Diagnostic value of bone marrow
J. G. GUERRA-CACERES~,E. GOTUZZO-HERENCIA~, E. CROSBY-DAGNINO~, M. MIRO-QUESADA~ AND
C. CARRILLO-PARODI’
lInstituto de Medicina Tropical Alexander Von Humboldt and 2Department of Medicine, Universidad Peruana
Cayetano Heredia, A.P. 5045, Lima, Peru
Summary
The diagnostic efficacy of bone-marrow culture,
serial blood cultures and agglutination tests was
compared in a prospective study of 60 patients with
typhoid fever, two thirds of whom had received
prior antibacterial therapy. Salmonella typhi was
recovered from marrow cultures in 95 q; of patients
but blood cultures were positive in only 43.3%
(P -Z0.001). Agglutination tests were eventually
diagnostic in 56 *7 “/o of patients, but in only 25 1,;
at the time of admission. If procedures had been
limited to blood cultures and agglutination tests,
diagnosis would have been missed in 21.7 “/& of
cases. The efficacy of marrow cultures was affected
not by the duration of disease but by the extent of
antibacterial therapy before presentation. Bac-
teriological recovery was faster from marrow
cultures.
Introduction
Before the advent of chloramphenicol several
workers advocated, apparently with little success, a
more extensive use of bone-marrow cultures for
accurate diagnosis of typhoid fever (STORTI, 1938;
OTT, 1938: LING et al.. 1940; SACKS & HACHTEL.
194i; PIA~GIO BLANCO -et a1.,‘1942; HIRSO&~TZ &
CASSEL, 1951). In recent years, marrow aspirates
have been claimed to be more efficient than blood
specimens for the isolation of Salmonella typhi in
place-s where prior administration of antibacterials
is a common finding in febrile patients reporting at
local hospitals (SCHLACK et al., 1966; MENDOZA-
HERNAND~Z et ai., 1974; GILMAN et al.; 1975).
A pilot study of 150 patients discharged from our
hospital with a diagnosis of tvvhoid fever showed
thai two thirds a&nowledgid taking antibiotics
before presentation. Serial blood cultures taken on
the day of admission were positive in less than 50 “,;,
the rest having been diagnosed largely by sequential
agglutination tests. These findings prompted us to
evaluate the benefits of introducing marrow culture
as a routine procedure in all patients admitted with
suspicion of typhoid fever. We here report the
results of a six-month diagnostic trial.
Patients and Methods
All patients admitted with a presumvtive diaenosis
of typhoid fever during a six-month period-were
included in this study. Duration of clinical disease
were recorded from day of onset of fever. Detailed
information of any pribr antibacterial therapy was
procured in each case.
AND HYGIENE, VOL. 73, No. 6, 1979
culture in typhoid fever
Two sets of bottles containing either 25 ml of
trypticase-soy broth or diphasic Ruiz-Castaneda
medium (RUIZ-CASTANEDA, 1947) were constantly
available. A set was randomly chosen for each
patient and the following procedures performed
during the first 24 hours after admission:
(9 5 ml blood were taken three times at intervals
of six hours and inoculated into the respective
medium;
(ii) 1 ml of bone marrow was aspirated from the
iliac crest or, exceptionally, the sternum.
0.5 ml was inoculated into the matching
medium and 0.5 ml was used for histologicaT
confirmation of marrow tissue;
(iii) agglutination tests using Difco antigens were
repeated every three days during the hospital
stay. Diagnostic titres were based on findings
in our pilot study and in a matched group of
febrile non-typhoid patients: (a) a reciprocal
titre for the 0 antigen of S. typhi greater than
160 on the initial serum specimen, and (b) a
rise in reciprocal titre for 0 antigen to a
.minimum
.. . . of 160.. .
Additionally, a stool specimen was collected from
non-constipated patients and a urine sample from
patients who manifested urinary symptoms. These
were plated on MacConkey’s and salmonella-
shinella agar.
&spici&s colonies were identified by standard
techniques (EDWARDS & EWING, 1972). Cultures
were not discarded as negative before 10 days.
Positive blood culture in a patient refers to the
earliest bacteriological recovery and definite bio-
chemical and serological identification of S. typhi
from any of the three blood specimens.
Typhoid fever was defined as an acute illness in
a patient with compatible symptoms and signs
accompanied by isolation of S. typhi from at least
one site or by diagnostic titres for the 0 antigen.
Only those patients who had been on antibacterials
for at least 48 hours before admission were con-
sidered as having had prior treatment.
Differences in proportions were analysed by the
standardized normal deviate (u) using binomial
approximation; the McNemar’s test was specifically
used in the case of paired samples. Means were
compared by the t-test (ARMITAGE, 1971).
Results
Procedures were completed as scheduled in 60 of
66 patients admitted with a clinical suspicion of
typhoid fever. The patients studied were 36 males
and 24 females between three and 54 years of age;
 J. E. GUERRA-CACERES et al. 681

