Phase I of the Surveillance for Enteric Fever in Asia Project (SEAP): An Overview and Lessons Learned

Abstract Objective The objective of Phase I of the Surveillance for Enteric Fever in Asia Project (SEAP), a multiphase surveillance study characterizing the burden of disease in South Asia, was to inform data collection for prospective surveillance and to capture clinical aspects of disease. Methods Through a retrospective record review conducted at hospitals in Bangladesh, India, Nepal, and Pakistan, we examined laboratory and clinical records to assess the culture positivity rate for Salmonella Typhi and Salmonella Paratyphi, age and sex distribution, and antimicrobial susceptability in each country. Results Of all blood cultures performed in Bangladesh, India, Nepal, and Pakistan, 1.5%, 0.43%, 2%, and 1.49%, respectively, were positive for S. Typhi and 0.24%, 0.1%, 0.5%, and 0.67%, respectively, were positive for S. Paratyphi. A higher proportion of laboratory-confirmed infections in Bangladesh and Pakistan were aged ≤5 years, while India and Nepal had a higher proportion of participants aged 15–25 years. In all countries, the sex of the majority of participants was male. The majority of isolates in all countries were resistant to fluoroquinolones, with a high proportion also resistant to ampicillin, chloramphenicol, and trimethoprim-sulfamethoxazole. Discussion Enteric fever remains endemic in South Asia. Data generated by this study can help inform strategies for implementation and evaluation of prevention and control measures.

Enteric fever, a preventable illness caused by the organisms Salmonella enterica subspecies enterica serovar Typhi or serovar Paratyphi, is estimated to cause >15 million illnesses and nearly 153 000 deaths worldwide annually [1,2]. These organisms are transmitted through food and water contaminated with fecal matter, and their control ultimately depends on access to potable water, as well as safely managed sanitation and hygienic practices in homes and in food production. Vaccines that protect against typhoid fever have been available for many years, and in March 2018, the World Health Organization (WHO) recommended the use of a newly prequalified typhoid conjugate vaccine (TCV) for typhoid fever prevention in countries of endemicity [3]. Compared with the 2 previously available vaccines, TCV provides longer-lasting protection, requires fewer doses, and is additionally suitable for children aged 6 months to <2 years, allowing delivery through routine childhood immunization programs [3][4][5]. Furthermore, the TCV was included in the 2019-2020 funding window of Gavi, the Vaccine Alliance [6]. To deliver the TCV to susceptible populations, it requires the support of country governments in many low-and middle-income countries, as well as other stakeholders, such as Gavi, the Vaccine Alliance. However, the absence of credible estimates of the burden of disease has made it difficult to provide policymakers with accurate estimates of the cost and potential impact of targeted interventions for prevention and control of enteric fever, including TCV.
One contributing factor to uncertainty about the burden of disease is the lack of reliable diagnostic testing. The most consistently valid method for laboratory diagnosis is blood culture, estimated to be only 61% sensitive [7] in a recent meta-analysis, and may be dependent on sample volume [8]. In many low-and middle-income countries, blood culture is also largely unavailable, particularly in rural areas. Additionally, the widespread use of antibiotics to treat typhoid empirically may reduce the sensitivity of blood culture and further reduce estimates of disease burden; in Nepal, 35% of participants had taken antibiotics prior to seeking healthcare at the study hospitals [9].
Over the last 2 decades, several surveillance studies have attempted to characterize the burden of enteric fever, including the Diseases of the Most Impoverished Program and the Typhoid Fever Surveillance in Africa Program [10,11]. While these studies were among the first prospective, population-based studies that estimated the incidence of enteric fever in Asia and sub-Saharan Africa, they have left several knowledge gaps, including the incidence of specific complications and the spectrum of disease and severity. Data on the cost of illness and the geographic spread of strains, including antimicrobial-resistant ones, are also limited. Since these studies, the Bill and Melinda Gates Foundation has invested in several new surveillance studies, including the Severe Typhoid in Africa project and the Surveillance for Enteric Fever in Asia Project (SEAP). SEAP is a multiphase enteric fever surveillance study designed to fill knowledge gaps about the burden of disease in Bangladesh, India, Nepal, and Pakistan. With the ability to characterize the burden of enteric fever in 3 major cities and 1 peri-urban area within the selected countries, results from this project will be able to help inform the use of TCVs and other typhoid prevention and control efforts.
The project has 2 phases: Phase I, a retrospective record review, and Phase II, a prospective study with defined enrollment criteria, including specific case definitions. Phase I had 3 objectives: (1) identify site hospitals in selected countries where enteric fever is endemic, to assess the burden of disease through record review and inform data collection for Phase II surveillance; (2) ensure that the hospital had capacity to participate in Phase II surveillance; and (3) map cases geographically to confirm the feasibility of creating a well-defined catchment area for each hospital in which the incidence of enteric disease can be estimated during Phase II surveillance. This article will present the results of the first objective.

