Illness Severity and Outcomes Among Enteric Fever Cases From Bangladesh, Nepal, and Pakistan: Data From the Surveillance for Enteric Fever in Asia Project, 2016–2019

Abstract Background Enteric fever can lead to prolonged hospital stays, clinical complications, and death. The Surveillance for Enteric Fever in Asia Project (SEAP), a prospective surveillance study, characterized the burden of enteric fever, including illness severity, in selected settings in Bangladesh, Nepal, and Pakistan. We assessed disease severity, including hospitalization, clinical complications, and death among SEAP participants. Methods We analyzed clinical and laboratory data from blood culture–confirmed enteric fever cases enrolled in SEAP hospitals and associated network laboratories from September 2016 to September 2019. We used hospitalization and duration of hospital stay as proxies for severity. We conducted a follow-up interview 6 weeks after enrollment to ascertain final outcomes. Results Of the 8705 blood culture-confirmed enteric fever cases enrolled, we identified 6 deaths (case-fatality ratio, .07%; 95% CI, .01–.13%), 2 from Nepal, 4 from Pakistan, and none from Bangladesh. Overall, 1.7% (90/5205) of patients recruited from SEAP hospitals experienced a clinical complication (Bangladesh, 0.6% [18/3032]; Nepal, 2.3% [12/531]; Pakistan, 3.7% [60/1642]). The most identified complications were hepatitis (n = 36), septic shock (n = 22), and pulmonary complications/pneumonia (n = 13). Across countries, 32% (2804/8669) of patients with hospitalization data available were hospitalized (Bangladesh, 27% [1295/4868]; Nepal, 29% [455/1595]; Pakistan, 48% [1054/2206]), with a median hospital stay of 5 days (IQR, 3–7). Conclusions While defined clinical complications and deaths were uncommon at the SEAP sites, the high proportion of hospitalizations and prolonged hospital stays highlight illness severity and the need for enteric fever control measures, including the use of typhoid conjugate vaccines.

Before the widescale implementation of antibiotics, an estimated 10-15% of patients with enteric fever developed clinical complications, the most severe including intestinal perforation, gastrointestinal bleeding, and neurological complications [4]. Other complications such as hepatitis, pneumonia, myocarditis, and cholecystitis are well described [5]; however, there is a paucity of recent data on the spectrum and frequency of clinical complications [6,7].
Additionally, determining the case fatality remains difficult for several reasons [7]. Global disease burden and mortality estimations have typically relied on available data from multiple sources; however, methodological differences, varying sample sizes, and the use of different confirmatory tests limit the ability to pool data across various studies. Existing reports of case fatality range from 0% to 23% depending on the geographic location and surveillance method used [6][7][8][9]. The most commonly estimated case-fatality ratio is 1% [10], although estimates are higher among those infected with multidrug-resistant strains and differ among age groups and regions [4,[7][8][9]11].
Recent data suggest a decline in enteric fever mortality, likely due to improvements in clinical capacity and access to effective treatments in endemic countries [8]. Updated mortality estimates, rates of complications, and severity data are necessary to understand the burden of severe typhoid and inform health policy, especially with growing antimicrobial resistance [7,12]. The Surveillance for Enteric Fever in Asia Project (SEAP), a multicountry, prospective, population-based surveillance project, was established to characterize the burden of enteric fever in Bangladesh, Nepal, and Pakistan [13]. Here, we describe illness severity, including hospitalization, duration of hospital stays, clinical complications, and deaths, among blood cultureconfirmed enteric fever cases over 36 months of surveillance in the SEAP study.

