Abstract
Background
In many resource-limited settings, hospitalization for community-acquired infection is common, but data regarding illness severity, etiology, and morbidity remain sparse.
Methods
We conducted a prospective observational study from May 2022 to August 2023 at 2 hospitals in northeast Thailand. Adults hospitalized with community-acquired infection were enrolled within 24 hours of admission and followed up to 28 days. We identified patients meeting sepsis criteria and assessed related epidemiology, management, and mortality risk factors.
Results
Of 1445 patients screened, 940 were enrolled. The median age was 60 years and preexisting diabetes mellitus was common (42%). Sixty-six percent of patients met sepsis criteria. Blood cultures and broad-spectrum antibiotics on admission were common (both >95%), although lactate measurement was performed in 43% of patients with sepsis. In patients with sepsis, critical illness outside the intensive care unit was common on medical ward admission, including respiratory failure (33%) and shock (21%). Tropical etiologies of infection included melioidosis (8%) and leptospirosis (4%), and gram-negative organisms accounted for 81% of bacteremia. Twenty percent of patients with sepsis died by 28 days. Sepsis-associated acute kidney injury (SA-AKI) on admission was independently associated with mortality (adjusted odds ratio, 2.07; 95% CI, 1.30–3.29; P = .002), and patients with SA-AKI had worse survival (P < .001) than those without.
Conclusions
In rural Southeast Asia, sepsis is common among patients hospitalized with infection and associated with substantial morbidity and mortality. Distinct pathogens and broad-spectrum antibiotics are common, even in the absence of sepsis. We identified several modifiable risk factors of death, including SA-AKI, potentially influencing initial management in similar settings.
Keywords: AKI, global infection, resource limited, sepsis, tropical infection
Sepsis disproportionally affects low- and middle-income countries (LMICs), with millions of estimated deaths annually [1]. In tropical regions of Southeast Asia, sepsis-related morbidity is particularly high and caused by a diverse group of pathogens [2]. The heterogeneity and global variation in this disease therefore necessitate further study to develop contextualized and comprehensive management plans [3].
The Third International Consensus Definition for Sepsis defined sepsis as infection-related host response dysregulation leading to organ dysfunction [4]. Implicit in this definition were measures of organ dysfunction typically requiring critical care interventions. As such, many studies of sepsis have focused on patients in the intensive care unit (ICU), even though patients with sepsis may be identified outside of such settings [5]. Furthermore, in resource-limited settings, patients frequently do not have access to critical care, or they receive critical care outside an ICU, requiring modifications of traditional management approaches [6, 7].
Studies of patients hospitalized with community-acquired infection are limited in Southeast Asia, despite its high estimated burden and unique tropical etiologies. Several regional studies of patients with suspected or confirmed sepsis in large academic or referral centers may miss broader populations of patients who are infected [2, 8]. Therefore, in this study, we sought to identify the frequency of sepsis among a large, prospectively enrolled cohort of patients admitted for community-acquired infection at 2 rural hospitals in northeast Thailand. We then assessed the causative infectious etiologies, management, risk factors, and infection-related outcomes in this unique and understudied population.
METHODS
Study Design, Participants, and Outcomes
We conducted a prospective observational study at 2 hospitals in northeastern Thailand from May 2022 through August 2023. Patients hospitalized at Mukdahan Hospital, Mukdahan province, Thailand, and Roi Et Hospital, Roi Et province, Thailand, were screened within 24 hours of hospital admission to determine if they met enrollment criteria of age ≥18 years and a primary admission diagnosis of suspected or confirmed infection. Exclusion criteria, regional and site information, and additional study design are available in the supplemental methods. Enrolled patients were then subsequently followed during their hospitalization. All patients, including postdischarge, were followed to or contacted at 28 days following enrollment to determine death after discharge. Discharged patients not contacted were not included in analyses related to 28-day mortality.
Definitions
Sepsis was defined in accordance with the Third International Consensus Definition for Sepsis [4]. A Sequential Organ Failure Assessment (SOFA) score was calculated by parameters within the first 24 hours of admission, and sepsis was defined as patients with suspected infection and a SOFA score ≥2. Modifications to the respiratory component of the SOFA score were necessary as arterial blood gas measurement is uncommon. For patients without an available partial pressure of arterial oxygen (PaO2), a ratio of the pulse-oximetric oxygen saturation to the fraction of inspired oxygen (SpO2/FiO2) was calculated for patients with an SpO2 <98% [9, 10]. As previously reported, including adoption in resource-constrained settings, SpO2/FiO2 equivalents for the SOFA PaO2/FiO2 cutoffs were calculated by the following equation: SpO2/FiO2 = 64 + [0.84 × (PaO2/FiO2)] [11, 12]. Otherwise, patients who were intubated or mechanically ventilated with SpO2 >98% and no available PaO2 were given a default respiratory SOFA score of 2 out of 4 (Supplementary Table 1) [8]. A Glasgow Coma Scale score was recorded at enrollment by the study team. To maintain consistency with prior sepsis investigations, when score components were not available (Supplementary Table 2), they were assumed to be normal [1, 13]. The dates of receipt of antibiotics, blood cultures, and lactate were available but not the minute/hour. Therefore, timing of sepsis bundles was limited to the calendar day of admission.
Acute kidney injury (AKI) was determined by criteria from the 2012 Kidney Disease: Improving Global Outcomes guidelines [14]. AKI was defined as a creatinine value within 24 hours of admission ≥150% of a baseline creatinine. If a baseline creatinine value was not available, an estimated value was calculated in patients without known chronic kidney disease per the “modification of diet in renal disease” equation, with an estimated glomerular filtration rate of 75 mL/min/1.73 m2. In all patients not requiring baseline peritoneal dialysis or hemodialysis, an increase in creatinine by 0.3 mg/dL during the first 48 hours of admission was also classified as AKI. Urine output was not available and so not included in the AKI definition. Additional definitions related to critical illness, presenting clinical syndromes, and infectious disease diagnoses can be found in the supplemental methods.
