Clinical Infectious Diseases

MAJOR ARTICLE

Prevention of Typhoid by Vi Conjugate Vaccine and

Achievable Improvements in Household Water, Sanitation,
and Hygiene: Evidence From a Cluster-Randomized Trial
in Dhaka, Bangladesh
Birkneh Tilahun Tadesse,1,a, Farhana Khanam,2,a Faisal Ahmmed,2 Justin Im,1 Md. Taufiqul Islam,2 Deok Ryun Kim,1 Sophie S. Y. Kang,1 Xinxue Liu,3
Fahima Chowdhury,2 Tasnuva Ahmed,2 Asma Binte Aziz,1 Masuma Hoque,2 Juyeon Park,1,4 Gideok Pak,1 Khalequ Zaman,2 Ashraful Islam Khan,2
Andrew J. Pollard,3 Jerome H. Kim,1 Florian Marks,1,4,5,6 Firdausi Qadri,2 and John D. Clemens1,2,7
1
International Vaccine Institute, Seoul, Republic of Korea; 2International Centre for Diarrhoeal Disease Research, Bangladesh, Dhaka, Bangladesh; 3Oxford Vaccine Group, Department of Pediatrics,
University of Oxford, and the National Institute for Health Research Oxford Biomedical Research Centre, Oxford, United Kingdom; 4Cambridge Institute of Therapeutic Immunology and Infectious
Disease, University of Cambridge School of Clinical Medicine, Cambridge Biomedical Campus, Cambridge, United Kingdom; 5Department of Microbiology and Parasitology, University of
Antananarivo, Antananarivo, Madagascar; 6Heidelberg Institute of Global Health, University of Heidelberg, Heidelberg, Germany; and 7Fielding School of Public Health, University of California,
Los Angeles, Los Angeles, California, USA

Background. Typhoid fever contributes to approximately 135 000 deaths annually. Achievable improvements in household
water, sanitation, and hygiene (WASH) combined with vaccination using typhoid conjugate vaccines (TCVs) may be an
effective preventive strategy. However, little is known about how improved WASH and vaccination interact to lower the risk of
typhoid.
Methods. A total of 61 654 urban Bangladeshi children aged 9 months to ,16 years, residing in 150 clusters with a baseline
population of 205 760 residents, were randomized 1:1 by cluster to Vi-tetanus toxoid TCV or Japanese encephalitis (JE) vaccine.
Surveillance for blood culture–confirmed typhoid fever was conducted over 2 years. Existing household WASH status was assessed
at baseline as Better or Not Better using previously validated criteria. The reduction in typhoid risk among all residents associated
with living in TCV clusters, Better WASH households, or both was evaluated using mixed-effects Poisson regression models.
Results. The adjusted reduced risk of typhoid among all residents living in the clusters assigned to TCV was 55% (95% confidence
interval [CI], 43%–65%; P , .001), and that of living in Better WASH households, regardless of cluster, was 37% (95% CI, 24%–48%;
P , .001). The highest risk of typhoid was observed in persons living in households with Not Better WASH in the JE clusters. In
comparison with these persons, those living in households with Better WASH in the TCV clusters had an adjusted reduced risk of
71% (95% CI, 59%–80%; P , .001).
Conclusions. Implementation of TCV programs combined with achievable and culturally acceptable household WASH practices
were independently associated with a significant reduction in typhoid risk.
Clinical Trials Registration. ISRCTN11643110.
Keywords. typhoid fever; WASH; typhoid conjugate vaccine; vaccine effectiveness.

