Typhoid fever in children aged less than 5 years
Anju Sinha, Sunil Sazawal, Ramesh Kumar, Seema Sood, Vankadara P Reddaiah, Bir Singh, Malla Rao,
Abdolla Naficy, John D Clemens, Maharaj K Bhan
Summary
Background Calculation of the incidence of typhoid fever
during preschool years is important to define the optimum
age of immunisation and the choice of vaccines for public-
health programmes in developing countries. Hospital-based
studies have suggested that children younger than 5 years
do not need vaccination against typhoid fever, but this view
needs to be re-examined in community-based longitudinal
studies. We undertook a prospective follow-up study of
residents of a low-income urban area of Delhi, India, with
active surveillance for case detection.
Methods A baseline census was undertaken in 1995.
Between Nov 1, 1995, and Oct 31, 1996, we visited 8172
residents of 1820 households in Kalkaji, Delhi, twice weekly
to detect febrile cases. Blood samples were obtained from
febrile patients, and those who tested positive for
Salmonella typhi were treated with ciprofloxacin.
Findings 63 culture-positive typhoid fever cases were
detected. Of these, 28 (44%) were in children aged under 5
years. The incidence rate of typhoid per 1000 person-years
was 27·3 at age under 5 years, 11·7 at 5–19 years, and 1·1
between 19 and 40 years. The difference in the incidence of
typhoid fever between those under 5 years and those aged
5–19 years (15·6 per 1000 person-years [95% CI
4·7–26·5]), and those aged 19–40 years (26·2 [16·0–36·3])
was significant (p<0·001 for both). The difference between
the incidence of typhoid at 5–19 years and the incidence at
19–40 years was also significant (10·6 [6·3–14·8],
p<0·001). Morbidity in those under 5 and in older people
was similar in terms of duration of fever, signs and
symptoms, and need for hospital admission.
Interpretation Our findings challenge the common view that
typhoid fever is a disorder of school-age children and of
adults. Typhoid is a common and significant cause of
morbidity between 1 and 5 years of age. The optimum age of
typhoid immunisation and the choice of vaccines needs to
be reassessed.
Lancet 1999; 354: 734–37
See Commentary page 698
Indian Council for Medical Research, Advanced Centre for
Diarrhoeal Disease Research, Division of Paediatric
Gastroenterology (A Sinha MD, S Sazawal MD, R Kumar MD,
Prof M K Bhan MD ), Department of Microbiology (S Sood MD ), and
Centre for Community Medicine (V P Reddaiah MD , B Singh MD), All
India Institute of Medical Sciences, New Delhi, India; Department
of International Health, School of Hygiene and Public Health, Johns
Hopkins University, Baltimore, MD (S Sazawal); and Department of
Health and Human Services, National Institute of Child Health and
Human Development, Bethesda, MD, USA (M Rao MPH,
A Naficy MPH, J D Clemens MD)
Correspondence to: Prof M K Bhan, Department of Pediatrics,
All India Institute of Medical Sciences, New Delhi 110-029, India
734
Introduction
Typhoid fever is a waterborne and foodborne disorder.
Unlike most other gastrointestinal infections, which
predominantly affect children aged 6 months to 3 years,
the incidence of typhoid fever peaks between 5 and 12
years.1 According to hospital-based data and limited
information from field studies that used passive
surveillance for case detection, typhoid fever is infrequent
under 3 years of age.2–5 This finding is repeated in
standard medical texts.6 A consequence of this
epidemiological profile is that antityphoid vaccines
currently available have been assessed only in children of
school age and in older people, and the optimum age for
immunisation as part of public-health programmes
remains unclear.
Several factors may explain the low rates of detection of
typhoid fever in preschool years. Typhoid fever may be
milder or atypical in presentation at this age,3,7,8 and as
such, it may be under-reported when case detection is
passive, as was the case in most previous studies.
