What proportion of Salmonella Typhi cases are detected by blood culture? A systematic literature review
© The Author(s) 2016
Received: 24 February 2016
Accepted: 9 May 2016
Published: 17 May 2016
Blood culture is often used in definitive diagnosis of typhoid fever while, bone marrow culture has a greater sensitivity and considered reference standard. The sensitivity of blood culture measured against bone marrow culture results in measurement bias because both tests are not fully sensitive. Here we propose a combination of the two cultures as a reference to define true positive S. Typhi cases. Based on a systematic literature review, we identified ten papers that had performed blood and bone marrow culture for S. Typhi in same subjects. We estimated the weighted mean of proportion of cases detected by culture measured against true S. Typhi positive cases using a random effects model. Of 529 true positive S. Typhi cases, 61 % (95 % CI 52–70 %) and 96 % (95 % CI 93–99 %) were detected by blood and bone marrow cultures respectively. Blood culture sensitivity was 66 % (95 % CI 56–75 %) when compared with bone marrow culture results. The use of blood culture sensitivity as a proxy measure to estimate the proportion of typhoid fever cases detected by blood culture is likely to be an underestimate. As blood culture sensitivity is used as a correction factor in estimating typhoid disease burden, epidemiologists and policy makers should account for the underestimation.
Typhoid fever is a fecal oral transmitted systemic disease caused by the Salmonella enterica serovar Typhi. Culturing the bacteria from body fluids is the definitive test for the diagnosis of typhoid fever although inconclusive serological methods such as Widal test are commonly employed in many health care settings [1, 2]. The culture of S. Typhi can be done from many body fluids such as blood, bone marrow, urine, rose spot extracts, duodenal aspirates and stool, while the blood culture remains the mainstay of definitive diagnosis [1, 3]. However, blood culture sometimes does not identify the bacteria even if it exists in the blood because of many procedural and technical issues. It is known that the maximum blood culture yield will be achieved when bacteremia is at peak such as in the first to third week from the onset of the illness  The quantity of blood sample collected may play a role, with a higher quantity more likely to give bacterial growth. Specimen collection, storage and transportation condition is likely to affect blood culture yield besides the culture media used [1, 4, 5]. Furthermore, febrile patients often take antibiotics; self-administered, prescribed or un-prescribed, likely to reduce the possibility of bacterial growth because antibiotics may inhibit the growth of S. Typhi. Considering the sub-optimal diagnostic yield of blood culture, studies that estimate typhoid fever disease burden tend to apply “a blood culture sensitivity correction factor” to account for missed cases [6, 7].
The sensitivity of a test is defined as the probability that the test correctly classifies people with disease as positive . In patients with typhoid fever, the blood culture sensitivity measures the proportion of the S. Typhi cases detected by blood culture compared to an independent test, commonly bone marrow culture. A previous literature review on the sensitivity of blood culture suggested 40 to 60 %  acknowledging the widely accepted sensitivity rate of 50 %  which is often used as a correction factor in many disease burden studies [6, 7, 10, 11]. The blood culture sensitivity is measured against bone marrow culture which is often considered superior because it can yield bacteria among those who took antibiotics and even after the bacteremia subsides in the blood [1, 12]. Ironically, bone marrow culture is also subject to same methodological and technical limitations as blood culture and misses some typhoid fever cases besides posing clinical difficulties as an invasive procedure. There are a few studies that document typhoid fever cases that are bone marrow negative and blood culture positive [13–15]. One of the clinical reviews has suggested a sensitivity of 40–80 % for blood culture and 55–67 % for bone marrow culture . This indicates that the current practice of using blood culture sensitivity as a proxy to represent proportion of cases detected in disease burden studies has intrinsic bias of using a sub-optimal reference test against which it is measured.
Besides being used as a correction factor in the measurement of disease burden estimates which play crucial role in policy making, the proportion of typhoid fever cases detected is important for clinicians. Even if the blood culture is negative for typhoid fever, the person still may be suffering from the disease and need treatment. Thus understanding how many typhoid fever cases are actually detected by blood culture has critical importance for clinicians who would treat the patients, epidemiologists who estimate the disease incidence and policy makers who would use the data for decisions on control measures. As we argued that the blood culture sensitivity measured against bone marrow culture is not the true measure of the proportion of typhoid fever cases detected, here we present a new method. This method will be helpful in measuring the precise proportion of typhoid fever cases detected when blood culture is deployed as a diagnostic test.