Statistical Analyses
Data were summarized by proportions for discrete variables and median (IQR) for continuous variables. Differences in proportions and medians between groups were assessed by χ2 and the Mann-Whitney test, respectively. To identify risk factors of 28-day mortality, 44 clinical characteristics available at enrollment were a priori identified. In univariate analyses, the association of each variable with 28-day mortality was calculated by logistic regression. To identify a set of variables for inclusion in multivariable models, all 44 clinical variables were subjected to logistic regression analysis by least absolute shrinkage and selection operator (LASSO) methodology, in which lambda was selected by the bayesian information criterion. The selected risk factors were confirmed with the lambda that minimizes the minimal mean squared prediction error [15]. The LASSO-identified risk factors were then assessed for their association with 28-day mortality in multivariable models, which included enrollment site. Curves for 28-day survival were compared via the log-rank test. P values <.05 were considered significant. Analyses were performed in Stata/SE version 14.2.
Patient Consent Statement
Written informed consent was obtained from all study participants or their surrogate decision makers. The study was approved by the ethics committees of Mukdahan Hospital (MEC 09/64), Roi Et Hospital (RE035/2564), the Mahidol University Faculty of Tropical Medicine (MUTM 2021-043-01), and the University of Washington (STUDY00012758).
RESULTS
Cohort Characteristics
Of 1445 patients screened, 940 were enrolled and included in analyses (Supplementary Figure 1). Enrolled patients were predominantly male (551/940, 59%) with a median age of 60 years (IQR, 49–70; Table 1, Supplementary Table 3). Notably, diabetes mellitus was common (396/940, 42%), and 44% (416/940) of patients were referred from another facility, typically on the same day as initial presentation. A total of 618 (66%) patients met sepsis criteria within 24 hours of admission. Patients with sepsis at enrollment tended to have more comorbidities vs those without sepsis by Charlson Comorbidity Index (P < .001), although diabetes mellitus was more common in patients without sepsis (P = .02). Patients with sepsis were more likely to have an admission diagnosis of pneumonia (272/618, 44%) than those without sepsis (82/322, 26%; P < .001). Conversely, patients with sepsis were less likely to have an admission diagnosis of an acute febrile illness (sepsis, 110/618, 18%; nonsepsis, 94/322, 29%; P < .001).
Table 1.
Characteristics of Patients With Community-Acquired Infection
| Median (IQR) or No. (%) | ||||
|---|---|---|---|---|
| Characteristics | Entire Cohort (N = 940) | Sepsis a (n = 618) | Nonsepsis (n = 322) | P Value |
| Demographics | ||||
| Age, y | 60 (49–70) | 61 (50–72) | 57 (46–68) | <.001 |
| Female sex | 389 (41) | 233 (38) | 156 (49) | <.01 |
| Rainy season presentation | 614 (65) | 390 (63) | 224 (70) | .05 |
| Preexisting conditions | ||||
| Charlson Comorbidity Index | 2 (1–4) | 3 (1–4) | 2 (1–3) | <.001 |
| Diabetes mellitus | 396 (42) | 243 (39) | 153 (48) | .02 |
| Hypertension | 356 (38) | 240 (38) | 116 (36) | .40 |
| Chronic kidney disease | 139 (15) | 115 (19) | 24 (8) | <.001 |
| Dyslipidemia | 97 (10) | 60 (10) | 37 (12) | .39 |
| Stroke | 60 (6) | 45 (7) | 15 (5) | .12 |
| Chronic lung disease | 55 (6) | 34 (6) | 20 (6) | .73 |
| Chronic cardiovascular disease | 41 (4) | 36 (6) | 5 (2) | <.01 |
| Chronic steroid use | 34 (4) | 22 (4) | 12 (4) | .90 |
| Cancer | 24 (3) | 17 (3) | 7 (2) | .59 |
| Chronic liver disease | 24 (3) | 24 (4) | 0 | <.001 |
| HIV | 17 (2) | 12 (2) | 5 (2) | .68 |
| Rheumatologic disorders | 16 (2) | 12 (2) | 4 (1) | .43 |
| Hospitalization | ||||
| Referral from another facility | 416 (44) | 350 (57) | 66 (21) | <.001 |
| Days to referral | 0 (0–0) | 0 (0–0) | 0 (0–0) | .93 |
| Admission ward | ||||
| Medical | 810 (86) | 509 (82) | 301 (93) | <.001 |
| Surgical | 28 (3) | 9 (2) | 19 (6) | |
| Intensive care unit | 102 (11) | 100 (16) | 2 (1) | |
| Length of hospitalization, d | 6 (3–13) | 6 (3–13) | 5 (3–11) | <.001 |
| Procedural drainage-debridement | 87 (9) | 56 (9) | 31 (10) | .78 |
| Presenting clinical syndromes b | ||||
| Pneumonia | 354 (38) | 272 (44) | 82 (26) | <.001 |
| Acute febrile illness | 204 (22) | 110 (18) | 94 (29) | <.001 |
| Gastrointestinal illness | 132 (14) | 81 (13) | 51 (16) | .25 |
| Skin or soft tissue infection | 27 (3) | 14 (2) | 13 (4) | .12 |
| Urinary tract infection | 73 (8) | 48 (8) | 25 (8) | .99 |
| Intracranial infection | 4 (0) | 3 (1) | 1 (0) | .7 |
| Abscess | 33 (4) | 14 (2) | 19 (6) | <.01 |
aSepsis defined as modified Sequential Organ Failure Assessment score ≥2 in patients with community-acquired infection.
bPresenting clinical syndromes based on primary admission diagnosis.