Subclinical or atypical presentation may be the result of
an underdeveloped reticuloendothelial system, the nidus
for multiplication of Salmonella typhi. Difficulties in
collecting the recommended 5 mL blood from preschool
children for conventional blood culture methods used in
endemic settings may also lead to underdiagnosis.9,10
For these reasons, it is important to re-examine the
incidence of typhoid fever, particularly in infants and
young children, by means of active surveillance11–13 and by
use of isolation methods that are highly sensitive even
when the amount of blood available for testing is small.
These data would help determine a suitable age for
immunisation in public-health programmes, and would
enable assessment of the effectiveness of antityphoid
vaccines currently available.14
We therefore undertook a prospective surveillance
study of a community-based cohort aged up to 40 years.
Blood culture for every eligible febrile episode used a
medium that gives high sensitivity when only a small
amount of blood is available for tests, as is common with
preschool children.
Methods
Study population
We studied residents of low socioeconomic status in an urban
area of Kalkaji, New Delhi, India, between Nov 1, 1995, and
Oct 31, 1996. The local climate has three distinctive seasons—a
hot dry summer (April–June), a wet and humid monsoon
(July–October), and a cool dry winter (November–March).
Residents of Kalkaji are migrants from neighbouring states.
Families live in jhuggies—clay structures with one or two rooms.
66% of water comes from hand pumps, and 34% from a piped
supply. The drainage system for sewage and wastewater is
inadequate. Typhoid immunisation is not a part of routine
health-care. We have undertaken continuous study of this
population for several years, which has enabled us to establish a
detailed demographic sampling frame.
In 1995, a computerised census of the population of Kalkaji
showed a total of 19 585 residents living in 4361 dwellings. All
THE LANCET • Vol 354 • August 28, 1999
Age at follow-up Total follow-up Culture-confirmed Typhoid incidence∗ vials, respectively (Becton Dickson, MD, USA). The latter
(years) (years) cases (n) (95% CI) medium contains antibiotic absorbing resins that enhance
Under 5 1027 28 27·3 (17·2 to 37·4)
isolation of bacteria if the patient has taken antibiotics before
0–1 166 0 ·· sampling. Specimens were transported to the laboratory
>1–2 202 5 24·8 (3·1 to 46·5) immediately and processed (Bactec NR 730, Becton Dickson).
>2–3 213 11 51·6 (21·1 to 82·2) The vials were probed for growth value twice on the first 2 days,
>3–4 225 5 22·2 (2·7 to 41·7) once on day 3 and day 4, and once on day 10. Positive vials were
>4–5 221 7 31·7 (8·2 to 55·2)
Over 5–19 2743 32 11·7 (7·9 to 15·7)
subcultured on blood, MacConkey, and chocolate agar plates.
肁5–12 1579 22 13·9 (8·1 to 19·8) Smears were gram-stained, and if gram-negative bacilli were
>12–19 1164 10 8·6 (3·3 to 13·9) found, further processing was done by the Sceptor MIC/ID
Over 19–40 2684 3 1·1 (⫺0·1 to 2·4) panel (Becton Dickson) for enteric bacilli. The panel was
Total 6454 63 9·8 (7·4 to 12·2) inoculated with the Sceptor inoculator (Becton Dickson) and
*Incidence per 1000 person-years. biochemical reactions, and minimum inhibitory concentration
Table 1: Age-specific incidence of culture-confirmed typhoid endpoints were then recorded by use of the template. The results
detected by active surveillance over a 1-year period in urban of biochemical and tests of minimum inhibitory concentration
Dehli were processed by a computer that used Sceptor software
(version 3.0) to identify the most likely organism present and to
residents were assigned a unique identification number, and this
calculate confidence values. Confirmation of Salmonella spp was
gave us the sample population for the study. To ensure uniform
done by agglutination with factor O, H, and Vi antiserum
distribution of randomly selected residents throughout the
(Murex Biotech, Dartford, UK). Culture-positive typhoid was
population and to prevent biases due to case clusters, the
defined as a febrile episode that yielded S typhi in blood culture.