Typhoid fever is a systemic disease caused by Salmonella enterica and around 135 000 typhoid fever–related deaths occur annual-
serovar Typhi (S. Typhi) and is a major public health concern in ly, with the highest burden in children residing in urban slums
low- and middle-income countries where the disease is endemic. [1, 2]. Emergence of multidrug-resistant S. Typhi further chal-
Globally, an estimated 14 million persons develop typhoid fever lenges conventional treatment regimens, resulting in a higher
frequency of complications and adverse health outcomes [3, 4].
As a human-restricted pathogen, transmission of S. Typhi
Received 11 January 2022; editorial decision 31 March 2022; published online 12 April 2022
a mainly occurs through ingestion of food or water contaminated
B. T. T. and F. K. contributed equally to this work.
Correspondence: B. T. Tadesse, International Vaccine Institute, SNU Research Park, 1 with feces of infected persons. The risk of transmission of
Gwanak-ro, Gwanak-gu, Seoul 08826, South Korea (birkneh.tadesse@ivi.int). S. Typhi in people who have no access to safe drinking water
Clinical Infectious Diseases® 2022;75(10):1681–7 is more than double that of those who have access to safe water
© The Author(s) 2022. Published by Oxford University Press on behalf of the Infectious Diseases
Society of America. [5]. Climate change and growth of economic opportunities in
This is an Open Access article distributed under the terms of the Creative Commons Attribution large cities has led to rapid urbanization of low- and
License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, dis-
tribution, and reproduction in any medium, provided the original work is properly cited.
middle-income countries; however, infrastructural develop-
https://doi.org/10.1093/cid/ciac289 ment, such as waste management systems and provision of

Typhoid Prevention by Vaccine and WASH • CID 2022:75 (15 November) • 1681
clean drinking water, has lagged behind. As a result, open sew- to a single dose of either the intervention (Vi-TT, Typbar-TCV,
age in urban slums often contaminates the domestic water Bharat Biotech International Ltd, Hyderabad, India) or control
source in urban dwellings, especially during the rainy season (SA 14-14-2 Japanese encephalitis [JE], Chengdu Institute of
[6]. Furthermore, unhygienic practices during food handling Biological Products, Chengdu, China) vaccine given to persons
exacerbate transmission within high-risk areas [7]. aged 9 months to ,16 years. Neither the JE nor Vi-TT vaccines
The combination of improvements in water, sanitation, and were available in the study area outside of the study setting.
hygiene (WASH) and vaccination using safe and efficacious vac- After the baseline immunization campaign, 3 catch-up vaccina-
cines will be pivotal for typhoid fever prevention and potential tions to the same age group were conducted at roughly
elimination in endemic countries. However, establishment of 6-month intervals to maintain stable vaccine coverage consid-
municipal infrastructure for waste and sewage disposal and en- ering the high migration rate within the study population.
suring a sustainable uncontaminated water supply are con- The baseline vaccination was held from 15 April to 15 May
strained by the lack of financial capital and political will. As 2018 followed by catch-up vaccination campaigns from
such, immunization with typhoid conjugate vaccines (TCVs), September to December 2018, April to May 2019, and
combined with building upon achievable and already existing im- October to November 2019.
provements in household WASH, may be an effective strategy for
typhoid prevention and potential elimination in hyperendemic
Typhoid Fever Surveillance
urban slums [8–11]. However, little is known about how already
Eight health facilities located in and around the study area par-
existing improvements in household WASH and typhoid vacci-
ticipated in passive surveillance serving the entire study popu-
nation interact to lower the risk of typhoid. In a recent analysis,
lation. These were all the public sector treatment centers for the
we demonstrated that salutary, already existing WASH practices
catchment population, dispersed throughout the study area. All
and behaviors that are therefore feasible and affordable predicted
microbiology tests were done at a single reference laboratory at
a lower occurrence of typhoid fever in urban slums of Dhaka,
the International Centre for Diarrhoeal Disease Research,
Bangladesh (Tadesse BT, Khanam F, unpublished data).
Bangladesh (icddr,b). Subjects were identified either by identi-
Earlier, we reported results from a cluster-randomized trial
fication cards distributed during the census or, lacking such
in urban Bangladesh that demonstrated that administration
cards, with use of the computerized census using tablets at
of a single dose of Vi polysaccharide conjugated to tetanus tox-
each surveillance site. Passive surveillance for typhoid fever
oid (Vi-TT) TCV to children 9 months to ,16 years of age con-
was initiated on 26 February 2018. Individuals with fever, de-
ferred 85% protection against typhoid in these children over an
fined as a history of fever for ≥2 days or axillary temperature
18-month period of surveillance [9]. Here, we extend these
of ≥38°C, were enrolled after giving informed consent. Blood
findings by addressing the public health question of what
samples were collected from study participants and assessed
were the combined effects of Vi-TT vaccination of children
by microbiological cultures for bacterial growth. Febrile visits
and the level of preexisting household WASH in protecting
were concatenated into febrile episodes when initial and subse-
the entire population of all ages against typhoid.
quent visits occurred within 14 days of one another. An episode
of typhoid fever was defined as a febrile episode in which
METHODS S. Typhi was isolated from blood culture and in which a
Study Population home visit confirmed that the person whose name was given
A cluster-randomized trial was conducted in a densely populat- at the treatment center had indeed sought care for treatment
ed urban area of Mirpur, Dhaka, Bangladesh, that has endemic of fever on the date of presentation.
typhoid fever [9, 12]. A baseline census and biannual updates
were conducted to enumerate the study population and to col- Definition of Household WASH
lect household-level demographic, socioeconomic, and WASH In a recent analysis (Tadesse BT, Khanam F, unpublished data),
information. A total of 205 760 individuals were enumerated in we developed a composite dichotomous variable for household
the baseline census, and 61 654 children aged 9 months to ,16 WASH (Better, Not Better) using individual WASH variables
years were eligible for the vaccination at baseline. Information collected during the baseline census for those present in the
on births, deaths, and migrations that occurred in or out of study area during the baseline census, and during the first cen-
the study area during the whole surveillance period were cap- sus update after the onset of their first participation for those
tured during biannual census updates. who moved into the clusters after the baseline census and for
births. A household with Better WASH had characteristics of
Vaccination owning a private toilet and the availability of a water filter in
After the baseline census, the whole study area was divided into the household during visit by study staff; or owning a private
150 geographical clusters, which were randomized at a 1:1 ratio toilet and access to safe drinking water, defined as having a