First, we chose two tests, blood culture and bone marrow culture to define an algorithm to estimate the proportion of bacteremic typhoid fever cases identified by blood culture. As noted before, both tests do not identify all typhoid fever cases  and hence choosing one of them as a reference standard to compare against the other would be an imperfect estimation of the proportion of cases detected. Therefore, we considered a composite reference standard of two tests to define true positive cases based on epidemiological principles [16, 17] as described below.
General decision matrix for blood and bone marrow diagnostic test
FPBC OR FNBMC
FNBC OR FPBMC
ProCase i = Proportion of typhoid fever cases detected by culture technique i in comparison to true positive cases detected based on both tests i and j.
ProCase BC vs BMC = Proportion of typhoid fever cases detected by blood culture in comparison to typhoid fever cases detected by bone marrow culture.
TP i = True positive cases by culture technique i (culture test shows bacterial growth by culture technique i when the bacteria is present in the sample based on culture technique j).
FP i = False positive cases by culture technique i (culture test shows bacterial growth by culture technique i when no bacteria was detected in the sample by culture technique j).
FN i = False negative cases by culture technique i (culture test shows no bacteria growth in the sample by culture technique i when bacteria is present in the sample based on culture technique j).
TP j&i = True positive test for both culture techniques j and i (culture test shows bacterial growth in culture technique j when the bacteria is present in the sample based on culture technique i).
FN j = False negative cases by culture technique j (culture test shows no bacteria growth in the sample by culture technique j when bacteria is present in the sample based on culture technique i).
BC = Blood culture test.
BMC = Bone marrow culture test.
Inclusion and exclusion criteria
Listed in PubMed, Embase database, WHO or PAHO databases
Published before 31st December 2013
Conducted in human subjects
Published in English language
Collected blood and bone marrow samples from same patients for S. Typhi detection
Identified by search terms defined in the text
Study design: Laboratory surveillance
Papers that do not distinguish S. Typhi from S. Paratyphi in blood culture results
Last, the selected publications were reviewed to calculate true positive, false positive, true negative and false negative cases from blood and bone marrow culture results in the same patients. We estimated the case weighted mean proportion of S. Typhi detected from selected studies using random effects model. The observations on culture media, volume of blood sample collected and duration of illness were presented descriptively.
S. Typhi isolation using blood and bone marrow cultures as sensitivity standard from 10 locations
No. of subjects included
No. of subjects tested by both BC and BMC
S. Typhi detected by BC or BMC
Total S. Typhi recovery from BC [x] (a + b)
Total S. Typhi recovery from BMC [y] (e + f)
Culture media used
Zaria, Nigeria 
B and BM: thioglycolate broth
B: 2 ml; BM: 1–2 ml
Transvaal, S. Africa 
Oxbile and nutrient broths
Karachi, Pakistan 
B and BM: thioglycolate/BHIB
B: 5 ml; BM: 0.5–1 ml
Semarang, Indonesia 
B and BM: oxgall broth
B: 3 and 10 ml; BM:1 ml
Semarang, Indonesia 
Luria-Betani (LB) broth
B: 8–10 ml; BM: 1–2 ml
Jakarta, Indonesia 
B: 3 ml; BM; 0.5–0.8 ml
Dong Thap, Vietnam 
B: oxgall, oxbile, BHIB
B: 5 and 15 ml; BM: 1 ml
Mexico City, Mexico 
B: peptone broth; BM: Peptone/Ruiz-castenada
B: 2 ml; BM: NP
Lima, Peru 
B&BM: Trypticase-soy broth/Ruiz-Castaneda
B: 5 ml; BM: 1 ml
Lima, Peru 
B: 3 ml; BM: 0.5 ml
The recovery of S. Typhi varied depending upon the bacteriological culture media used [21, 22], the volume of blood samples collected  and duration of the disease [15, 24]. The Luria-Betani (LB) broth media resulted in better recovery compared to Difco Oxgal with Selenite F broth media (44/54 vs. 19/43 respectively) [15, 20]. Nevertheless the lower yield from Difco Oxgal broth could be because many studies used blood volumes of less than 3 ml and the same media showed improved recovery where ≥5 ml blood was used [14, 22]. Some studies showed that the collection of a 15 ml of blood sample increased the sensitivity of blood culture comparable to that of the bone marrow culture [9, 22]. The earlier a patient reports to a health facility, the better the recovery of S. Typhi organism in blood cultures and bone marrow cultures [14, 20, 22].