Table I-Distribution of patients and frequency of isolation of Salmonella typhi according to medium
used for blood and marrow cultures
Characteristics of Tryp;F:a;;le-soy Ruiz-Castaneda
cultural groups medium

Number 30 (100.0%) 30 (100.0%)

Prior antibacterial therapy 20 (66.7%) 17 (56.7%)
Duration of disease :
< 14 days 20 (66.7%) 19 (63.3%)
> 14 days 10 (33.3%) 11 (36.7%)
Bacteriological recovery :
Blood 12 (40.0%) 14 (46.7%)
Marrow 29 (96.7 %) 28 (93.3%)

Table II-Frequency of isolation of Salmonella typhi from blood and marrow cultures by duration
of clinical disease in patients with typhoid fever
Positive Cultures
Duration of No. of
disease patients Blood Marrow

l- 7 days 141)(lOO.O%) 71 (38 *5 “/u) 13 3 (974%)

8-14 days 251 81 25J
15-21 days 101 71 101
l-(90-5%)
222 days 11 r’ (lOO*O”,;) 4f(524%) 9J
Total 60 (100~0%) 26 (43.3%) 57 (95.0%)

Table III-Relation between prior antibacterial therapy and isolation of Salmonella typhi from blood
and marrow cultures in patients with typhoid fever
Positive Cultures
Prior Antibacterial No. of
Therapy patients Blood Marrow

Nil 23 (100.0%) 13 (56.5%) 23 (lOO*O%)

Single drug 22 (100.0 %) 9 (40*9%) 22 (100.0%)
Combined drugs 15 (100.0%) 4 (26.7%) 12 (80.0%)
Total 60 (lOO*O%) 26 (43.3%) 57 (95.0%)

70% were less than 19 years old. Marrow tissue
was confirmed histologically in all the aspirates. No
complications arose nor were analgesics required
after bone-marrow aspiration in any case. All 60
patients met the criteria for diagnosis of typhoid
fever. S. typhi was isolated from 57 and reciprocal
0 titres of 360 were detected in the other three.
Equal numbers of patients had their blood and
marrow specimens inoculated into trypticase-soy
broth (TSB) and into Ruiz-Castaneda medium
(RCM). Table I shows that the distribution of
patients by prior antibacterial therapy and by
duration of disease was very similar in both cultured
groups. Blood cultures were positive in 40% of the
patients in the TSB group and in 46.7% of those
in RCM group (u = 0.46, p = 0*65), while the
positivity of marrow cultures was 96.7% and
93.3%, respectively (u = O-44, p = 0.66). Given
no significant difference between the groups, either
in composition or in bacteriological recovery from
identical sites, data was pooled for further analysis.
The frequency of isolation of S. typhi from blood
and marrow specimens by duration of clinical
disease is seen in Table II. 57 patients (95 *0 %) had
a positive marrow culture in contrast to 26 (43 *3 %)
who had a positive blood culture (u = 5 -39,
p < 0.001). The difference in positivity was
significant at any period. After the second week of
illness there was an apparent increase in the efficacy
of blood cultures (u = 0.98, p = O-33) and de-
682 DIAGNOSTIC VALUE OF BONE MARROW CULTURE IN TYPHOID FEVER