1682 • CID 2022:75 (15 November) • Tadesse et al

private tap, private well, bottled water, or water vendor We evaluated the impact of vaccination with Vi-TT by com-
(Supplementary Figure 1). paring the rate of blood culture–confirmed typhoid fever epi-
sodes in the overall population residing in the Vi-TT clusters
Statistical Analysis with those living in the JE clusters. Only first typhoid episodes
The analysis was conducted in a dynamic population that con- occurring 1 or more days after the start date of residence were
sidered births and in-migrations as well as catch-up vaccination included in the analysis. The association between living in a
after the baseline vaccination. Follow-up began at the “start Better WASH household and typhoid was assessed using the
date of residence.” For those present at baseline the start date same start date and follow-up strategy. We then evaluated the
of residence was the date of vaccination for vaccinees and the overall protection of all individuals living in Vi-TT vs JE clusters
midpoint of the baseline vaccination (30 April 2018) for non- according to whether they lived in Better vs Not Better WASH
vaccinees. For those who in-migrated or were newly born after households, and the protection of living in Better vs Not Better
baseline, the date of first in-migration and the date of birth were WASH households by their residence in Vi-TT vs JE clusters.
considered as the start date of residence in the study area. Each Finally, we evaluated the evidence for a trend of the incidence
individual was analyzed following the “first in, first out” princi- of typhoid considering 4 groups: persons living in Not Better
ple where individuals were included at the time of the start date WASH households in JE clusters (the reference group), persons
of residence in the study cluster and were censored at death, living in Better WASH households in JE clusters, persons living
end of surveillance (15 March 2021), or migration out of the in Not Better WASH households in Vi-TT clusters, and persons
cluster, whichever came first. living in Better WASH households in Vi-TT clusters.