Study Setting
We conducted clinical and laboratory surveillance for enteric fever from September 2016 to September 2019 at selected hospitals and through enhanced laboratory surveillance at additional participating hospitals and clinics (laboratory network sites) that provide both outpatient and inpatient services. SEAP sites in Dhaka, Bangladesh, included 2 large pediatric hospitals and 3 laboratories (serving both adult and pediatric populations); in Nepal, 1 hospital in Kathmandu and 1 in Kavrepalanchok, plus 8 laboratories; and in Karachi, Pakistan, 2 tertiary hospitals, 3 laboratories, and surgical wards in 2 additional hospitals. Study site selection was described previously; in brief, Bangladesh, Nepal, and Pakistan were selected as countries with a known burden of enteric fever. A review of health facilities and laboratories led to the inclusion of several sites in each country that had existing evidence of enteric fever burden, the microbiological infrastructure to identify enteric fever by blood culture, inpatient and outpatient facilities, and the ability to obtain population-based sociodemographic data of the facility catchment area [14].

Definitions
Complications were defined as a diagnosis of a comorbid condition confirmed by the treating physician as a complication of enteric fever. Hospital admission and duration of hospital stay were used as proxies for disease severity. We defined a hospitalization day as one that included an overnight stay. Fluoroquinolone resistance included reduced susceptibility to ciprofloxacin, ofloxacin, levofloxacin, moxifloxacin, and gatifloxacin. Third-generation cephalosporin resistance included resistance to ceftriaxone, cefixime, cefotaxime, cefpodoxime, and ceftazidime. Extensively drug-resistant (XDR) S. Typhi is characterized as resistant to first-line antibiotics (chloramphenicol, ampicillin, and trimethoprimsulfamethoxazole), fluoroquinolones, and third-generation cephalosporins [12].

Inclusion Criteria
We selected patients with blood culture-confirmed Typhi or Paratyphi A of any age who (1) presented to outpatient departments, were identified in inpatient/surgical wards, or identified through hospital laboratories at SEAP hospitals or (2) were identified at laboratory network sites. Inclusion criteria by enrollment location are described in Table 1.

Data Collection
Study teams reviewed the medical charts of all participants from SEAP hospitals at the time of the enrollment visit and collected data on patient age, medications prescribed, duration of hospital admission for those hospitalized, complications, and clinical outcomes. Medical charts from laboratory network patients were not available for review. To ascertain final patient outcomes (eg, resolved, died), study teams attempted to contact all patients with blood culture-confirmed enteric fever enrolled in the study by telephone by calling up to 5 times over a 2-week period beginning 6 weeks following enrollment. The study teams also collected phone numbers of family members to contact if the patient could not be reached directly. If a patient sought medical care related to his or her illness during the 6 weeks between enrollment and follow-up, study teams attempted to obtain medical records from those visits to perform chart reviews.

Statistical Analysis
We calculated frequencies of complications and outcomes. We conducted bivariate comparisons using Pearson's chi-square test and used the Wilcoxon rank-sum test for assessing the differences in the median values of continuous variables. Country-specific univariable and multivariable logistic regression models were used to identify potential predictors of hospitalization among cases identified from the SEAP sites, including age, gender, serovar (S. Typhi or S. Paratyphi A), fluoroquinolone resistance, third-generation cephalosporin resistance, prior antibiotic use, and time in days from fever onset to hospital presentation. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated from the models.
Similarly, country-specific univariable and multivariable negative binomial regression models were used to identify predictors of the duration of hospital stay among patients hospitalized at the SEAP hospitals, including the same predictors stated above and empiric treatment at hospital presentation. Relative risks and 95% CIs were estimated from the models. For all analyses, a P value less than .05 was considered statistically significant.

Ethical Considerations
Informed consent was obtained from all participants or parents/ guardians of participants, as well as verbal assent from participants aged 11-17 years old. The study protocol was approved by the Bangladesh Institute of Child Health Ethical Review Committee, Nepal Health Research Council Ethical Review Board, Pakistan Ethical Review Committee, and Stanford University Institutional Review Board. This project was approved by the Centers for Disease Control and Prevention (CDC) human subjects review as research with CDC not directly engaged, so a full institutional review board approval was not required.