Initial Management and Outcomes by Sepsis Classification
On admission to the study hospitals, blood cultures were sent on 96% of patients (902/940; Table 2, Supplementary Table 4). Antibiotics were frequently given on admission (917/940, 98%) and overwhelmingly included broad gram-negative coverage (912/940, 97%). Ceftazidime and ceftriaxone were the most frequently received antibiotics at the study hospitals or referring sites (Supplementary Table 5). Antibiotics started on admission were continued for a median 6 days (sepsis, 6 days [IQR, 3–13]; nonsepsis, 4 days [IQR, 3–11]; P = .002). Finally, lactate was measured for 32% (302/940) of patients on admission and was more frequently measured for patients with sepsis as opposed to those without (264/618 [43%] vs 38/322 [12%]; P < .001). Given the inconsistency of lactate measurement, adherence to the recommended sepsis bundle on admission (blood cultures, antibiotics, and lactate measurement) per the 2021 Surviving Sepsis Campaign (SSC) occurred in only 41% of patients with sepsis (255/618) [16].
Table 2.
Management, Severity of Illness, and Outcomes of Patients With Community-Acquired Infection
| No. (%) or Median (IQR) | ||||
|---|---|---|---|---|
| Characteristic | Entire Cohort (N = 940) | Sepsis (n = 618) | Nonsepsis (n = 322) | P Value |
| Admission management | ||||
| Blood cultures sent on admission | 902 (96) | 596 (96) | 306 (95) | .30 |
| Antibiotics received on admission | 917 (98) | 604 (98) | 313 (97) | .62 |
| Broad gram-negative coverage a | 912 (97) | 603 (98) | 309 (96) | .17 |
| MRSA coverage b | 185 (20) | 126 (20) | 59 (18) | .45 |
| Duration, d | 6 (3–12) | 6 (3–13) | 4 (3–11) | .002 |
| Lactate measured on admission | 302 (32) | 264 (43) | 38 (12) | <.001 |
| Admission illness characteristics | ||||
| SOFA | 3 (1–7) | 5 (3–9) | 0 (0–1) | <.001 |
| qSOFA | 1 (1–2) | 2 (1–2) | 1 (0–1) | <.001 |
| Acute kidney injury c | 261 (28) | 239 (39) | 22 (7) | <.001 |
| Outcomes during hospitalization | ||||
| Mechanical ventilation | 322 (34) | 315 (51) | 7 (2) | <.001 |
| Vasoactive medication | 275 (29) | 267 (43) | 8 (3) | <.001 |
| New kidney replacement therapy | 44 (5) | 43 (7) | 1 (0) | <.001 |
| Intermittent hemodialysis | 44 (5) | 43 (7) | 1 (0) | <.001 |
| Peritoneal dialysis | 0 | 0 | 0 | |
| Mortality outcomes | ||||
| 7 d | 78 (8) | 75 (12) | 3 (1) | <.001 |
| Days to death | 2 (1–5) | 2 (1–5) | 5 (1–6) | .35 |
| 28 d | 132 (14) | 125 (20) | 7 (2) | <.001 |
| Days to death | 6 (2–13) | 6 (2–12) | 20 (6–23) | .10 |
| Lost to 28-d follow-up | 6 (1) | 4 (1) | 2 (1) | >.99 |
Abbreviations: KDIGO, Kidney Disease: Improving Global Outcomes; MRSA, methicillin-resistant Staphylococcus aureus; qSOFA, quick Sequential Organ Failure Assessment; modified SOFA, Sequential Organ Failure Assessment.
aGram-negative coverage on or prior to admission, including receipt of a third- or fourth-generation cephalosporin, carbapenem, piperacillin-tazobactam, or levofloxacin.
bCommunity-acquired MRSA coverage on or prior to admission, including receipt of vancomycin, doxycycline, or clindamycin.
cAcute kidney injury based on KDIGO definition of 150% of estimated baseline creatinine in patients without chronic kidney disease or increase in creatinine by 0.3 mg/dL within 48 hours in any patient.
A total of 132 patients (14%) died within 28 days after enrollment (Table 2, Supplementary Table 4). Highlighting local cultural practices, 67 (51%) of those who died by 28 days were discharged from the hospital prior to their death, and 41 (62%) died within 1 day of discharge. Death within 28 days of enrollment was more common in patients with sepsis (125/618, 20%) than those without sepsis (7/322, 2%; P < .001). Out of 940 enrolled patients, 6 were lost to follow-up at 28 days after enrollment.
Infectious Diseases by Sepsis Classification
We next characterized the infectious diseases represented in the cohort (Table 3, Supplementary Table 6). Overall, 18% (164/940) had a positive blood culture from admission. Bacteremia was more common in patients with sepsis (130/618, 21%) than those without sepsis (34/322, 10%; P < .001). Gram-negative bacteria accounted for most positive blood cultures (133/164, 81%), including patients with sepsis (102/130, 78%) and without (31/34, 91%). Burkholderia pseudomallei, the antibiotic-resistant bacteria endemic in northeast Thailand and the cause of the severe tropical infection melioidosis, was the most common bacteria identified in blood cultures. We compared other tropical infectious diseases based on final hospital diagnosis. Leptospirosis was diagnosed in 36 of 940 (4%) patients, with 92% (33/36) meeting sepsis criteria. Other infectious diseases were diagnosed in the cohort, such as acute tuberculosis (24/940, 3%), scrub typhus (14/940, 2%), dengue (7/940, 1%), and nonbacteremic melioidosis (20/940, 2%). Notably, no specific bacteremia pathogen differed by sepsis category, and the majority of the cohort did not have a specific infectious etiology identified.
Table 3.