Figure 1. Consolidated Standards of Reporting Trials (CONSORT) diagram showing population at first and last census and incident typhoid cases in Vi polysaccharide
conjugated to tetanus toxoid (Vi-TT) and Japanese encephalitis (JE) clusters. *Three study participants (1 from the JE and 2 from the Vi-TT clusters) died by 15 March 2020.

Typhoid Prevention by Vaccine and WASH • CID 2022:75 (15 November) • 1683
Table 1. Baseline Demographic Characteristics of Study Participants An interaction term for cluster assignment and household
Living in the Japanese Encephalitis and Vi–Tetanus Toxoid Clusters at WASH was fitted in the models to evaluate the possibility of
Baseline Census
synergistic protection against typhoid between living in
Parameter JE Group Vi-TT Group Vi-TT clusters and living in a household with Better WASH.
All residents 163 373 163 421
A 2-tailed P value , .05 was considered as the threshold for sig-
Mean age + SDa 25.4 + 17.3 25.3 + 17.2 nificance. All analyses were carried out using R version 4.1.0
Sex and the glmmTMB package was used to fit the mixed-effects
Male 81 134 (49.7) 81 097 (49.6) Poisson regression models [15, 16].
Female 82 239 (50.3) 82 324 (50.4)
Religion
Muslim 161 591 (98.9) 161 150 (98.6) RESULTS
Others 1782 (1.1) 2271 (1.4)
Trial Participants
Ward of residence
2 62 791 (38.4) 66 077 (40.4)
The disposition of trial participants using the Consolidated
3 38 551 (23.6) 44 168 (27.0) Standards of Reporting Trials (CONSORT) diagram is present-
5 62 031 (38.0) 53 176 (32.5) ed in Figure 1. In total, 205 760 individuals lived in the study
Resident’s household WASH status area during the baseline census; of these, 102 696 and 103 064
Better 56 969 (34.9) 54 281 (33.2)
resided in the Vi-TT and JE clusters, respectively. During 24
Not better 106 404 (65.1) 109 140 (66.8)
months of follow-up, we recorded 94 572 in-migrations and
Better WASH coverage, %, Mean + SD 37.6 + 16.5 35.7 + 18.3
6462 births and 85 772 out-migrations and 1529 deaths. The to-
Abbreviations: JE, Japanese encephalitis vaccine; SD, standard deviation; Vi-TT, Vi
polysaccharide conjugated to tetanus toxoid; WASH, water, sanitation, and hygiene. tal population residing in the study area was 239 493 at the last
a
Age in years measured at baseline census.
census and cumulatively 326 794 at any time during the entire
24 months of follow-up. A total of 33 727 and 33 315 individu-
als received a single dose of Vi-TT and JE vaccine, respectively.
In simple analyses, Kaplan-Meier survival curves were con- The mean age (+ standard deviation) of the residents at start
structed to evaluate protection against typhoid fever associated date of residence was 25.3 (+17.2) years in the Vi-TT clusters
with residence in a Vi-TT cluster or with residence in a house- and 25.4 (+17.3) years in the JE clusters. As shown in Table 1,
hold with Better WASH, statistically assessed using the log- the populations residing in the Vi-TT and JE clusters were also
rank test [13]. We further applied mixed-effects Poisson regres- comparable at the start date of residence for distribution by sex,
sion models to estimate the incidence rate ratios (IRRs) related religion, ward of residence, household WASH status, and prev-
to residence in a Vi-TT cluster or in a household with Better alence of Better WASH household status (35% in the Vi-TT
WASH while adjusting for the design effect of cluster- clusters and 33% in the JE clusters) (Table 1). The distribution
randomization including those used to define the strata for ran- of baseline characteristics by residence in Better and Not
domization (geographical ward, distance to study clinics), the WASH household is presented in Supplementary Table 1.
number of eligible children at baseline, and other baseline pre-
specified covariates during protocol development, including Protection Against Typhoid Associated With Living in Vi-TT Clusters and
age at baseline census and sex [14]. The IRRs and their P values Better WASH Households
and 95% confidence intervals (CIs) were estimated by exponen- In the cumulative population ever followed in the clusters, vaccine
tiating the coefficients for dichotomous variables for the coverage was 21% in the Vi-TT clusters and 20% in the JE clusters;
vaccine arm of residence and for residence in a Better vs in both Vi-TT and JE clusters, vaccine coverage of the age-eligible
Not Better WASH household, and protective effectiveness population was maintained at around 64%. A total of 633 con-
(PE) associated with residence in a Vi-TT cluster and in a firmed typhoid fever cases were diagnosed in the study population
Better WASH household was calculated as (1 – IRR) × 100%. during the approximately 24 months of follow-up. Of these,