population was divided into clusters of 70 households with a
mean of 4·49 people per household, and 26 such clusters were
randomly selected by computer for active surveillance. Statistical analysis
Ethical approval was obtained from the Institutional Review We estimated a sample size of 8000 people for active surveillance
Boards of the All India Institute of Medical Sciences, New which would give 7000 person-years of follow-up on the
Delhi, India, and the National Institute of Child Health and assumption of 12% loss to follow-up. By enumeration of all
Development, Bethesda, MD, USA—the two participating possible outcomes from our binomial distribution, we estimated
institutions. Informed consent was obtained from the study that the follow-up of 7000 person-years would give an 85%
participants before enrolment. probability that the lower limit of the 95% CI for the incidence
of typhoid fever would be at least six cases per 1000 person-
years, assuming an incidence of ten cases per 1000 person-years.
Fieldwork
We estimated the incidence of typhoid for those people under
Between Nov 1, 1995, and Oct 31, 1996, residents in active surveillance by dividing the number of episodes by the
households selected for active surveillance were visited at home actual number of years of follow up, which was calculated to
twice a week by trained field assistants. During these visits, the allow for people to move to higher age categories. Incidence was
continuing residence status of each study participant was calculated per 1000 person-years. Age at typhoid onset was used
verified, and newcomers to the households were enrolled if they to categorise cases in age groups. We estimated 95% CIs for
proposed to stay for the duration of the study. Each resident was incidence rates and differences across age categories using
questioned about any fever. Duration of fever in children was conventional methods.15 We compared proportions using ␹2, ␹2
determined by parental assessment. People reported to be ill but for trend, and Fisher’s exact tests.
not available at the visit were revisited in the evening, as were
those households that were found locked. If residents could not
be contacted, we recorded them as “not available” until the next Results
scheduled visit. The total population number studied was 8172, of whom
When fever was reported, core body temperature was 7159 were less than 40 years old. We visited these 7159
measured by tympanic thermometer (Thermoscan Pro-LT, San initially enrolled residents at their households twice
Diego, CA, USA). Blood samples were obtained if body weekly for 12 months: 1126 were under 5 years old; 2908
temperature was 38°C or more for those 5 years or younger, aged between 5 and 19 years, and 3125 were over 19
irrespective of the duration of fever. People older than 5 years years of age. Only 11 880 (6·8%) of 174 720 visits failed
had to have had continuous fever for the past 3 days. Patients to yield morbidity information. During the year there
eligible for blood sampling were referred to the study clinic for
were 171 births and 47 deaths, and 383 people
examination by a physician and for blood sampling after
informed consent was obtained. Patients gave only one sample
outmigrated. The new births and 1798 immigrants were
during a febrile episode. included in the surveillance.
All febrile cases were followed up by a physician twice during Overall, there were 1454 febrile episodes eligible for
the first week of illness, and weekly thereafter until recovery. On blood culture. Of these, cultures were actually obtained
each follow-up visit, patients were assessed for vital signs, in 1217 cases. S typhi were isolated in 63 cases,
duration and character of fever, toxic effects, presence of salmonella other than S typhi in 24 cases, and other
palpable spleen, and other relevant physical signs. Febrile cases bacteria were isolated in 18 cases. Of the total culture-
were initially given an antipyretic and antimalarial drugs. positive S typhi cases in the cohort, 16 (25%) occurred in
Patients whose blood culture was positive for S typhi, or those children under 3 years of age and 28 (44%) in those
whose clinical profile was consistent with typhoid fever, were under 5 years. During 1 year of active surveillance, 63
treated with oral ciprofloxacin 15 mg per kg bodyweight for at
culture-positive cases were observed. The overall
least 7 days, and for 4 days after the cessation of fever.
Ciprofloxacin is currently used for primary treatment of typhoid incidence of typhoid fever in patients up to 40 years of
fever in India because multidrug resistant strains are common. age was 9·8 per 1000 person-years.