crease in that of marrow cultures (u = 1.04, 1978). Prevention of life-threatening complications

p = 0*30), which lacked statistical significance. In
no instance was S. typhi isolated from blood while
the simultaneous marrow culture was sterile.
Bacteriological recovery from marrow aspirates
took a mean time of 2 *9 days as opposed to 4.0 days
from blood specimens (t = 2.78, p < 0.01); the
median times were two and four days, respectively.
The relationship between prior antibacterial
therapy and bacteriological diagnosis is shown in
Table III. The fall in efficacv of blood cultures
from 56.596 in untreated patients to 40.99: in
those on single drugs (u = 6.97, p = 0.33) and to
26.7% in vatients on combined drugs fu = 1.82.
p = 6;07) -was not statistically signifi;ant. Marrow
cultures were lOOoh positive in untreated patients
and in those on one drug, but positivity fell to
80 q/, in patients on combined drugs (u = 2.09,
p = 0.04). Marrow cultures were significantly more
efficient than blood cultures in each of the three
groups of patients. Single therapy consisted of
chloramphenicol in 11 cases, ampicillin in four,
tetracyclines in two and miscellaneous in five; daily
doses were more than 1 *O g in the 22 patients. The
three bacteriologically negative patients manifested
having had prior treatment with a combination of
drugs : trimethoprim-sulphamethoxazole, penicillin-
dicloxacillin, and chloramphenicol-paromomycin-
streptomycin, respectively.
Agglutination tests were diagnostic in 15 patients
(25::) on admission-in six of the 39 patients
(15.446) with less than two weeks of illness and in
nine of the 21 (42.8 $6) admitted in the later stages.
Significant titres were eventually found in 34 of the
60 patients (56.776) in a mean time of six days.
12 patients (20 y;), all with positive marrow cultures
but only seven with positive blood cultures, had
persistently negative 0 titres. Had the investigations
been limited to blood cultures and agglutmation
tests. diagnosis would have been vossible in 47
patients (78.39/o), the difference being still signifi-
cant with respect to the efficacy of marrow cultures
(u = 2.25, p = 0.02).
Onlv 21 natients, all of whom eventually had
positive marrow cultures, had a stool specimen
cultured. This was nositive in five cases (23.8O:).
two of whom had- simultaneous negative blood
cultures but diagnostic agglutination titres. None
of the urine cultures taken from 15 patients with
urinary complaints yielded S. typhi. However,
concomittant urinary tract infections with Escher-
ichia coli and Klebsiella enterobacter, respectively,
were identified in two females who had negative
blood but positive marrow cultures.
Discussion
Current clinical patterns of typhoid fever in
endemic areas often pose diagnostic difficulties
even for experienced clinicians (GULATTI et al.,
1968; WICKS et al.. 1971; FLORES-ESPINOZA. 1973).
Promiscuous use -of antibiotics seems to- hinder
bacteriological recovery of S. typhi from blood and
other classic culture sites (SCHLACK et al., 1966;
MENDOZA-HERNANDEZ et al., 1974; GILMAN et al.,
1975), while agglutination tests have a limited value
in these areas (SCHROEDER, 1968; LEVINE et al.,