Table 2. Protection Against Typhoid Associated With Residence in a Vi–Tetanus Toxoid Cluster and Residence in a Better Water, Sanitation, and Hygiene
Household

Vaccine/WASH No. of Individuals No. of Cases PY IRa Crude PE, % (95% CI) P Value Adjusted PEa, % (95% CI) P Value

JE 163 373 435 208 414 209 Ref … Ref …

Vi-TT 163 421 198 206 867 96 56 (43–66) ,.001 55 (43–65) ,.001
Not Better 215 544 474 261 425 181 Ref … Ref …
Better 111 250 159 153 856 103 42 (30–52) ,.001 37 (24–48) ,.001
Abbreviations: CI, confidence interval; IR, incidence rate; JE, Japanese encephalitis vaccine; PE, protective effectiveness; PY, person-years; Ref, reference group; Vi-TT, Vi polysaccharide
vaccine conjugated with tetanus toxoid; WASH, water, sanitation, and hygiene.
a
Model adjusted for cluster-randomization variables (ward [ward 2, ward 3, ward 5], number of eligible children at baseline, design effect, age at baseline census, and sex).

1684 • CID 2022:75 (15 November) • Tadesse et al

Table 3. Protection Against Typhoid Associated With Residence in a Vi-TT Cluster, Stratified by Household Water, Sanitation, and Hygiene (WASH), and
With Residence in a Household With Better WASH, Stratified by Vaccination Cluster

Crude PE, % Adjusted PEa, %

Strata Group No. of Individuals No. of Cases PY IR (95% CI) P Value (95% CI) P Value

Not Better WASH JE 106 404 325 129 088 252 Ref … Ref …
Vi-TT 109 140 149 132 337 113 58 (44–69) ,.001 58 (45–69) ,.001
Better WASH JE 56 969 110 79 326 139 Ref … Ref …
Vi-TT 54 281 49 74 530 66 52 (30–67) ,.001 48 (26–64) ,.001
JE Clusters Not Better WASH 106 404 325 129 088 252 Ref … Ref …
Better WASH 56 969 110 79 326 139 44 (30–55) ,.001 39 (24–51) ,.001
Vi-TT Clusters Not Better WASH 109 140 149 132 337 113 Ref … Ref …
Better WASH 54 281 49 74 530 66 40 (16–57) .003 34 (7–53) .016
Abbreviations: CI, confidence interval; IR, incidence rate; JE, Japanese encephalitis vaccine; PE, protective effectiveness; PY, person-years; Ref, reference group; Vi-TT, Vi polysaccharide
vaccine conjugated with tetanus toxoid; WASH, water, sanitation, and hygiene.
a
Model adjusted for cluster-randomization variables (ward [ward 2, ward 3, ward 5], number of eligible children at baseline, design effect, age at baseline census, and sex).