RESULTS
A total of 34 748 patients were enrolled from SEAP sites in Bangladesh, Nepal, and Pakistan, of whom 8705 (25%) were blood culture-confirmed for enteric fever; 5215 (60%) cases were identified from the SEAP hospitals and 3490 (40%) cases were identified through laboratory network sites. Of the confirmed cases, 7591 (87%) were positive for S. Typhi and 1114 (13%) were positive for S. Paratyphi A ( Table 2). There was no laboratory evidence of bacterial coinfection among the blood culture-confirmed cases.

Six-Week Follow-up
Of the 8705 confirmed cases, 3 died during hospitalization. Of the remaining 8702 patients, 8092 (93%) patients or patient caregivers participated in the follow-up interview at 6 weeks: Bangladesh, 94% (4562/4873); Nepal, 94% (1498/1602); and Pakistan, 91% (2029/2230). Study teams were unable to interview 610 (7%) patients; 9 refused to participate and the others could not be reached. In Pakistan, 64% of patients with typhoid with known outcomes were infected with strains resistant to third-generation cephalosporins, as were 71% of those lost to follow-up and 100% of the fatal cases (Table 4). A total of 9% (716/8092) of participants contacted reported symptoms following the enrollment visit; 64% (455/716) sought additional medical care (Table 2), and 69% (313/455) of those patients had medical records available for review. One patient was diagnosed with an intestinal perforation 1 day following the hospital outpatient visit, an 18-year-old male from Pakistan with XDR typhoid.

Case-fatality Ratio
Of the 8095 participants with known outcomes (including the 3 who died during hospitalization), we identified 6 deaths, all culture-positive for S. Typhi (overall case-fatality ratio of .07%; 95% CI, .01-.13%). Two fatal cases occurred in Nepal, both adult males (33 and 65 years old) who presented to laboratory network sites and reportedly died after hospital discharge. Four fatal cases occurred in Pakistan, 2 male and 2 female patients (8 months-8 years old). Three of these patients died at the SEAP hospitals during hospitalization and 1 died after the enrollment visit. No deaths were identified in Bangladesh. None of the fatal cases included a diagnosis of intestinal perforations. In Pakistan, 3 patients who died reportedly consumed antibiotics prior to enrollment. All isolates from fatal cases were resistant to fluoroquinolones, and those in Pakistan were all XDR.

Hospitalization
Of the 8669 patients with confirmed enteric fever with available hospitalization data, 2804 (32%) were hospitalized. Overall, 2576 (92%) of the hospitalized patients with confirmed enteric fever were infected with S. . Among patients infected with S. Typhi, the highest proportion of hospitalizations occurred among children younger than 2 years old in Bangladesh and Nepal and among those 16 years or older in Pakistan. Across countries, the highest proportion of hospitalizations among those with S. Paratyphi A infection was among children younger than 5 years old ( Figure 1).
Infection with S. Typhi, reported prior antibiotic use, and delayed care-seeking were significant common predictors of hospitalization across countries in the univariable models. In the multivariable models, there was a consistent trend in the direction of the effect for Typhi infection and prior antibiotic use, which was statistically significant in 2 of the 3 countries.
Resistance to fluoroquinolones and third-generation cephalosporins was associated with higher odds of hospitalization in Pakistan (Table 5). Data on duration of hospital stay were available for 96% (1911/1981) of patients admitted to SEAP hospitals (Bangladesh, 939/940; Nepal, 150/150; Pakistan 822/891). The median duration of hospital stay was the same for patients with Typhi and Paratyphi A, 5 days (interquartile range, 3-7 days) ( Table 2). There were no common predictors for duration of hospital stay across countries. Third-generation cephalosporin resistance was associated with 48-54% longer hospital stays in Pakistan (Table 6).