This review presents a summary of the proportion of S. Typhi cases detected by blood culture based on the combined results of blood and bone marrow culture in the same individuals. The results indicate that 2 out of 5 people infected with S. Typhi would remain undiagnosed if blood culture were deployed as the diagnostic test. The results also imply that even if bone marrow culture were used, 1 in 25 people infected with S. Typhi would remain undiagnosed. Clinicians should be aware and remember these missed diagnoses in case management. It is likely that even the combination of blood culture and bone marrow will not diagnose all patients with typhoid fever but in the absence of a perfect reference standard test this is difficult to confirm. Most importantly, the findings indicate that the proportion of typhoid fever cases detected by blood culture is lower than the blood culture sensitivity measured against bone marrow culture. This has important public health ramifications in typhoid fever control because blood culture sensitivity is often used as s a proxy for the proportion of S. Typhi detected in correcting disease burden estimates.
Previous global disease burden estimates [10, 11] and incidence studies [6, 7] have used 50 % blood culture sensitivity as a proxy measure for the proportion of cases detected by blood culture while this review suggests a higher value of 61 %. If this higher proportion were used as a correction factor in estimating the incidence, the disease burden decreases. In recent typhoid disease burden estimations in low and middle income countries, the authors have clearly demonstrated these differences through a sensitivity analysis. The study estimated 25.3 million cases when 50 % blood culture sensitivity correction was used which declined to 20.6 million cases when 61 % blood culture correction was deployed . Such variation in the disease burden estimation may influence global policy and financing decisions, hence careful considerations in using these results are necessary. Notably we argue that the blood culture sensitivity should not be used as a proxy for the proportion of S. Typhi cases detected because it undermines the disease burden estimates. In this review we found blood culture sensitivity was 66 % using bone marrow culture as the reference standard. If we use this sensitivity in previous disease burden estimates , the typhoid fever burden in low and middle income countries decreases by 1.5 million; from estimated 20.6 million to 19.1 million.
Our study has several limitations which should be considered carefully before applying the summary information to clinical, epidemiological and policy decisions. First, it is well known that the bacterial growth in culture is dependent on the time from the onset of illness and sample collection; greater recovery was observed in early bacteremic phase. We could not account for time factor in our analysis as such detailed information was unavailable in the papers identified. It is possible that some studies could have collected samples early after the disease onset and others in a later stage resulting in inter observational biases. Had we known the time of sample collection for all studies, we could have presented sensitivity as a factor of time. Second, sampling the same febrile case in our analysis may not mean that blood and bone marrow samples were collected simultaneously. We only know that same person had provided both blood and bone marrow culture samples. As commonly practiced, bone marrow sample might have been collected in a later stage of the disease management or after antibiotic prescription. In this case, it could be possible that those who were tested negative for blood culture might have been sampled for bone marrow which would create a blood culture sensitivity underestimation bias. Third, the volume of specimen collected will have differed between blood and bone marrow and between different studies which might have introduced the bias in bacterial isolation as larger the volume of the sample, higher the yield would be. Studies of the quantitative bacteriology of the blood and bone marrow suggest that the bacterial counts in bone marrow may be 10-fold higher than blood so larger volumes of blood maybe comparable with smaller volumes of bone marrow aspirate . Many studies identified in this review were conducted several decades ago and had collected smaller volume of blood as per the prevalent practices at that time. Similarly, different culture media were employed in different studies which may have influenced the results as the potential for bacterial growth may differ between the culture media. None of the studies identified by this review used the commercial media that are now commonly used for culture . Also, as these studies were conducted in different time periods (1955–2001), in different geographical regions, in different endemicity locations, and in different clinical and laboratory settings; inter study variation may have influenced the results. Forth, three papers did not separate reports for S. Typhi and S. Paratyphi A cases, which led to the exclusion of those papers from the analysis. We cannot predict if the inclusion of such studies would have altered the results. Fifth, as both blood and bone marrow cultures are partially sensitive, the composite reference standard may have still missed some typhoid fever cases. In that situation, the proportion or the S. Typhi detected could be an overestimation. The true proportion of cases detected and true sensitivity of culture can only be measured against a perfect test that is fully sensitive and specific. Molecular diagnostic tests may hold the promise for future to provide better composite reference standard, if not perfect . Sixth, as bone marrow culture is an invasive procedure, there is an intrinsic procedural difficulty that may have discouraged the clinicians from performing the procedure; or may have restrained patients or guardians from giving procedural consent. This may have introduced differential sample collection bias between blood and bone marrow culture. We could not measure this bias in our study because information on failed procedures or people did not consent for procedure was not presented in the papers. Seventh, the blood and bone marrow cultures could only be performed in health facilities sufficiently equipped for the complexity of both procedures and with accessible skilled personnel. This means the selected papers could represent the proportion of cases detected and the sensitivity at higher quality health facilities which need not be the same as the health facilities where only blood culture is performed. Thus the true blood culture sensitivity or proportion of cases identified at a primary health facility or a remote health facility or a community based health facility where typhoid fever surveillance is conducted may differ. Eight, most studies had small sample size; largest being 103, resulted in wide confidence intervals. Studies with larger sample size in future will help in reducing the uncertainty. Finally, the literature search only included studies published in English which may have resulted in missing some papers published in other languages.