435 cases occurred in the JE clusters and 198 in the Vi-TT clusters groups defined by living in Vi-TT vs JE clusters, and by living
with respective typhoid fever incidence rates of 209 and 96 per 100 in households with Better vs Not Better WASH. We observed
000 person-years of observation (PYO). The adjusted overall PE the lowest typhoid incidence in persons who lived in Better
associated with living in the Vi-TT clusters was 55% (95% CI, WASH households located within the Vi-TT clusters, with an
43%–65%; P , .001). Similarly, 159 cases of typhoid fever were di- incidence rate of 66 per 100 000 PYO and an adjusted PE of
agnosed in persons living in Better WASH households compared 71% (95% CI, 59%–80%; P , .001) in comparison with the pop-
with 474 cases in those living in Not Better WASH households ulation living in Not Better WASH households in the JE
with respective incidence rates of 103 and 181 per 100 000 PYO. clusters (Table 4). This PE was higher than that observed
The PE associated with living in Better WASH households was for living in Vi-TT clusters but within Not Better WASH
37% (95% CI, 24%–48%; P , .001; Table 2). households, with an adjusted PE of 57% (95% CI, 44%–67%;
We next evaluated the overall PE associated with living in P , .001), and those living in Better WASH households but
Vi-TT clusters stratified by household WASH status, and within the JE clusters, with an adjusted PE of 39% (95% CI,
with living in Better WASH households stratified by vaccine 24%–51%, P , .001) (Table 4). The proportion of vaccinees
cluster arm assignment. For persons living in Not Better in these 4 categories was comparable (18%–21%). No evidence
WASH households, the adjusted overall PE associated with liv- of statistical interaction between household WASH status and
ing in Vi-TT clusters was 58% (95% CI, 45%–69%; P , .001) cluster assignment was observed (P = .740).
while for those living in Better WASH households, the adjusted
PE was 48% (95% CI, 26%–64%; P , .001). Similarly, living in
DISCUSSION
Better WASH households was associated with a PE of 39%
(95% CI, 24%–51%; P , .001) in persons living in the JE clus- Our findings show that already existing and therefore culturally
ters compared to 34% (95% CI, 7%–53%; P , .001) in those liv- acceptable and achievable improvements to household WASH
ing in the Vi-TT clusters (Table 3). practices and behaviors, combined with vaccination with an ef-
ficacious typhoid vaccine, were associated with a significant re-
Combined Impact of Living in Better WASH Households and Within Vi-TT duction in the burden of typhoid fever in an endemic urban
Clusters slum in Bangladesh. Living in a cluster receiving Vi-TT (55%
We further analyzed the combined impact of living in Better overall protection) or residing in a household with Better
WASH households within the Vi-TT clusters by examining 4 WASH (37% overall protection) were each independently

Table 4. Protection Against Typhoid by Vaccine Cluster of Residence and Water, Sanitation, and Hygiene in the Household

Vaccine + WASH No. of Individuals No. of Cases PY IR Crude PE, % (95% CI) P Value Adjusted PEa, % (95% CI) P Value

JE + Not Better WASH 106 404 325 129 088 252 Ref … Ref …
JE + Better WASH 56 969 110 79 326 139 44 (30–55) ,.001 39 (24–51) ,.001
Vi-TT + Not Better WASH 109 140 149 132 337 113 57 (43–67) ,.001 57 (44–67) ,.001
Vi-TT + Better WASH 54 281 49 74 530 66 74 (63–82) ,.001 71 (59–80) ,.001
Abbreviations: CI, confidence interval; IR, incidence rate; JE, Japanese encephalitis vaccine; PE, protective effectiveness; PY, person-years; Ref, reference group; Vi-TT, Vi polysaccharide
vaccine conjugated with tetanus toxoid; WASH, water, sanitation, and hygiene.
a
Model adjusted for cluster-randomization variables (ward [ward 2, ward 3, ward 5], number of eligible children at baseline, design effect, age at baseline census, and sex).