Infectious Diseases in Patients With Community-Acquired Infection
| Characteristic | Entire Cohort (N = 940) | Sepsis (n = 618) | Nonsepsis (n = 322) | P Value |
|---|---|---|---|---|
| Bloodstream infection | 164 (18) | 130 (21) | 34 (10) | <.001 |
| Gram-negative bacteremia | 133 (14) | 102 (17) | 31 (10) | .004 |
| Burkholderia pseudomallei | 58 (6) | 43 (7) | 15 (5) | .17 |
| Escherichia coli | 39 (4) | 27 (4) | 12 (4) | .64 |
| Klebsiella spp | 21 (2) | 16 (3) | 5 (2) | .37 |
| Pseudomonas spp | 3 (0) | 3 (1) | 0 | .56 |
| Acinetobacter baumannii | 5 (1) | 5 (1) | 0 | .17 |
| Salmonella serogroup D | 5 (1) | 5 (1) | 0 | .17 |
| Citrobacter koseri | 2 (0) | 2 (0) | 0 | .55 |
| Vibrio parahemolyticus | 1 (0) | 1 (0) | 0 | >.99 |
| Aeromonas spp | 1 (0) | 1 (0) | 0 | >.99 |
| Gram-positive bacteremia | 32 (3) | 29 (5) | 3 (1) | .002 |
| Staphylococcus aureus | 11 (1) | 10 (2) | 1 (0) | .11 |
| Streptococcus pneumoniae | 4 (0) | 4 (1) | 0 | .31 |
| Streptococcus pyogenes | 6 (1) | 5 (1) | 1 (0) | .67 |
| Group G Streptococcus | 3 (0) | 3 (1) | 0 | .56 |
| Group D Streptococcus | 2 (0) | 2 (0) | 0 | .55 |
| Other Streptococcus spp | 2 (0) | 2 (0) | 0 | .55 |
| Enterococcus faecalis | 4 (0) | 3 (1) | 1 (0) | .67 |
| Polymicrobial | 5 (1) | 4 (1) | 1 (0) | >.99 |
| Contaminant organisms a | 35 (4) | 28 (5) | 7 (2) | .07 |
| Other infectious etiologies b | ||||
| Leptospirosis | 36 (4) | 33 (5) | 3 (1) | <.001 |
| Acute tuberculosis disease | 24 (3) | 10 (2) | 14 (4) | .02 |
| Nonbacteremic melioidosis | 20 (2) | 12 (2) | 8 (3) | .64 |
| Scrub typhus | 14 (2) | 12 (2) | 2 (1) | .16 |
| Dengue | 7 (1) | 4 (1) | 3 (1) | .70 |
| Malaria | 1 (0) | 1 (0) | 0 | >.99 |
aContaminant organisms include coagulase-negative Staphylococcus spp, alpha-hemolytic Streptococcus spp, Propionibacterium spp, Corynebacterium spp, Burkholderia cepacia, or Bacillus spp if no other clinical evidence existed suggestive of infection.
bListed infectious etiologies were based on final diagnoses according to local hospital testing and diagnostic protocols. Nonbacteremic melioidosis diagnoses were based on positivity of any nonblood culture for B pseudomallei.
Critical Illness in Sepsis
As many patients in resource-limited settings receive medical care outside of traditional ICUs, we next sought to characterize the medical management of patients with sepsis inside and outside the ICU (Table 4). In 618 patients with sepsis, 176 (28%) were initially admitted to an ICU while 444 (72%) were admitted to a ward. Notably, critical care was still common in patients with sepsis outside the ICU, as 27% (121/444) required mechanical ventilation and 21% (92/444) required vasoactive medications within 24 hours of admission. Patients with sepsis initially admitted to the ICU had higher 28-day mortality (34%) and frequency of AKI on admission (54%) as compared with those with critical illness but admitted to the medical ward (19% [P = .001] and 38% [P = .002], respectively; Supplementary Table 7).
Table 4.
Clinical Management and Outcomes of Patients With Sepsis by Admission Ward
| Median (IQR) or No. (%) | ||||
|---|---|---|---|---|
| Characteristic | All Sepsis (n = 618) | ICU (n = 174) | Ward (n = 444) | P Value |
| Severity of illness at enrollment | ||||
| SOFA | 5 (3–9) | 10 (7–13) | 4 (3–6) | <.001 |
| APACHE II | … | 22 (17–25) | … | … |
| Acute kidney injury a | 239 (39) | 94 (53) | 145 (33) | <.001 |
| Critical care at enrollment | ||||
| Respiratory failure b | 306 (50) | 159 (90) | 147 (33) | <.001 |
| Mechanical ventilation | 277 (45) | 156 (89) | 121 (27) | <.001 |
| Vasoactive medications | 198 (32) | 106 (60) | 92 (21) | <.001 |
| New kidney replacement therapy | 15 (3) | 13 (7) | 2 (1) | <.001 |
| Admission management | ||||
| Blood cultures sent | 596 (96) | 172 (99) | 424 (96) | .05 |
| Antibiotics received | 604 (98) | 172 (99) | 432 (97) | .37 |
| Procedural drainage/debridement | 56 (9) | 16 (9) | 40 (9) | .94 |
| Lactate concentration measured | 264 (43) | 96 (55) | 168 (38) | <.001 |
| Death | ||||
| 7 d | 75 (12) | 38 (22) | 37 (8) | <.001 |
| Days to death | 2 (1–5) | 2 (1–5) | 3 (1–5) | .48 |
| 28 d | 125 (20) | 60 (34) | 65 (15) | <.001 |
| Days to death | 6 (2–12) | 5 (1–12) | 6 (2–13) | .34 |
| Lost to 28-d follow-up | 4 (1) | 0 | 4 (1) | .58 |
Abbreviations: ICU, intensive care unit; KDIGO, Kidney Disease: Improving Global Outcomes; modified SOFA, Sequential Organ Failure Assessment.
aAcute kidney injury based on KDIGO definition of 150% of estimated baseline creatinine in patients without chronic kidney disease or increase in creatinine by 0.3 mg/dL within 48 hours in any patient not requiring dialysis at baseline.
bRespiratory failure defined as requiring endotracheal intubation, mechanical ventilation, or noninvasive positive pressure ventilation.