METHODS
Based on the available data in the literature on the geographic burden of enteric fever, we chose Bangladesh, India, Nepal, and Pakistan as potential country sites.

Study Site Selection
We conducted a standard assessment of existing hospitals, health clinics, and laboratories in the area to determine possible study sites. Key variables in the health facility assessment were as follows: (1) existing evidence on the prevalence of enteric fever; (2) a population that is representative of the overall demographic and socioeconomic characteristics of the surrounding population; (3) a strong laboratory capacity and inpatient (medical and surgical) and outpatient facilities; (4) a location geographically close to the population served by the study hospital; (5) the availability of population-based sociodemographic indicators from reliable sources (eg, bureau of statistics of the city/district/national level or demographic health surveys) or the ability to obtain a geographically representative sample, which will allow extrapolation of population size and demographic characteristics; and (6) a willingness and commitment to conduct the study on the basis of agreed upon aims and objectives. Based on this assessment, we chose several sites in each country, specified below, for inclusion in the study.

Bangladesh
Two sites were selected in Bangladesh: Dhaka Shishu Hospital, a 640-bed tertiary-care hospital, and Shishu Sasthya Foundation Hospital, a 200-bed hospital providing primary care only. Both serve a catchment area of 3.4 million people and are the 2 largest pediatric hospitals in Dhaka (serving patients aged <18 years). Owing to proximity, both hospitals are within the same catchment area, in an urban and peri-urban setting. Both hospitals are also sentinel sites in the WHO-coordinated Invasive Bacterial Vaccine-Preventable Disease surveillance network, which captures information electronically on hospitalized patients with suspected or laboratory-confirmed pneumonia, sepsis, and meningitis. The hospitals were able to use this surveillance network to create a platform to capture enteric fever cases, which facilitated data collection for Phase I.

India
Five sites were selected in India: the Postgraduate Institute of Medical Sciences, a mixed public-private tertiary hospital in Chandigarh with 1960 beds; Medanta Hospital, a private tertiary hospital in Gurugram (previously known as Gurgaon), Haryana, with 1250 beds; Christian Medical College, a private tertiary hospital in Vellore, Tamil Nadu, with 2800 beds; Apollo Gleneagles Hospital, a private tertiary hospital in Kolkata, West Bengal, with 750 beds; and Kasturba Medical College -Manipal University Hospital, a private tertiary hospital in Manipal, Karnataka, with around 2000 beds. All sites serve urban populations, except Kasturba Medical College, which serves a peri-urban population.

Nepal
Sites were chosen because of a prospective enteric fever surveillance study already in place [9], which facilitated data collection. The 4 selected sites were: Bayalpata Hospital, a 15-bed hospital serving a rural population in Achlam; Damauli Hospital, a district hospital serving the town of Vyas, which has a population of 43 000; Kirnetar Health Center, a clinic in rural Dolakha; and Dhulikhel Hospital in Kavrepalanchowk, serving a peri-urban population in the greater Kathmandu Valley and with catchment area of approximately 330 000 people.

Pakistan
Three sites in Pakistan were selected, consisting of the 3 hospitals in the Aga Khan University Hospital network: Aga Khan University Hospital, a 700-bed tertiary care hospital in Karachi; Aga Khan Hospital for Women and Children, a 48-bed secondary care hospital in Karachi; and Aga Khan Hospital Hyderabad, an 87-bed secondary care hospital in Hyderabad. Both Aga Khan Hospital for Women and Children and Aga Khan Hospital Hyderabad only provide care for women and children. These hospitals cover a catchment area of 3 million people.

Study Period, Data Sources, and Case Definitions
After site selection was completed, we conducted a retrospective review of data available on enteric fever from the selected hospitals, including suspected cases and blood culture-confirmed S. Typhi and S. Paratyphi infection. A minimum of 2 years of data were collected to assess any seasonal variation at the selected sites. Specific methods for the sites, including case definitions, study periods, and data sources, are described in Table 1.