The proportion of S. Typhi detected by blood culture estimated from this systematic literature review provides a point estimate and range for correcting missed typhoid fever cases in disease burden studies based on a scientific rationale. The estimated blood culture sensitivity in this review is higher than the commonly reported value. We question the use of blood culture sensitivity as a proxy for the proportion of S. Typhi cases detected in disease burden measures as it underestimates the real problem. This information should be used by clinicians, epidemiologists and decision makers in making rational and logical decisions. Future research should focus on measuring proportion of typhoid fever cases detected by blood culture based on standardized concurrent blood and bone marrow culture or other reasonable reference standard in the field settings which will help in understanding the true proportion of S. Typhi cases identified by blood culture.
VM has conceptualized the study, advised on developing search terms, served as the third reviewer in conducting the search, guided the analysis, interpreted the results, rewrote and finalized the manuscript.ER has served as first reviewer in conducting the search, extracted data, conducted the analysis, worked on early draft manuscript. VVM has served as second reviewer in conducting the search, developed search terms, assisted in data extraction and analysis, reviewed and edited the manuscript. JYP estimated weighted mean proportion of positive blood culture using random effects model and drew forest plots. All authors read and approved the final manuscript.
This study was supported by the Bill and Melinda Gates Foundation through the funds made available for Vi-based Vaccines for Asia (VIVA) Initiative. The Korean and Swedish Governments provide core funding to International Vaccine Institute. We thank Dr. Jin Kyung Park for statistical support and Dr. Thomas Wierzba and Dr. Christopher Parry for reviewing the manuscript and providing valuable advice.
The authors declare that they have no competing interests.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- WHO. Background document: the diagnosis, treatment and prevention of typhoid fever. Geneva: World Health Organization; 2003.Google Scholar
- WHO. Typhoid vaccines: WHO position paper. Wkly Epidemiol Rec Health Sect Secr Leag Nations. 2008;83:49–59.Google Scholar
- Bhutta ZA. Current concepts in the diagnosis and treatment of typhoid fever. BMJ. 2006;333:78–82.View ArticlePubMedPubMed CentralGoogle Scholar
- Vaught JB. Blood collection, shipment, processing, and storage. Cancer Epidemiol Biomarkers Prev. 2006;15:1582–4.View ArticlePubMedGoogle Scholar
- Gillio-Meina C, Cepinskas G, Cecchini EL, Fraser DD. Translational research in pediatrics II: blood collection, processing, shipping, and storage. Pediatrics. 2013;131:754–66.View ArticlePubMedGoogle Scholar
- Srikantiah P, Girgis FY, Luby SP, Jennings G, Wasfy MO, Crump JA, Hoekstra RM, Anwer M, Mahoney FJ. Population-based surveillance of typhoid fever in Egypt. Am J Trop Med Hyg. 2006;74:114–9.PubMedGoogle Scholar
- Thriemer K, Ley BB, Ame SS, Deen JL, de Pak G, Chang NY, Hashim R, Schmied WH, Busch CJ-L, Nixon S, Morrissey A, Puri MK, Ochiai RL, Wierzba T, Clemens JD, Ali M, Jiddawi MS, von Seidlein L, Ali SM. Clinical and epidemiological features of typhoid fever in Pemba, Zanzibar: assessment of the performance of the WHO case definitions. PLoS ONE. 2012;7(12):518.View ArticleGoogle Scholar
- Modern epidemiology. 2nd ed. Philadelphia: Lippincott-Raven; 1998.Google Scholar
- Parry CM, Wijedoru L, Arjyal A, Baker S. The utility of diagnostic tests for enteric fever in endemic locations. Expert Rev Anti Infect Ther. 2011;9:711–25.View ArticlePubMedGoogle Scholar
- Crump JA, Luby SP, Mintz ED. The global burden of typhoid fever. Bull World Health Organ. 2004;82:346–53.PubMedPubMed CentralGoogle Scholar