Risk Factors of 28-Day Mortality in Sepsis
Finally, we investigated whether specific risk factors present at enrollment were associated with 28-day mortality in patients with sepsis (Supplementary Tables 8 and 9). Using 44 routinely obtained enrollment variables, such as patient characteristics, vital signs, and laboratory data (Supplementary Table 10), we performed a LASSO regression to identify a parsimonious set of variables associated with 28-day sepsis mortality. Six variables were identified, and all were associated with 28-day sepsis mortality in unadjusted models (Table 5). In a multivariable model including enrollment site, 5 risk factors were associated with an increased risk of 28-day sepsis mortality (all P < .01; Table 5): a preexisting HIV diagnosis, higher heart rate, lower hemoglobin, respiratory failure, and AKI (adjusted odds ratio, 2.07; 95% CI, 1.30–3.29; P = .002). As 39% (239/618) of patients with sepsis had AKI at enrollment, we next looked at the outcomes of sepsis cases with and without AKI. In patients with sepsis-associated AKI (SA-AKI), 15% (35/239) required kidney replacement therapy (KRT) during their hospitalization, and 30% (71/239) died within 28 days as compared with 14% (54/375) without AKI (P < .001). Finally, patients with sepsis with AKI at enrollment had significantly worse survival over 28-days when compared those without AKI (P < .001; Figure 1).
Table 5.
Risk Factors at Presentation for 28-Day Mortality in Patients With Sepsis (n = 614)
| Unadjusted | Adjusted a | |||
|---|---|---|---|---|
| Variable b | OR (95% CI) | P Value | OR (95% CI) | P Value |
| Preexisting conditions | ||||
| HIV | 4.06 (1.28–12.8) | .02 | 5.28 (1.82–15.3) | .002 |
| Presenting clinical/laboratory data | ||||
| Heart rate | 1.03 (1.02–1.04) | <.001 | 1.03 (1.02–1.04) | <.001 |
| Blood urea nitrogen | 3.87 (1.99–7.51) | <.001 | 2.21 (.98–4.97) | .06 |
| Hemoglobin | 0.90 (.83–.97) | <.001 | 0.83 (.76–.92) | <.001 |
| Presenting characteristics | ||||
| Acute kidney injury | 2.51 (1.68–3.75) | <.001 | 2.07 (1.30–3.29) | .002 |
| Respiratory failure | 3.64 (2.35–5.64) | <.001 | 2.34 (1.42–3.86) | .001 |
Abbreviations: LASSO, least absolute shrinkage and selection operator; OR, odds ratio.
aAll listed variables were included in the multivariable model, which adjusted for enrollment site (adjusted OR, 1.49; 95% CI, .94–2.37; P = .09).
bVariables included in unadjusted models and the multivariable model were selected through LASSO regression for 44 variables of patient characteristics. The sample (n = 614) does not include 4 patients lost to follow-up at 28 days after enrollment.
Figure 1.
Survival curve of sepsis cases with and without AKI. Kaplan-Meier survival curves over 28 days demonstrate worse survival in patients with sepsis-associated AKI (gray) at enrollment vs those without AKI (black dashed; P < .001, log-rank test). AKI, acute kidney injury.
DISCUSSION
This observational prospective cohort study is one of the largest multicenter studies of patients hospitalized with suspected community-acquired infection in a resource-limited setting, including Southeast Asia. We report that sepsis is common in this cohort, is caused by different pathogens as compared with high-income country (HIC) settings, and is associated with significant mortality. Furthermore, we report that critical illness is not only common but also frequently managed outside a traditional ICU. Finally, we identify SA-AKI as a significant risk factor for sepsis mortality.
Limited prospective data exist regarding the frequency of sepsis among patients hospitalized for community-acquired infection in resource-limited settings. Previously reported large prospective studies in Southeast and South Asia enrolled patients with suspected or confirmed sepsis, often only in ICUs, limiting broader application in the region [2, 5, 8, 17]. In HIC settings, 25% to 44% of patients presenting to emergency departments with community-acquired infection typically meet criteria for sepsis [18, 19]. However, in our study, 66% of patients hospitalized with community-acquired infection met sepsis criteria. Notably, patients in this study frequently received mechanical ventilation and vasoactive medications outside the ICU. These findings confirm the frequency of sepsis in this region but also highlight the challenges of applying traditional definitions of sepsis to locations with resource limitations [20, 21]. Additionally, our findings suggest that prospective enrollment of patients with sepsis in an ICU may miss patients meeting sepsis criteria and those with critical illness in similar settings. Whether patients outside the ICU requiring mechanical ventilation or vasoactive medications may benefit from an early escalation of care is unknown, although only 4% of those initially admitted to the ward with sepsis were eventually transferred to the ICU, despite a 15% mortality rate. In the United States, admission to the ward rather than the ICU may be associated with improved outcomes among patients with sepsis not requiring life support [22]. However, the 2021 SSC guidelines recommend ICU admission within 6 hours for patients requiring critical care, and early ICU transfer is associated with improved mortality in sepsis [16, 23]. In resource-limited settings where ICU transfer may not be possible for all patients who are critically ill, further study is necessary to determine the optimal management strategy and outcome effect of critical care outside the ICU.
According to the 2021 SSC guidelines, early empiric broad-spectrum antibiotics and blood cultures are recommended for patients with community-acquired sepsis [16]. Sepsis bundles have been widely implemented in Thailand, although usage varies countrywide [24]. Reflecting this, nearly all patients in our cohort received empiric antibiotics and had blood cultures sent at admission. Conversely, lactate measurement was less common, occurring in less than half of patients with sepsis. Lactate levels are associated with sepsis mortality, including that in northeast Thailand, and its measurement is recommended in the SSC hour 1 sepsis bundle [16, 25]. A recent multinational study of patients with sepsis admitted to tertiary ICUs throughout Asia reported poor compliance with 1- and 3-hour sepsis bundles [17]. However, the utility of lactate levels to guide resuscitation in patients with sepsis is unknown in LMICs and remains a topic of debate in HICs [26, 27]. Therefore, further study is necessary to determine whether broader adoption of SSC bundles is cost-efficient and improves outcomes in similar settings.