Laboratory Data Collection
For each month of data collection, the following information was obtained from laboratories: the number of blood cultures performed, the number of blood cultures positive for S. Typhi and S. Paratyphi, and any available patient-level demographic information, such as age, sex, neighborhood of residence, and specimen collection date. For the retrospective phase, patient address was only collected up to the smallest administrative unit, to map the catchment area. For Bangladesh and Nepal, data were extracted retrospectively from prospective studies already in place.

Inpatient Department Data Collection
Using hospital record numbers and where available, study staff obtained charts of hospitalized participants with laboratory-confirmed S. Typhi or S. Paratyphi infection and abstracted clinical information, including demographic characteristics, treatments, investigations, clinical manifestations, and patient outcomes.
Data were abstracted using standard instruments and entered electronically for analysis, which was performed at the country level. Differences in sex and age distributions were analyzed using Wilcoxon rank sum tests and differences in proportions. Blood culture positivity rates and antimicrobial resistance patterns were descriptively reviewed. More details about the analyses performed are available in the country-specific articles published in this supplement [9,[12][13][14].

Ethical Review
The study protocols were reviewed and approved by the Centers for Disease Control and Prevention (Atlanta, GA); institutional ethics committees at Pakistan); and the Ethical Review Committee of Bangladesh Institute of Child Health (Dhaka, Bangladesh). Identifying personal information was accessible only to evaluation staff at the sites. All data reported to investigators were deidentified.

RESULTS
Overall results from sites in each country, including the number of blood cultures and blood culture positivity rate (total and by organism), are shown in Table 2  The age distributions of participants with laboratory-confirmed cases varied between the 4 countries (see Figure 1). Bangladesh had the youngest participants for both S. Typhi (median, 3 years; interquartile range [IQR], 2-6 years) and S. Paratyphi (3 years; IQR, 2-6 years) infection, but this was expected as the 2 site hospitals served pediatric patients only. In Pakistan, which also captured more pediatric cases, participants with S. Typhi infection were significantly younger, compared with S. . The difference in ages between organisms was significant in India (P = .01) but not in Nepal. In all 4 countries, the sex of the majority of the participants was male (68% in India, 61% in Pakistan, 55% in Bangladesh, and 52% in Nepal).
Antimicrobial resistance patterns for each country, by organism, are shown in Figure 2. Overall, a higher proportion of S. Typhi isolates were resistant to chloramphenicol, ampicillin, and trimethoprim-sulfamethoxazole, compared with S. Complete results are available in the country-specific articles published in this supplement [9,[12][13][14].