DISCUSSION
The SEAP study generated comprehensive data on illness severity and outcomes among a large cohort of blood cultureconfirmed cases of enteric fever from high-burden settings in Bangladesh, Nepal, and Pakistan. These data are important for understanding the burden of severe typhoid, and the burden that enteric fever places on the healthcare system. This is particularly relevant in settings like South Asia where many severely ill patients are refused admission due to hospital capacity limitations [15].
Mortality is a critical measure of disease severity and is the main driver of typhoid disease burden estimates. A recent metaanalysis estimated the overall case fatality of enteric fever to be 2.49% [9]. While this is markedly higher than the case-fatality ratio identified from SEAP, several of the studies included in the meta-analysis with higher case-fatality rates utilized small sample sizes or included studies from as early as 1960. The 2017 Global Burden of Disease Study estimated an overall casefatality rate of 0.95%, with the South Asia-specific estimates closer to 1.1% [2]. The widely invoked case-fatality ratio of 1% was estimated using published data from hospital-based studies and separate community-based studies [10]. Because enteric fever is generally treated as an outpatient disease, estimates derived from inpatient surveillance exclude mild cases not admitted to the hospital and may overestimate the case-fatality ratio [16][17][18][19]. However, active surveillance at the community level detects and treats disease early, providing very limited information on uncommon outcomes like severe complications and death [16,20]. We observed such variation in mortality in the SEAP data, in which the case-fatality ratio among inpatients was almost 10 times higher than that of outpatients.
Alternatives to the tertiary hospitals in the SEAP settings are laboratory network sites, data from which are often not included in hospital-based surveillance. These facilities include community-based private consultation centers, clinics, and secondary hospitals, and constitute an important component of a comprehensive enteric fever surveillance system [19]. By  including laboratory network sites alongside our hospital-based surveillance, we were able to more accurately estimate the casefatality ratio compared with utilizing tertiary hospitals alone. While our overall case-fatality ratio of less than 1% is considerably lower than published estimates, it is comparable to recent findings from Dhaka, Bangladesh, that utilized a similar methodology as SEAP and included patients diagnosed through laboratory network sites [20,21].
Previous studies have reported higher mortality among patients infected with multidrug-resistant S. Typhi [7][8][9]. With improved clinical services and the use of effective antibiotics, case fatality among blood culture-confirmed enteric fever has reportedly decreased in recent decades [8,20,21]. However, with widespread antimicrobial resistance, there is a risk of increased severe outcomes. While we did not detect any cephalosporin resistance in Bangladesh or Nepal [22], a recent molecular surveillance study projected that resistance will develop in the region because there are documented cases in Pakistan and India [12,23]. All deaths from our population were among patients infected with drug-resistant S. Typhi, including XDR typhoid among all deaths in Pakistan, highlighting the threat that antimicrobial resistance poses to the currently low case-fatality ratio.
Similarly, clinical complications attributed to enteric fever by treating physicians were infrequently identified among SEAP patients. Hepatitis, septic shock, and pulmonary complications-the complications we commonly identified-are well documented in the literature [8]. Like the fatal cases, most patients with a complication were infected with organisms resistant to antibiotics commonly prescribed in the region. A study from Vietnam reported complications among an estimated 15% of patients with typhoid and found that clinical complications were associated with intermediate fluoroquinolone resistance during a time when these drugs were widely used to treat enteric fever [24]. Because we observed that patients with third-generation cephalosporin resistance were more likely to develop a typhoid-related complication, we anticipate the rates of severe disease to increase as this class of antimicrobials, including ceftriaxone and cefixime, is commonly prescribed empirically in the SEAP settings [25].