Analysis of age-specific incidence showed some
Laboratory methods unexpected results (table 1). The difference between the
Blood sampling used aseptic precautions. 3–5 mL blood was incidence of typhoid fever at age under 5 years and the
taken from patients older than 5 years, at least 2 mL blood was incidence at 5–19 years (15·6 cases per 1000 person years
taken from children under 5 years, and samples were inoculated [95% CI 4·7–26·5]) and at 19–40 years (26·2
into Bactec 26 A plus (aerobic) and Bactec red plus (aerobic) [16·0–36·3]) was significant (p<0·001). The difference

THE LANCET • Vol 354 • August 28, 1999 735

Indicator of Age Difference in p from patients under 5 years and in 47% of isolates from
severity
<5 years 肁5 years
proportions (95% CI) those older than 5 years. Sensitivity to ciprofloxacin
(n=28) (n=35) occurred in 96% of isolates from those aged under 5 years
Toxicity 18 (64%) 19 (54%) 10 (⫺34·2 to 14·2) 0·59 and 97% of isolates from those aged over 5 years. The
Splenomegaly 4 (14%) 9 (26%) ⫺12 (⫺30·9 to 8·0) 0·42 phage types of all the S typhi isolates were determined by
Diarrhoea 4 (14%) ·· ·· ··
Vomiting 1 (4%) 1 (3%) 1 (⫺9·5 to 8·1) 1·0
standard procedures at the National Reference Centre for
Hospital admission 5 (18%) 1 (3%) 15 (⫺0·2 to 30·2) 0·08 salmonella phage typing. The strains were of several
Fever duration (days) different phage types, predominantly E1 (54·7%), A
聿3 1 (4%) 1 (3%) 1 (⫺9·5 to 8·1) 0·7† (15%), and 35 (15%). All phage types were distributed
>3–7 ·· 3 (9%) ·· ··
>7–10 9 (32%) 4 (11%) 21 (⫺0·4 to 40·9) ·· evenly throughout the study area.
>10 18 (65%) 27 (77%) ⫺12 (⫺35·4 to 9·7) ·· Nine patients did not recover after a 10-day course of
Loss of 1 or more full 22 (78%) 33 (94%) ⫺16 (⫺32·7 to 1·3) 0·12 ciprofloxacin despite sensitivity to the drug in vitro.
days normal activity*
These patients were admitted to hospital and treated with
*For workers=loss of full workday, for children=loss of full day at school, for preschool
children=loss of full day of normal activity needing special attention of mother. † ␹2 for
ceftriaxone. Seven of the nine patients were under 5 years
trend. old.
Table 2: Severity of culture-confirmed typhoid cases detected
in active surveillance during a 1-year period in urban Delhi Discussion
between typhoid incidence at 6–19 years and later ages Our findings challenge the common view of typhoid fever
was also significant (10·6 [6·3–14·8], p<0·001). The as a disorder that affects mainly children of school age
incidence of culture-positive typhoid peaked at age and adults.16 These findings also contradict the current
3 years at 51·6 cases per 1000 person-years (95% CI view that typhoid in children under 5 years is mild and
subclinical. Typhoid infection can be a significant cause
21·1–82·2). No case was confirmed by culture in the first
of morbidity between 1 and 5 years of age. In Kalkaji,
year of life.
44% of all S typhi infections occurred in children aged
Incidence rates of typhoid varied seasonally. The
between 1 and 5 years. We suggest that current strategies
maximum incidence (18·8 cases per 1000 person-years)
for vaccination against typhoid fever need urgent review.
occurred during the monsoon (July–October), and lower
The relatively mild and atypical nature of illness related
rates of 5·4 and 4·7 per 1000 person-years occurred
to S typhi in preschool children may explain the low
during summer and winter seasons, respectively.
number of hospital admissions in this age group.3,7,8,17
We studied a dynamic population cohort, in which
However, the data on severity in our study should be
estimates of disease incidence may be biased by different
judged in the context of the early and aggressive
incidence rates among immigrants. To assess whether
treatment provided. Although we found no cases of
this bias affected our results we also calculated the
perforation and haemorrhage, our findings show that
incidence of typhoid in the fixed cohort of 6849 people
typhoid infection in young children does not cause a
for whom data were available at the beginning and the merely transient, benign bacteraemia.