DISCUSSION
This study generated data on enteric fever prevalence among febrile patients, by age, sex, and time (month/year), along with information on other clinical factors and antimicrobial resistance among hospitalized participants with enteric fever from sentinel sites in 4 countries, allowing for cross-site comparison. These results were critical to informing data collection for the prospective phase of the SEAP. While prospective surveillance will occur in India through a separate surveillance system (the National Surveillance System for Enteric Fever in India), we have identified sites in Bangladesh, Nepal, and Pakistan that continue to experience a sustained burden of enteric fever and are capable of contributing useful information on disease burden in prospective surveillance.
In Pakistan, S. Typhi infections were concentrated around younger ages, and while Bangladesh had pediatric cases only, because the sites were pediatric hospitals, nearly a quarter of S. Typhi-infected participants were <2 years old. This is an important observation, since these ages represent a population that can be protected with TCVs if a vaccination program is implemented. While overall the burden has shifted to an older population in India and Nepal as compared to Pakistan and Bangladesh, there remains a considerable burden of S. Typhi at the younger ages and a need for TCV implementation; the older population would also benefit from the WHO's recommendation for a catch-up campaign beginning at 15 years [3]. These differences in ages should be viewed in the context of the study design, though, because the differences could be partially explained by trends in healthcare-seeking behavior. This is also the case for sex distribution; literature has shown that generally a higher proportion of males seek care at the hospital, especially in younger ages [15]. Prospective surveillance performed in these countries will allow us to adjust incidence estimates based on reported care-seeking behavior and to estimate rates of infection by sex and age group. We were also able to review current patterns of antimicrobial resistance in all 4 countries. There is still a high proportion of S. Typhi isolates resistant to chloramphenicol, ampicillin, and trimethoprim-sulfamethoxazole, especially in Pakistan and Bangladesh [12,14]. Since the proportion of isolates in these countries that are nonsusceptible to ciprofloxacin is nearing 100%, this leaves third-generation cephalosporins, azithromycin, and carbapenems as the last remaining oral antibiotics for treating typhoid. As we have identified a few isolates in Nepal and India resistant to these agents [9,13], continued close monitoring of antimicrobial resistance is essential to inform appropriate treatment recommendations.
SEAP Phase I has yielded valuable information on the enteric fever burden across multiple settings in South Asia, but the results are best considered in the context of the chosen study settings. There may be an upward bias in incidence estimates generated in urban populations when compared with rural populations, likely because of transmission drivers such as population density and scarce access to clean water and sanitation [16]. Indeed, data from the Nepal study described in this supplement found that the prevalence of enteric fever among febrile individuals was strongly correlated with population density [9].
In addition to limitations in in-country generalizability due to the selected sites, SEAP Phase I provides an important perspective on key issues and limitations of retrospective data. Data obtained were limited in terms of comparability across settings and period. The data from each setting were highly variable with regard to availability, including variability for the surveillance period, age groups included in surveillance, and surveillance methods. Perhaps most significant was the variability in case definitions used for participants with suspected cases for whom a blood culture was requested. The lack of standardized criteria for requesting blood culture makes the diagnosis of enteric fever highly subjective to the treating clinicians' opinions. This is illustrated in Nepal, where half of individuals with a fever duration of >3 days were given a provisional clinical diagnosis of typhoid, but fewer than 5% of these individuals had culture-confirmed infection [9]. Furthermore, clinicians in resource-poor settings may be less likely to prescribe testing for poorer patients, who are possibly at a higher risk of enteric fever and subsequent poor outcomes. Additionally, hospital-based studies have an inherent limitation in that they underestimate the disease burden in the community, owing to differential health-seeking behaviors [17]. Active, community-based surveillance, is also more likely to be able to identify complications and sequelae that occur in patients with typhoid fever after discharge from a hospital or treatment at an outpatient clinic. These limitations and others [18], combined with nonstandardized case definitions, have important implications for disease burden and incidence estimations, which form the basis of national and global policies [1,2]. The data from Phase I were also limited to what was available retrospectively through existing data sources, primarily clinical records from a few sites. While laboratory data were relatively more complete and available, linking laboratory and clinical data sources presented difficulties. Clinical records were either missing or did not adequately capture the required information for a large proportion of laboratory-confirmed cases, specifically among individuals seen in outpatient departments. This prohibited sufficient characterization of the clinical profile of laboratory-confirmed cases and differences in clinical profiles by antimicrobial resistance. This is an especially critical lesson learned in the era of rising antimicrobial resistance and a potentially increased severity of illness associated with resistant strains [19][20][21]. Phase II of the SEAP will address many of these limitations inherent to retrospective study design. For example, the protocol for Phase II has a clearly defined study population, using a created catchment area to define eligibility, and standardized case definitions, which differ by enrollment location within the hospital (eg, outpatient, inpatient, and surgical settings). We also plan to prospectively collect robust data through multiple sources, including patient interviews, laboratory records, and medical charts, to improve the quality and amount of available information and its potential to be appropriately linked and analyzed. Unlike Phase I, Phase II will be able to estimate incidence on the basis of data from hospital-based surveillance, using the low-cost hybrid method outlined by Luby et al [17]. We plan to obtain the proportion of participants with disease matching our case definition who sought healthcare and create age-stratified incidence by using healthcare-seeking behavior, as well as to adjust for a variety of factors related to potential biases as described by another article in this supplement [22], to deliver more-discrete and accurate estimates to country governments and decision makers. SEAP Phase I is one of the few multicenter studies in South Asia that have attempted to retrospectively assess the enteric fever burden by using existing clinical and disease surveillance systems. With investment in capacity building, SEAP is building a low-cost and sustainable surveillance system that can be leveraged for ongoing surveillance to evaluate the impact of interventions, including vaccines, by integrating data capture into existing site infrastructure. It is unlikely that vaccines will entirely eliminate typhoid, shown by modeling in South Asia, but that a combination of water, sanitation, and hand hygiene interventions and vaccination may eliminate typhoid in settings where it is endemic [23]. Since the recent availability of a licensed and WHO-prequalified typhoid conjugate vaccine (Typbar-TCV; Bharat Biotech, India), which has been shown to be immunogenic in children aged <2 years, demonstrating vaccine effectiveness and public health impact is more important than ever [5,[24][25][26]. SEAP Phase II will produce not only accurate estimates of the burden of disease in selected areas in Bangladesh, Nepal, and Pakistan, but also baseline data for future vaccine effectiveness studies in the site catchment areas, as well as evaluate the impact of additional long-term interventions, such as improvements in water and sanitation infrastructure.