- Buckle GC, Walker CLF, Black RE. Typhoid fever and paratyphoid fever: systematic review to estimate global morbidity and mortality for 2010. J Glob Health. 2012;2:010401.View ArticlePubMedPubMed CentralGoogle Scholar
- Malla T, Malla KK, Thapalial A, Shaw C. Enteric fever: a retrospective 6-year analysis of 82 paediatric cases in a teaching hospital. Kathmandu Univ Med J KUMJ. 2007;5:181–7.PubMedGoogle Scholar
- Akoh JA. Relative sensitivity of blood and bone marrow cultures in typhoid fever. Trop Doct. 1991;21:174–6.PubMedGoogle Scholar
- Gasem MH, Dolmans WM, Isbandrio BB, Wahyono H, Keuter M, Djokomoeljanto R. Culture of Salmonella typhi and Salmonella paratyphi from blood and bone marrow in suspected typhoid fever. Trop Geogr Med. 1995;47:164–7.PubMedGoogle Scholar
- Vallenas C, Hernandez H, Kay B, Black R, Gotuzzo E. Efficacy of bone marrow, blood, stool and duodenal contents cultures for bacteriologic confirmation of typhoid fever in children. Pediatr Infect Dis. 1985;4:496–8.View ArticlePubMedGoogle Scholar
- Gehlbach SH. Interpreting the medical literature5th. New York: McGraw-Hill, Medical Pub. Division; 2006.Google Scholar
- Hawass NE. Comparing the sensitivities and specificities of two diagnostic procedures performed on the same group of patients. Br J Radiol. 1997;70:360–6.View ArticlePubMedGoogle Scholar
- Hirsowitz L, Cassel R. Sternal marrow cultures in typhoid fever. Br Med J. 1951;1:862.View ArticlePubMedPubMed CentralGoogle Scholar
- Farooqui BJ, Khurshid M, Ashfaq MK, Khan MA. Comparative yield of Salmonella typhi from blood and bone marrow cultures in patients with fever of unknown origin. J Clin Pathol. 1991;44:258–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Gasem MH, Smits HL, Goris MGA, Dolmans WMV. Evaluation of a simple and rapid dipstick assay for the diagnosis of typhoid fever in Indonesia. J Med Microbiol. 2002;51:173–7.View ArticlePubMedGoogle Scholar
- Chaicumpa W, Ruangkunaporn Y, Burr D, Chongsa-Nguan M, Echeverria P. Diagnosis of typhoid fever by detection of Salmonella typhi antigen in urine. J Clin Microbiol. 1992;30:2513–5.PubMedPubMed CentralGoogle Scholar
- Wain J, Diep TS, Bay PV, Walsh AL, Vinh H, Duong NM, Ho VA, Hien TT, Farrar J, White NJ, Parry CM, Day NP. Specimens and culture media for the laboratory diagnosis of typhoid fever. J Infect Dev Ctries. 2008;2:469–74.PubMedGoogle Scholar
- Gilman RH, Terminel M, Levine MM, Hernandez-Mendoza P, Hornick RB. Relative efficacy of blood, urine, rectal swab, bone-marrow, and rose-spot cultures for recovery of Salmonella typhi in typhoid fever. Lancet. 1975;1:1211–3.View ArticlePubMedGoogle Scholar
- Guerra-Caceres JG, Gotuzzo-Herencia E, Crosby-Dagnino E, Miro-Quesada M, Carrillo-Parodi C. Diagnostic value of bone marrow culture in typhoid fever. Trans R Soc Trop Med Hyg. 1979;73:680–3.View ArticlePubMedGoogle Scholar
- Mogasale V, Maskery B, Ochiai RL, Lee JS, Mogasale VV, Ramani E, Kim YE, Park JK, Wierzba TF. Burden of typhoid fever in low-income and middle-income countries: a systematic, literature-based update with risk-factor adjustment. Lancet Glob Health. 2014;2:e570–80.View ArticlePubMedGoogle Scholar
- Wain J, Bay PVB, Vinh H, Duong NM, Diep TS, Walsh AL, Parry CM, Hasserjian RP, Ho VA, Hien TT, Farrar J, White NJ, Day NPJ. Quantitation of bacteria in bone marrow from patients with typhoid fever: relationship between counts and clinical features. J Clin Microbiol. 2001;39:1571–6.View ArticlePubMedPubMed CentralGoogle Scholar
- Reller ME, Zaidi AKM, Sultana S, Azeem S, Hanif B, Qureshi S, Hasan R, Bhutta Z, Akhter R, Goldmann DA. Controlled evaluation of bactec peds plus/F and bactec lytic/10 anaerobic/F media for isolation of Salmonella enterica Serovars Typhi and Paratyphi A from blood. J Clin Microbiol. 2009;47:245–6.View ArticlePubMedPubMed CentralGoogle Scholar