The adoption of broad, early empiric antibiotics in patients with suspected infection may contribute to antibiotic resistance, a growing concern in LMICs, including Thailand [28, 29]. Indeed, in northeast Thailand, antimicrobial-resistant infections, such as Escherichia coli, Klebsiella species, and Acinetobacter species, are associated with excess mortality in hospitalized patients [30]. Given this concern, the necessity of broad empiric gram-negative antibiotics in hospitalized patients in HICs has become a matter of debate [31]. In our cohort, nearly all patients, regardless of illness severity or sepsis status, received broad gram-negative coverage at admission, most commonly with ceftazidime, a third-generation cephalosporin with antipseudomonal activity, similar to prior reports in the region [32]. This approach may reflect the high frequency of gram-negative bacteremia in our cohort, a trend noted across Southeast and South Asia but distinct from many HICs [2, 5, 8, 33]. Empiric antibiotics were not discontinued quickly, as the median antibiotic duration was 6 days in patients with sepsis and 4 days in those without sepsis. Therefore, it is possible that in some patients, particularly those without sepsis and negative cultures, empiric antibiotics could have been avoided or coverage de-escalated quickly. As sepsis management in tropical settings continues to evolve, the evolution of antimicrobial stewardship programs, including biomarker-based or other de-escalation protocols, may be a reasonable approach to help address the potential for future antimicrobial resistance [7, 34].
Reflecting the increasing prevalence of diabetes mellitus in Asia, 42% of enrolled patients in this cohort had a preexisting diagnosis of diabetes [35]. The high prevalence of diabetes among patients with sepsis has been described in similar large sepsis studies in the region [2, 5, 8, 17]. In Thailand, diabetes associated with poor glycemic control is particularly prevalent in the northeast region of the country [36]. Additionally, diabetes is a major risk factor for melioidosis, perhaps reflecting the high rates of that disease in our cohort [37]. As diabetes prevalence increases in lower-resourced tropical settings, particular attention needs to be paid to how this demographic change may affect infection etiologies and management.
An intriguing finding in our study relates to the frequency and importance of SA-AKI. AKI at enrollment was common in patients with sepsis and independently associated with 28-day mortality. The burden of AKI is particularly high in LMICs, where an estimated 85% of global cases occur [38]. However, access to KRT is often limited in these settings, and continuous KRT is frequently restricted to urban quaternary facilities [39]. Only limited intermittent hemodialysis was available at the study hospitals in our report, although 15% of patients with SA-AKI at enrollment received this intervention during their hospitalization. The incidence of KRT in SA-AKI has frequently been reported as >20%, so it is possible that KRT usage would have been higher in our cohort with greater access [39, 40]. Notably, some tropical infections, including those common in rural Southeast Asia, such as leptospirosis and melioidosis, are associated with particularly increased risks of AKI [41, 42]. While it is unclear whether SA-AKI is preventable, optimized management may reduce the risk of disease progression [43]. Additionally, recent reports have identified subphenotypes of HIC AKI cases with potentially unique responses to therapeutic interventions [44]. Critically, few studies have assessed the epidemiology of SA-AKI in resource-constrained settings globally, including Southeast Asia [45]. Given the prevalence and importance of this sepsis-associated disease, further contextualized consideration of SA-AKI in global settings is urgently needed.
This study has several strengths. To our knowledge, this is the largest multicenter prospective cohort study of patients hospitalized with community-acquired infection in a resource-limited setting. Our study included all-cause 28-day mortality, important given the local culture practice of discharging moribund patients to die at home, with minimal loss to follow-up. Study data were collected by a trained study team, yielding minimal missing data. Additionally, strict, broadly accepted definitions were used to define patients with sepsis and AKI.
Our study also has several limitations. Highlighting the resource constraints of the study hospitals, arterial blood gases were not routinely obtained, requiring modification of SOFA. Additionally, the timing of specific interventions, including antibiotics, was not available on a minute or hour level of specificity, rather the calendar day. Identification of infectious etiologies was limited to the testing performed at the study hospitals and did not include prospective specific pathogen testing. Antibiotic susceptibility data were not available either. Both study hospitals were referral hospitals within their rural provinces; therefore, care delivery may have differed among hospitals and may have been affected by the prospective study design. Additionally, patient demographics and management may differ as compared with local health care centers or larger academic quaternary care centers within Thailand and other global settings. The study was also conducted over a 16-month period and may not completely reflect the seasonal variations of some infections; moreover, it did not enroll patients infected with SAR-CoV-2.
In conclusion, sepsis and sepsis-associated critical illness are common and associated with mortality in adults with community-acquired infection in rural Southeast Asia. SA-AKI on presentation was associated with an increased risk of mortality. Given the paucity of high-quality existing data on sepsis in resource-constrained settings, this study provides critical and timely data to inform future clinical and research approaches in global settings.
Supplementary Material
Notes
Acknowledgments. The authors thank the patients and staff at Mukdahan Hospital and Roi Et Hospital.
Author contributions. R. P. and S. W. W. participated in project design and analysis and wrote the first draft of the manuscript. T. D. C. also analyzed the data. P. P., A. D., S. S., and B. M. oversaw data collection and management and participated in project design. N. S., S. C., and R. J. participated in project design, directed recruitment, and facilitated data collection. T. E. W., N. C., and S. W. W. developed the project and oversaw the study. All authors contributed to the writing or revision of the manuscript.
Availability of supporting data. The dataset supporting these findings is available in a repository (10.6084/m9.figshare.29886080).
Financial support. This work was supported by the US National Institutes of Health (grants F32HL168809 to T. D. C., K08HL157562 to S. W. W., and R01AI137111 to T. E. W.); and a Firland Foundation grant to S. W. W.