A considerable proportion of blood culture-confirmed enteric fever cases were hospitalized across the 3 countries, especially among children who would benefit from typhoid conjugate vaccines (TCVs), which are safe for children as young as 6 months old [26]. Similar to our findings from SEAP, a surveillance study for enteric fever in Bangladesh identified that almost 41% of patients with blood culture-confirmed enteric fever younger than 2 years old were hospitalized [21]. With the introduction of TCVs into routine immunization and improved access to safe water, food, and sanitation and hygiene facilities, we expect a reduction in typhoid hospitalizations, making hospital beds available for patients requiring inpatient care for other conditions [15]. Antibiotics are widely available without prescription in the communities under SEAP surveillance. We observed that hospitalization was associated with antimicrobial use and resistance, especially in Pakistan, which is currently experiencing an ongoing outbreak of XDR typhoid [12]. Delayed care-seeking and self-reported prior antibiotic use, while only moderately reliable [27], were also associated with higher odds of severe disease, consistent with previous findings [8,28]. Reasons for delayed care-seeking are unknown; possibly some patients presented to the study sites only after treatment failure, especially in cases of drug resistance [29]. Empiric treatment was also common in our sites [25]. Such treatment, in addition to indiscriminate antibiotic usage in settings already detecting drug-resistant enteric fever, may put patients at risk for poor outcomes due to delays in appropriate therapy [4,11,30]. Indeed, empiric treatment in Pakistan was associated with a longer duration of hospital stay, most probably due to ineffective initial treatment. Similarly, patients from Pakistan with evidence of XDR typhoid were admitted for approximately 50% longer than those with non-XDR typhoid. The longer duration of hospital stay may be attributed to morbidity but may also be driven by the need for intravenous carbapenem therapy. Typhoid vaccination could potentially reduce the overprescribing of antibiotics, the subsequent spread of antimicrobial resistance [26], and the overall burden on the healthcare system. An important limitation to our study is the potential to miss the most severe cases of enteric fever. To ascertain complications, data were abstracted from medical records of blood culture-confirmed patients who sought care at the SEAP hospitals. However, most patients with enteric fever likely seek care at pharmacies or informal care systems and never present to hospitals [31]. We may have missed severe disease and enteric fever deaths among those who did not present to a hospital or who did not obtain diagnostic tests, as these are costly and not always required before treatment [15,16]. Nevertheless, we were able to review over 5000 medical records of blood culture-confirmed cases. We may have missed deaths among patients lost to follow up, especially patients in Pakistan with XDR typhoid infections who would be more likely to have severe outcomes, including death. While our loss to follow-up was low, our estimated case-fatality ratio is likely biased downwards. Because it may be more difficult to contact families of individuals who died, those reached for follow-up may not be representative of the true case-fatality ratio. However, our hospitalization estimates-an important measure of severity-may be overestimated because we calculated the proportion hospitalized only among those patients with enteric fever presenting to a hospital or laboratory network site, which does not consider the large proportion of cases who seek care at pharmacies or other clinics, especially those with mild disease. Last, blood culture sensitivity is estimated to be approximately 60%-or lower for those with prior antibiotic use [32]. Therefore, we likely missed additional enteric fever cases presenting to our study sites who were not confirmed by blood culture, including patients with enteric fever-related intestinal perforations [33]. Future surveillance should prioritize reducing the loss to follow-up of blood culture-confirmed cases to improve the accuracy of case fatality and severe disease estimates.

Conclusions
While clinical complications and deaths were rarely identified in our large dataset, the high proportion of hospitalizations and prolonged hospital stays, reflecting substantial costs and morbidity, underscores the benefit of immediate interventions. Case fatality from enteric fever is not a static figure; it varies according to context and time. We found a lower-than-expected case-fatality ratio but noted that extensive drug resistance is associated with more severe disease and morality, and global data demonstrate a continuing trend in the emergence and spread of drug-resistant strains. Thus, continuing to monitor enteric fever mortality is important and can be added to existing hybrid surveillance at a low marginal cost. The World Health Organization's Strategic Advisory Group of Experts (SAGE) on Immunization has recommended TCV introduction in countries with a high burden of typhoid fever or with a high burden of antimicrobial-resistant S. Typhi, both scenarios applicable at the SEAP sites [34]. Patient outcomes, disease severity, and the impact of antimicrobial resistance should drive decision making on enteric fever interventions, including vaccine deployment. With support from Gavi, the Vaccine Alliance to introduce TCVs, more countries will need reliable burden data for vaccine implementation. It is also important to note that there is an identified burden of S. Paratyphi A, which is not covered by the current vaccines, further emphasizing the need to monitor trends in disease burden and severity among patients with enteric fever. Notes