with anvrovriate theravv (ROWLAND, 1961) devends
largely bn *a prompt and accurate -diagnosis: The
vossibilitv of infection with chloramvhenicol-
iesistant -strains (OVERTURRI et al., 19733 BUTLER
et al., 1973; GONZALEZ-CORTES et al., 1973) further
emphasizes. the need to pursue vigorously the
isolation of S. typhi in all patients suspected of
typhoid fever.
The group we have studied seems fairly repre-
sentative of the type of patients with typhoid fever
seen in urban endemic areas. Approximately two
thirds of them had taken antibacterials before
presentation. Our data demonstrates that a bone-
marrow aspirate, whether cultured in trypticase-soy
broth or in Ruiz-Castaneda medium, significantly
increases the chance of bacteriological confirmation
in a patient with typhoid fever. Marrow cultures
were positive in all the untreated patients and in all
those who had taken one drug alone. Had investiga-
tions been restricted to serial blood cultures and
agglutination tests, the diagnosis would have been
missed in 13 of the 60 vatients (21*6%X The
efficacy of marrow cultures^was not affected’by the
duration of disease, as suggested by GILMAN et aZ.
(1975), but by the extent of prior antibacterial
therapy. Additionally, the organism was recovered
in a shorter time from marrow than from blood
cultures.
Suppression of S. typhi bacteraemia by prior
administration of antibacterials probably accounted
for the low efficacy of serial blood cultures in this
series (43.3 y/o). It is likely that some of the patients
considered as untreated had concealed the fact that
they were taking drugs. This possibility and the
small number of patients in the combined-drugs
group may explain why the fall in efficacy of blood
cultures with increasing use of antibacterials lacked
statistical significance.
The efficacy of stool and urine cultures cannot be
fully assessed because these were performed only in
selected patients. However, no one had a positive
culture from stool alone and all urine cultures from
15 patients with urinary complaints were negative
for S. typhi. The diagnosis of typhoid fever could
have been missed in two females with Gram-negative
urinary tract infections if marrow aspirates had not
been cultured. The association of typhoid fever and
urinary infection seems to be more frequent than
can be explained by the prevalence of asympto-
matic bacteriuria in certain groups (COLAN et al.,
1975).
We have corroborated the limited but con-
tributory value of sequential agglutination tests in
the diagnosis of typhoid fever in endemic areas.
Diagnostic titres were more frequent the longer the
duration of disease, even when antibacterials had
been administered. However, 209; of our patients
had persistent negative 0 titres and the diagnosis
could have been missed in half of them had marrow
cultures not been performed.
Intracellular bacilli have been identified within
mononuclear phagocytic cells in the bone marrow
(QUESADA, 1969),as &ey were previously in intestinal
and mesenteric lesions (ADAMS, 1939). The concen-
tration of S. typhi in marrow -tissue’ is likely to be
higher than in peripheral blood where less than
 J. E. GUERRA-CACERES et 01. 683