Contributor Information
Rungnapa Phunpang, Department of Microbiology and Immunology, Faculty of Tropical Medicine, Mahidol University, Bangkok, Thailand.
Prapassorn Poolchanuan, Department of Microbiology and Immunology, Faculty of Tropical Medicine, Mahidol University, Bangkok, Thailand.
Taylor D Coston, Division of Pulmonary, Critical Care and Sleep Medicine, Department of Medicine, University of Washington, Seattle, Washington, USA.
Adul Dulsuk, Department of Microbiology and Immunology, Faculty of Tropical Medicine, Mahidol University, Bangkok, Thailand.
Sopha Saeyang, Department of Microbiology and Immunology, Faculty of Tropical Medicine, Mahidol University, Bangkok, Thailand.
Boonthanom Moonmueangsan, Department of Microbiology and Immunology, Faculty of Tropical Medicine, Mahidol University, Bangkok, Thailand.
Narongchai Sangsa, Department of Medicine, Roi Et Hospital, Roi Et, Thailand.
Sermchart Chinnakarnsawas, Department of Medicine, Roi Et Hospital, Roi Et, Thailand.
Rachan Janon, Department of Medicine, Mukdahan Hospital, Mukdahan, Thailand.
T Eoin West, Division of Pulmonary, Critical Care and Sleep Medicine, Department of Medicine, University of Washington, Seattle, Washington, USA; Department of Global Health, University of Washington, Seattle, Washington, USA.
Narisara Chantratita, Department of Microbiology and Immunology, Faculty of Tropical Medicine, Mahidol University, Bangkok, Thailand; Mahidol-Oxford Tropical Medicine Research Unit, Faculty of Tropical Medicine, Mahidol University, Bangkok, Thailand.
Shelton W Wright, Division of Pediatric Critical Care Medicine, Department of Pediatrics, University of Washington, Seattle, Washington, USA.
Supplementary Data
Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
References
- 1. Rudd KE, Seymour CW, Aluisio AR, et al. Association of the quick Sequential (Sepsis-Related) Organ Failure Assessment (qSOFA) score with excess hospital mortality in adults with suspected infection in low- and middle-income countries. JAMA 2018; 319:2202–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Sudarmono P, Aman AT, Arif M, et al. Causes and outcomes of sepsis in Southeast Asia: a multinational multicentre cross-sectional study. Lancet Glob Health 2017; 5:e157–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Cummings MJ, Jacob ST. Equitable endotyping is essential to achieve a global standard of precise, effective, and locally-relevant sepsis care. EBioMedicine 2022:86:104348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Singer M, Deutschman CS, Seymour C, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA 2016; 315:801–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Virk HS, Biemond JJ, Earny VA, et al. Unraveling sepsis epidemiology in a low- and middle-income intensive care setting reveals the alarming burden of tropical infections and antimicrobial resistance: a prospective observational study (MARS-India). Clin Infect Dis 2025; 80:101–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Rudd KE, Tutaryebwa LK, West TE. Presentation, management, and outcomes of sepsis in adults and children admitted to a rural Ugandan hospital: a prospective observational cohort study. PLoS One 2017; 12:1–13. [Google Scholar]
- 7. Thwaites L, Nasa P, Abbenbroek B, et al. Management of adult sepsis in resource-limited settings: global expert consensus statements using a Delphi method. Intensive Care Med 2025; 51:21–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Hantrakun V, Somayaji R, Teparrukkul P, et al. Clinical epidemiology and outcomes of community acquired infection and sepsis among hospitalized patients in a resource limited setting in Northeast Thailand: a prospective observational study (Ubon-sepsis). PLoS One 2018; 13:1–14. [Google Scholar]
- 9. Wick KD, Matthay MA, Ware LB. Pulse oximetry for the diagnosis and management of acute respiratory distress syndrome. Lancet Respir Med 2022; 10:1086–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Schenck EJ, Hoffman KL, Oromendia C, et al. A comparative analysis of the respiratory subscore of the Sequential Organ Failure Assessment scoring system. Ann Am Thorac Soc 2021; 18:1849–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Catoire P, Tellier E, de la Rivière C, et al. Assessment of the SpO2/FiO2 ratio as a tool for hypoxemia screening in the emergency department. Am J Emerg Med 2021; 44:116–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Riviello ED, Kiviri W, Twagirumugabe T, et al. Hospital incidence and outcomes of the acute respiratory distress syndrome using the Kigali modification of the Berlin definition. Am J Respir Crit Care Med 2016; 193:52–9. [DOI] [PubMed] [Google Scholar]
- 13. Seymour CW, Liu VX, Iwashyna TJ, et al. Assessment of clinical criteria for sepsis for the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA 2016; 315:762–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Kellum JA, Lameire N, Aspelin P, et al. et al. KDIGO clinical practice guideline for acute kidney injury. Kidney Int Suppl 2012; 2:1–138. [Google Scholar]
- 15. Tibshirani R. Regression shrinkage and selection via the lasso. J R Statist Soc B 1996; 58:267–88. [Google Scholar]
- 16. Evans L, Rhodes A, Alhazzani W, et al. Surviving Sepsis Campaign: international guidelines for management of sepsis and septic shock 2021. Intensive Care Med 2021; 47:1181–247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Li A, Ling L, Qin H, et al. Epidemiology, management, and outcomes of sepsis in ICUs among countries of differing national wealth across Asia. Am J Respir Crit Care Med 2022; 206:1107–16. [DOI] [PubMed] [Google Scholar]
- 18. Oberlin M, Balen F, Bertrand L, et al. Sepsis prevalence among patients with suspected infection in emergency department: a multicenter prospective cohort study. Eur J Emerg Med 2020; 27:373–8. [DOI] [PubMed] [Google Scholar]
- 19. Freund Y, Lemachatti N, Krastinova E, et al. Prognostic accuracy of Sepsis-3 criteria for in-hospital mortality among patients with suspected infection presenting to the emergency department. JAMA 2017; 317:301–8. [DOI] [PubMed] [Google Scholar]