end of 1 year’s surveillance. The age-specific incidence Two moderately effective and well tolerated vaccines
rates under 5 years of age in this fixed cohort (24·1 per against typhoid fever are currently available, but neither is
1000 person-years in the under 5 years, 4·5 in those older being used in public-health programmes in developing
than 5 years) were similar to the rates in the dynamic countries. Our data and experience of feasibility suggest
cohort. that the optimum age for initial immunisation against
We took blood samples from children under 5 years old typhoid in a setting such as Kalkaji is the same age as
from the first day of fever, but in older patients the measles immunisation.18 Ty21a typhoid vaccine, which
sample was obtained only after at least 3 days of fever. To has been tested for efficacy only in school children and
assess whether this method accounted for differential adults,19,20 is not suitable for large-scale use in developing
incidence rates in those under 5 and those older we countries for logistical and cost reasons. This vaccine
analysed rates of culture positively by day of fever at the requires at least three oral doses and cannot, therefore, be
time of blood collection. At 1–2 days, 3–7 days, and more given at one visit alone at 9 months of age. The Vi
than 7 days, the positivity rates in children up to 5 years polysaccharide vaccine could be given, but it is not
of age were 2·0%, 4·3%, and 30·4% respectively. In older sufficiently and durably immunogenic at this age since it
patients, blood samples were not obtained on fever days is a T-independent antigen,21 although data on this issue
1–2, and the positivity rates at 3–7 days and more than are sparse. Our findings support the need for
7 days from blood collection were 6·5% and 15·2%. In development of new vaccines such as conjugates of the Vi
both age groups, maximum culture positivity rates polysaccharide and others that may be effective when
occurred after 7 days of fever. Therefore, early blood given in late infancy.
sampling did not affect the positivity rate. The incidence of typhoid fever and the age distribution
The clinical profiles of typhoid fever in children under of cases varies between developing countries.22 Therefore,
5 years and in older patients are compared in table 2. the age patterns of typhoid fever observed in our urban
There was no significant difference in any indicators of study area may differ from those in rural areas within
typhoid severity between these two age categories. India or in other developing countries. Ferreccio and
However, typhoid fever was associated with significant colleagues3 showed a low incidence of mild typhoid in
morbidity in children under 5 years of age. 18 of 28 infants and young children in an endemic area in
patients under 5 years old had signs of toxicity, all but Santiago. Similar epidemiological data on typhoid fever
one had fever for more than 7 days despite prompt are required in different regions of the world and other
treatment, and five needed admission to hospital. developing countries to allow estimates of cost
The minimum inhibitory concentration technique effectiveness and formulation of rational public-health
showed sensitivity to chloramphenicol in 64% of isolates policy for typhoid immunisation.

736 THE LANCET • Vol 354 • August 28, 1999

Contributors
Maharaj K Bhan was the principal investigator in India and was involved
in protocol design, study conduct, and data interpretation. Sunil Sazawal
was involved in study conception and design, obtained all baseline data
used in defining sampling strategy, and contributed to statistical analysis
and preparation of the paper. Anju Sinha supervised the fieldwork, and
assisted in data analysis and preparation of the paper. Ramesh Kumar and
Seema Sood supervised the laboratory work. Vankadara P Reddaiah and
Bir Singh provided statistical assistance. John D Clemens, Malla Rao, and
Abdolla Naficy, our US collaborators, helped at all stages of the study,
particularly during data analysis.
Acknowledgments
The study was supported by the Indo-US vaccine action programme.
We thank Charles Lowe of NIH for help with design; Dharmendra
Kashyap for computer software development and data management;
Geeta Mehta of the National Reference Centre for salmonella phage
typing at the Lady Hardinge Medical College, New Delhi, India; the
Department of Biotechnology, India and National Institutes of Health,
Bethesda, MD, USA, for financial support; and the Norwegian University
Committee for Development Research and Education, for core support.