Typhoid Prevention by Vaccine and WASH • CID 2022:75 (15 November) • 1685
associated with a lower risk of typhoid. Moreover, examination relation to the absence of both, though, as noted earlier, our es-
of the combination of residence in a Better WASH household timates of the combined effect of vaccination and WASH were
and residence in a cluster receiving Vi-TT revealed a stepwise likely conservative for methodological reasons. We therefore
increase in protection, from living in a Better WASH household suggest that this reduction probably represents the minimum
in a JE cluster (39%), to living in a Vi-TT cluster but in a Not that can be achieved in realistic programs of vaccination and ef-
Better WASH household (57%), to living in a Better WASH fective and feasible WASH adaptations. Future well-designed
household within a Vi-TT cluster (71%). studies of WASH interventions and vaccination will be needed
Our study had several potential limitations. Households to design effective elimination strategies.
were not randomized to Better vs Not Better WASH. Taken together, our findings help elucidate the separate and
However, controlling for confounding variables did not affect combined impact of vaccination with TCV and already existing
our results. Furthermore, the composite household WASH improvements in WASH practices and behaviors in preventing
variable was built based on a simplified evaluation of typhoid in hyperendemic urban slums. These findings can be
household-level WASH features ascertained via a single ques- taken as a “proof of principle” to support the assertion that im-
tionnaire at baseline. This could have led to loss of some infor- plementation of achievable, high-impact WASH interventions
mation, which might have led to underestimation of the actual that leverage already existing salutary practices, in combination
impact of Better WASH household status. A more comprehen- with vaccination programs, may control typhoid fever in such
sive and specific definition of WASH behaviors and facilities re- settings.
sponsible for reducing typhoid transmission would have led to
higher estimates of protection. Additionally, the clusters were Supplementary Data
not true epidemiological units of person-to-person typhoid Supplementary materials are available at Clinical Infectious Diseases online.
transmission and likely experienced a great deal of transmis- Consisting of data provided by the authors to benefit the reader, the posted
sion originating from the outside, which would have attenuated materials are not copyedited and are the sole responsibility of the authors,
so questions or comments should be addressed to the corresponding
estimates of overall Vi-TT protection [9]. The relatively low author.
sensitivity of blood culture for the detection of S. Typhi could
also have led to underestimation of the absolute effect size of Notes
the interventions, though because differential sensitivity of Financial support. This work was supported by the Bill & Melinda Gates
blood cultures by vaccine or WASH status would seem unlike- Foundation (INV-025386). A. P. reports funding from the UK National
ly, the relative reductions of typhoid rates reported should not Institute for Health Research.
Potential conflicts of interest. A. P. reports grants or contracts from
have been affected. Finally, it is also important to stress that AstraZeneca for work on coronavirus disease 2019 vaccines (payment to
the WASH features studied in this analysis represent existing institution) and a leadership role as Chair of the UK Joint Committee on
WASH behaviors and practices of this population. Whether Vaccination and Immunisation and member of the World Health
Organization Strategic Advisory Group of Experts on Immunization until
WASH interventions delivered to at-risk populations can rep- 2022, both unpaid. F. M. reports participation on a data and safety moni-
licate or even improve upon these behaviors and practices in toring board for Ebola vaccine study by the London School of Hygiene
entire populations represents a significant challenge for the and Tropical Medicine. J. I. reports support for attending meetings and/or
travel by the Bill & Melinda Gates Foundation. All other authors report no
future.
potential conflicts of interest.
Despite these limitations, our analysis had several strengths. All authors have submitted the ICMJE Form for Disclosure of Potential
First, it was based on a rigorously planned and implemented Conflicts of Interest. Conflicts that the editors consider relevant to the con-
cluster-randomized trial in which vaccine assignment was un- tent of the manuscript have been disclosed.