10 organisms per ml may be present even in severe endemic for typhoid fever. American Journal of
bacteraemia (WATSON, 1975). The survival of intra- Tropical Medicine and Hygiene, 27, 795-800.
cellular bacilli despite the administration of anti- Ling, C. C., Taur, S. S., Hsueh, I?. C. & Yang,
biotics has been highlighted in a study of 60 S. Y. (1940). Medullo-culture in the diagnosis of
asymptomatic people who, one year after appropriate typhoid and paratyphoid fevers: An analysis of
treatment of bacteriologically proved typhoid fever, 38 cases. Chinese Medical Journal, 57, 11-26.
had stool cultures positive for S. typhi in 3.3% of Mendoza-Hernandez, I?., Terminel-Valenzuela, M.
cases as opposed to positive marrow cultures in & Ruiz-Maya, L. (1974). Experiencias bacterio-
16.6% (MENDOZA-HERNANDEZ et al., 1974). logicas, clinicas y terapeuticas en 1976 cases de
Our study has confirmed that a single bone- fiebre tifoidea. Gaceta Medica de Mexico, 108,
marrow culture is the best procedure for a rapid and 85-92.
accurate diagnosis of typhoid fever, especially in Ott, A. (1938). Uber die Bedeutung der Knochen-
patients on prior antibacterial therapy. The method markskultur fur den Typhus und Paratyphus-
is non-traumatic and safe in the hands of trained bacillennachweis. Klinische Wochenschrift, 17,
personnel. 1475-1476.
Overturri, G., Marton, K. I. & Mathies, A. E.
References (1973). Antibiotic resistance in typhoid fever.
Adams, J. W. (1939). Intracellular bacilli in New EnglandJournal of Medicine, 289,463-465.
intestinal and mesenteric lesions of typhoid fever. Piaggio-Blanco, R. A., Paseyro, I’. & Sanguinetti,
American Journal of Pathology, 15, 561-566. C. M. (1942). El medulocultivo coma metodo de
Armitage, I?. (1971). Statistical Methods in Medical diagnostic0 en la fiebre tifoidea. Archives Uru-
Research (1st edit.). Oxford: Blackwell Scientific guayos de Medicina, Cirugia y Especialidades, 5,
Publications, DD. 116-131. 413-423.
Butler, T., Arnold, A., Linh, N. N. & Pollock, M. Quesada, N. (1969). La celula tifica en la medula
(1973). Chloramphenicol-resistant typhoid fever osea. Archives Peruanos de Patologia y Clinica,
in Vietnam associated with R. factor. Lancet. ii, 23, 181-198.
983-985. Rowland, H. A. K. (1961). The complications of
Colan, A. R., Gross, D. R. & Tamer, M. A. (1975). tvnhoid fever. ‘fournal of Troaical Medicine and
Typhoid fever in children. Pediatrics, 56,606-609. Higiene, 64, 143-152. s A
Edwards, I’. R. & Ewing, W. H. (1972). Zdentzjka- Ruiz-Castaneda, M. (1947). A practical method for
tion of enterobacteriaciae (3rd edit.). Minneapolis : routine blood cultures in brucellosis. Proceedings
Burgess Publishing Company, pp. 7-20 & 146- of the Society for Experimental Biology
-_ and
258. Medicine, 64,114-115. -
Flores-Espinoza, J. (1973). Tifoidea en Mexico: Sacks, M. S. & Hachtel, F. W. (1941). A note on
Analisis clinic0 de la epidemia de 1972. Gaceta the bacteriologic culture of bone marrow in
Medica de Mexico, 106. 11-36. typhoid fever. Journal of Laboratory and Clinical
Gilman, R. H., Terminel, M., Levine, M. M., Medicine, 26, 1024-1029.
Hernandez-Mendoza, I’. & Hornick, R. B. (1975). Schlack, L., Pino, M. & Wiederhold, A. (1966).
Relative efficacy of blood, urine, rectal swab, El mielocultivo en el diagnostic0 de fiebre
bone marrow, and rose-spot cultures for recovery tifoidea y paratifoidea : Analisis comparative de
of Salmonella typhi in typhoid fever. Lancet, i, 135 cases a su ingreso hospitalario. Revista
1211-1213. Chilena de Pediatria, 37, 213-220.
Gonzalez-Cortes, A., Bessudo, D., Sanchez-Leiva, Schroeder, S. A. (1968). Interpretation of serologic
R., Fregoso, R., Hinojosa, M. & Becerril, P. tests for typhoid fever. Journal of the American
(1973). Water-borne transmission of chloram- Medical Association, 206, 839-840.
phenicol-resistant S. typhi in Mexico. Lancet, ii, Storti, E. (1938). Bakteriologische und Parasit-
605-607. ologische Untersuchungen des Knochenmarks.
Gulatti, P. D., Saxena, S. N., Gupta, I’. S. & Klinische Wochenschrijt, 17, 596-599.
Chuttani, H. K. (1968). Changing pattern of Watson, K. C. (1975). Laboratory diagnosis of
typhoid fever. American Journal of Medicine, typhoid fever. Lancet, i, 1377.
45, 544-548. Wicks, A. C. B., Holmes, G. S. & Davidson, L.
Hirsowitz, L. & Cassel, R. (1951). Sternal marrow (1971). Endemic typhoid fever: A diagnostic
cultures in typhoid fever. British Medical pitfall. Quarterly Journal of Medicine, New Series,
Journal, i, 862-863. 40, 341-354.
Levine, M. M., Grados, O., Gilman, R. H., Wood-
ward, W. E., Solis-Plaza, R. & Waldman, W.
(1978). Diagnostic value of the Widal test in areas Accepted for publication 2nd May, 1979.