- 20. Rudd KE, Kissoon N, Limmathurotsakul D, et al. The global burden of sepsis: barriers and potential solutions. Crit Care 2018; 22:1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Rudd KE, Johnson SC, Agesa KM, et al. Global, regional, and national sepsis incidence and mortality, 1990–2017: analysis for the Global Burden of Disease Study. Lancet 2020; 6736:1–12. [Google Scholar]
- 22. Anesi GL, Liu VX, Chowdhury M, et al. Association of ICU admission and outcomes in sepsis and acute respiratory failure. Am J Respir Crit Care Med 2022; 205:520–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Shibata J, Osawa I, Fukuchi K, Goto T. The association between time from emergency department visit to ICU admission and mortality in patients with sepsis. Crit Care Explor 2023; 5:E0915. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Booraphun S, Hantrakun V, Siriboon S, et al. Effectiveness of a sepsis programme in a resource-limited setting: a retrospective analysis of data of a prospective observational study (Ubon-sepsis). BMJ Open 2021:11:e041022. [Google Scholar]
- 25. Wright SW, Hantrakun V, Rudd KE, et al. Enhanced bedside mortality prediction combining point-of-care lactate and the quick Sequential Organ Failure Assessment (qSOFA) score in patients hospitalised with suspected infection in Southeast Asia: a cohort study. Lancet Glob Health 2022; 10:e1281–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Chen H, Zhao C, Wei Y, Jin J. Early lactate measurement is associated with better outcomes in septic patients with an elevated serum lactate level. Crit Care 2019:23:351. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Zampieri FG, Damiani LP, Bakker J, et al. Effects of a resuscitation strategy targeting peripheral perfusion status versus serum lactate levels among patients with septic shock: a bayesian reanalysis of the ANDROMEDA-SHOCK trial. Am J Respir Crit Care Med 2020; 201:423–9. [DOI] [PubMed] [Google Scholar]
- 28. Tuamsuwan K, Chamawan P, Boonyarit P, et al. Frequency of antimicrobial-resistant bloodstream infections in 111 hospitals in Thailand, 2022. J Infect 2024; 89:106249. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. World Health Organization . Global action plan on antimicrobial resistance. 2015. Available at: https://www.who.int/publications/i/item/9789241509763
- 30. Lim C, Teparrukkul P, Nuntalohit S, et al. Excess mortality attributable to hospital-acquired antimicrobial-resistant infections: a 2-year prospective surveillance study in northeast Thailand. Open Forum Infect Dis 2022; 9:ofac305. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Baghdadi JD, Goodman KE, Magder LS, Claeys KC, Sutherland ME, Harris AD. Association between delayed broad-spectrum gram-negative antibiotics and clinical outcomes: how much does getting it right with empiric antibiotics matter? Clin Infect Dis 2025; 80:949–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Rudd KE, Hantrakun V, Somayaji R, et al. Early management of sepsis in medical patients in rural Thailand: a single-center prospective observational study. J Intensive Care 2019; 7:1–8. [Google Scholar]
- 33. Todi S, Mehta Y, Zirpe K, et al. A multicentre prospective registry of one thousand sepsis patients admitted in Indian ICUs: (SEPSIS INDIA) study. Crit Care 2024; 28:375. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Schuetz P, Wirz Y, Sager R, et al. Effect of procalcitonin-guided antibiotic treatment on mortality in acute respiratory infections: a patient level meta-analysis. Lancet Infect Dis 2018; 18:95–107. [DOI] [PubMed] [Google Scholar]
- 35. Nanditha A, Ma RCW, Ramachandran A, et al. Diabetes in Asia and the Pacific: implications for the global epidemic. Diabetes Care 2016; 39:472–85. [DOI] [PubMed] [Google Scholar]
- 36. Sakboonyarat B, Pima W, Chokbumrungsuk C, et al. National trends in the prevalence of glycemic control among patients with type 2 diabetes receiving continuous care in Thailand from 2011 to 2018. Sci Rep 2021:11:14260. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Chantratita N, Phunpang R, Yarasai A, et al. Characteristics and one year outcomes of melioidosis patients in northeastern Thailand: a prospective, multicenter cohort study. Lancet Reg Health Southeast Asia 2023; 9:100118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Batte A, Shahrin L, Claure-Del Granado R, Luyckx VA, Conroy AL. Infections and acute kidney injury: a global perspective. Semin Nephrol 2023; 43:151466. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Cullis B, Calice da Silva V, McCulloch M, Ulasi I, Wijewickrama E, Iyengar A. Access to dialysis for acute kidney injury in low-resource settings. Semin Nephrol 2022; 42:151313. [DOI] [PubMed] [Google Scholar]
- 40. Chiang HY, Liang CC, Hsiao YL, et al. Sepsis-associated acute kidney disease incidence, trajectory, and outcomes. Kidney Med 2025; 7:100959. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Sethi A, Kumar TP, Vinod KS, et al. Kidney involvement in leptospirosis: a systematic review and meta-analysis. Infection 2025; 53:785–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Prabhu RA, Shaw T, Rao IR, et al. Acute kidney injury and its outcomes in melioidosis. J Nephrol 2021; 34:1941–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Zarbock A, Nadim MK, Pickkers P, et al. Sepsis-associated acute kidney injury: consensus report of the 28th Acute Disease Quality Initiative Workgroup. Nat Rev Nephrol 2023; 19:401–17. [DOI] [PubMed] [Google Scholar]
- 44. Legrand M, Bagshaw SM, Bhatraju PK, et al. Sepsis-associated acute kidney injury: recent advances in enrichment strategies, sub-phenotyping and clinical trials. Crit Care 2024:28:92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Liu J, Xie H, Ye Z, Li F, Wang L. Rates, predictors, and mortality of sepsis-associated acute kidney injury: a systematic review and meta-analysis. BMC Nephrol 2020:21:318. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.