ambiguous. In this trial, the diagnosis of typhoid was made

blinded to vaccine assignment. As well, persons conducting References
surveillance of typhoid had no foreknowledge of our hypothesis 1. Global Burden of Disease Collaborative Network. GBD 2019 cause and risk sum-
maries: typhoid fever—level 4 cause. 2020. Available at: http://www.
regarding the relationship between WASH and typhoid. healthdataorg/results/gbd_summaries/2019. Accessed 20 September 2021.
Second, we used a WASH composite variable that was previ- 2. Sinha A, Sazawal S, Kumar R, et al. Typhoid fever in children aged less than 5
years. Lancet 1999; 354:734–7.
ously developed in the nonintervention clusters in the same
3. Rasheed MK, Hasan SS, Babar ZU, Ahmed SI. Extensively drug-resistant typhoid
community and was validated in a population separate from fever in Pakistan. Lancet Infect Dis 2019; 19:242–3.
the population in which the composite variable was derived. 4. Yousafzai MT, Qamar FN, Shakoor S, et al. Ceftriaxone-resistant Salmonella
Typhi outbreak in Hyderabad City of Sindh, Pakistan: high time for the introduc-
Our findings support a concerted approach to typhoid elim- tion of typhoid conjugate vaccine. Clin Infect Dis 2019; 68(Suppl_1):S16–21.
ination through vaccination using safe and effective TCVs with 5. Mogasale VV, Ramani E, Mogasale V, Park JY, Wierzba TF. Estimating typhoid
fever risk associated with lack of access to safe water: a systematic literature re-
implementation of achievable WASH improvements in high- view. J Environ Public Health 2018; 2018:9589208.
risk communities. It is, however, important to note that the 6. Ross AG, Rahman M, Alam M, Zaman K, Qadri F. Can we ‘WaSH’ infectious dis-
eases out of slums? Int J Infect Dis 2020; 92:130–2.
combination of vaccination using TCVs and living in a Better 7. Brockett S, Wolfe MK, Hamot A, Appiah GD, Mintz ED, Lantagne D.
WASH household still reflected only a 70% reduction in Associations among water, sanitation, and hygiene, and food exposures and

1686 • CID 2022:75 (15 November) • Tadesse et al

 typhoid fever in case-control studies: a systematic review and meta-analysis. Am J 12. Theiss-Nyland K, Qadri F, Colin-Jones R, et al. Assessing the impact of a
Trop Med Hyg 2020; 103:1020–31. Vi-polysaccharide conjugate vaccine in preventing typhoid infection among
8. Patel PD, Patel P, Liang Y, et al. Safety and efficacy of a typhoid conjugate vaccine Bangladeshi children: a protocol for a phase IIIb trial. Clin Infect Dis 2019; 68-
in Malawian children. N Engl J Med 2021; 385:1104–15. (Suppl 2):S74–82.
9. Qadri F, Khanam F, Liu X, et al. Protection by vaccination of children against ty- 13. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations.
phoid fever with a Vi-tetanus toxoid conjugate vaccine in urban Bangladesh: a J Am Stat Assoc 1958; 53:457–81.
cluster-randomised trial. Lancet 2021; 398:675–84. 14. Breslow NE, Clayton DG. Approximate inference in generalized linear mixed
10. Shakya M, Colin-Jones R, Theiss-Nyland K, et al. Phase 3 efficacy analysis models. J Am Stat Assoc 1993; 88:9–25.
of a typhoid conjugate vaccine trial in Nepal. N Engl J Med 2019; 381: 15. R Core Team. R: a language and environment for statistical computing. Vienna,
2209–18. Austria: R Foundation for Statistical Computing, 2019.
11. Im J, Islam MT, Ahmmed F, et al. Can existing improvements of water, sanitation, 16. Brooks ME, Kristensen K, van Benthem KJ, et al. glmmTMB balances speed and
and hygiene (WASH) in urban slums reduce the burden of typhoid fever in these flexibility among packages for zero-inflated generalized linear mixed modeling. R
settings? Clin Infect Dis 2021;72:e720–6. J 2017; 9:378–400.

Typhoid Prevention by Vaccine and WASH • CID 2022:75 (15 November) • 1687