References
1 Patnaik KC, Kapoor PN. A note on incidence of typhoid in Delhi.
Indian J Med Res 1967; 55: 228–39.
2 Sen SK, Mahakur AC. Enteric fever—a comparative study of adult
and paediatric cases. Indian J Pediatr 1972; 39: 354–60.
3 Ferreccio C, Manterola A, Prenzel I, et al. Benign bacteremia caused
by Salmonella typhi and parathypi in children younger than 2 years.
J Pediatr 1984; 104: 899–901.
4 Kapoor JP, Mohan M, Talwar V, et al. Typhoid fever in young
children. Indian Pediatr 1985; 22: 811–13.
5 Mahle WT, Levine MM. Salmonella typhi infection in children
younger than five years of age. Pediatr Infect Dis J 1993; 12: 627–31.
6 Levine MM. Typhoid fever vaccines. In: Plotkin S, Mortimer E.
Vaccines (2nd ed). Philadelphia: WB Saunders, 1994: 601.
7 Mulligan TO. Typhoid fever in young children. BMJ 1971; 4:
665–67.
THE LANCET • Vol 354 • August 28, 1999
8 Arora RK, Gupta A, Joshi NM, et al Multidrug resistant typhoid fever:
study of an outbreak in Calcutta. Indian Pediatr 1992; 29: 61–66.
9 Edelman R, Levine MM. Summary of an international workshop on
typhoid fever. Rev Infect Dis 1986; 8: 329–49.
10 Mishra S, Patwari AK, Anand VK, et al. A clinical profile of multidrug
resistant typhoid fever. Indian Pediatr 1991; 28: 1171–74.
11 Acharya IL, Lowe CU, Thapa R, et al. Prevention of typhoid fever in
Nepal with the Vi capsular polysaccharide of salmonella typhi. N Engl
J Med 1987; 317: 1101–04.
12 Klugman KP, Koornhof HJ, Schneerson R, et al. Protective activity of
Vi capsular polysaccharide vaccine against typhoid fever. Lancet 1987;
ii: 1165–69.
13 Chuttani CS, Prakash K, Vergese A, et al. Controlled field trials of
oral killed typhoid vaccines in India. Int J Epidemiol 1972; 1: 39–43.
14 Gupta A, Jalla S, Sazawal S, et al. Advances in vaccines for typhoid
fever. Indian J Pediatr 1994; 61: 321–39.
15 Smith PG, Morrow RH. Methods for field trials of interventions
against tropical diseases: a ‘toolbox’. Oxford: Oxford University Press,
1991: 292–93.
16 Levine MM, Tacket CO, Galen JE, et al. Progress in development of
new attenuated strains of Salmonella typhi as live oral vaccines against
typhoid fever. In: Levine MM, Woodrow GC, eds. New generation
vaccines. New York: Marcel Dekker, 1997: 437–46.
17 Middlekamp JN. Typhoid fever in infants and children. Med Times
1965; 93: 956–62.
18 Ivanoff B, Levine MM, Lambert PH. Vaccination against typhoid
fever: present status. Bull World Health Organ 1994; 72: 957–71.
19 Ferreccio C, Levine MM, Rodriguez H, et al.Comparative efficacy of
two, three, or four doses of Ty21a live oral vaccine in enteric coated
capsules: a field trial in an endemic area. J Infect Dis 1989; 159:
766–69.
20 Hackett J. Salmonella-based vaccines. Vaccine 1990; 8: 5–11.
21 Plotkin SA, Cam NB. A new typhoid vaccine composed of the Vi
capsular polysaccharide. Arch Intern Med 1995; 155: 2293–99.
22 Simanjuntak CH, Paleologo FP, Punjabi NH, et al. Oral
immunisation against typhoid fever in Indonesia with Ty21a vaccine.
Lancet 1991; 338: 1055–59.
737