Among febrile pediatric inpatients in Tanzania, where malaria transmission was intense, invasive nontyphoidal Salmonella was common and Salmonella Typhi was uncommon, whereas the inverse was observed at a site with low malaria transmission.
Keywords: typhoid fever, Salmonella, sub-Saharan Africa, malaria
Abstract
Background. The epidemiology of Salmonella Typhi and invasive nontyphoidal Salmonella (NTS) differs, and prevalence of these pathogens among children in sub-Saharan Africa may vary in relation to malaria transmission intensity.
Methods. We compared the prevalence of bacteremia among febrile pediatric inpatients aged 2 months to 13 years recruited at sites of high and low malaria endemicity in Tanzania. Enrollment at Teule Hospital, the high malaria transmission site, was from June 2006 through May 2007, and at Kilimanjaro Christian Medical Centre (KCMC), the low malaria transmission site, from September 2007 through August 2008. Automated blood culture, malaria microscopy with Giemsa-stained blood films, and human immunodeficiency virus testing were performed.
Results. At Teule, 3639 children were enrolled compared to 467 at KCMC. Smear-positive malaria was detected in 2195 of 3639 (60.3%) children at Teule and 11 of 460 (2.4%) at KCMC (P < .001). Bacteremia was present in 336 of 3639 (9.2%) children at Teule and 20 of 463 (4.3%) at KCMC (P < .001). NTS was isolated in 162 of 3639 (4.5%) children at Teule and 1 of 463 (0.2%) at KCMC (P < .001). Salmonella Typhi was isolated from 11 (0.3%) children at Teule and 6 (1.3%) at KCMC (P = .008). With NTS excluded, the prevalence of bacteremia at Teule was 5.0% and at KCMC 4.1% (P = .391).
Conclusions. Where malaria transmission was intense, invasive NTS was common and Salmonella Typhi was uncommon, whereas the inverse was observed at a low malaria transmission site. The relationship between these pathogens, the environment, and the host is a compelling area for further research.
(See the Editorial Commentary by MacLennan on pages 648–50.)
Salmonella enterica is a major cause of community-acquired bloodstream infection and death among children across sub-Saharan Africa [1, 2]. However, the proportion of disease caused by infection with Salmonella enterica serovars Typhi and Paratyphi (typhoidal Salmonella) vs nontyphoidal S. enterica serovars (NTS), shows considerable temporal and geographic variation. The epidemiology of typhoidal Salmonella appears to be driven by environmental risk factors including overcrowding, poor sanitation, and unsafe food and water [3]. By contrast, hospital-based studies of invasive NTS (iNTS) among African children and adults implicate host-associated risk factors including human immunodeficiency virus (HIV)–related immunosuppression, malaria, anemia, and malnutrition [4].
The association of malaria with bacteremia, particularly NTS, has been documented at multiple locations in Africa [5–7]. Malaria epidemiology in Africa is heterogenous, and in northeast Tanzania malaria transmission rates vary geographically from intense and perennial in coastal areas, to low or absent in the highland areas of the Kilimanjaro Region [8]. In this study we compare the prevalence of malaria and bacteremia, focusing on NTS and Salmonella Typhi, among febrile hospitalized children at 2 sites of different malaria transmission in Tanzania.
METHODS
Study Sites
The study was conducted in 2 locations 258 km apart in northern Tanzania: Moshi in the Kilimanjaro Region and Muheza in the Tanga Region (Figure 1). Moshi is situated at 890 m above sea level, and Muheza is 96 m above sea level. The climate in both locations is characterized by short and long rainy periods. In Moshi, Plasmodium falciparum malaria transmission is low and seasonal [9], whereas in Muheza malaria transmission at the time of the study was intense and perennial, with 2 seasonal peaks corresponding with the end of the rainy seasons [8]. In Moshi, the study was conducted at Kilimanjaro Christian Medical Centre (KCMC), a 458-bed referral hospital serving several regions in northern Tanzania. In Muheza, the study was conducted at Teule Hospital (Teule), a 330-bed district hospital serving the Tanga Region of northeastern Tanzania.
Figure 1.
Map of study sites at Teule Hospital, Muheza, and Kilimanjaro Christian Medical Centre, Moshi, Tanzania.
Participants
At both study sites, as described in detail elsewhere [10, 11], pediatric inpatients aged 2 months to 13 years admitted within the past 24 hours were eligible for enrollment if they had a history of fever in the past 48 hours, an axillary temperature ≥37.5°C, or a rectal temperature of ≥38.0°C. Patients were excluded if they had known malignancy, renal or hepatic failure, bone marrow aplasia, trauma, or surgery. Participants were prospectively enrolled 5 days a week; study enrollment occurred at Teule from 1 June 2006 through 31 May 2007 and at KCMC from 17 September 2007 through 31 August 2008.
Data Collection
Case report forms, including clinical history, physical examination, and demographics, were standardized between the study sites and completed by trained clinical officers who were study team members. Blood was drawn using aseptic technique for aerobic blood culture, blood parasite examination, and HIV testing. Additionally, point-of-care tests for hemoglobin, glucose, and serum lactate and a rapid diagnostic test for P. falciparum malaria were performed.
Laboratory Methods
Blood culture was performed using BacT/ALERT pediatric fan bottles that were loaded into the BacT/ALERT 3D Microbial Detection System (bioMérieux, Durham, North Carolina), where they were incubated for 5 days. Standard methods were used for identifying bloodstream isolates. Giemsa-stained blood films were examined for blood parasites by oil immersion microscopy. Slides were independently double-read and discrepancies resolved by an independent third reader. HIV antibody testing was performed using a previously described algorithm [10]. Point-of-care testing was performed for hemoglobin and glucose using Hemocue (Angelholm, Sweden), and for lactate using Lactate-Pro (Arkray, Kyoto, Japan). A histidine-rich protein 2 (HRP-2)–based rapid diagnostic test for P. falciparum was done using site-validated Paracheck (Orchid Biomedical, Mumbai, India).
Study Definitions
Malaria was defined as P. falciparum asexual parasitemia of any density on microscopy. Recent malaria was defined as negative for P. falciparum asexual parasites by microscopy with a positive HRP-2–based rapid diagnostic test. This definition is based on the persistence of detectable HRP-2 for ≥1 month after clearance of parasites [12, 13]. Parasite densities were calculated from the geometric mean of the 2 closest counts of asexual parasites per 200 white blood cells and the actual white blood count or, if missing, 8000 cells/µL. Severe anemia was defined as a hemoglobin level <5 g/dL. Hypoglycemia was defined as a glucose level of <45 mg/dL [14]. Severe acute malnutrition was defined for children aged 6 months to <60 months as mid-upper arm circumference <115 mm, weight-for-height z score less than −3, or presence of bilateral pitting pedal edema [14, 15]; for children ≥60 months, severe malnutrition was defined as body mass index-for-age z score less than −3 or presence of bilateral pitting pedal edema [16, 17]. Participant residence was classified as rural, urban, or mixed based on the 2002 Tanzania Population and Housing Census [18].
Statistical Analysis
Data were entered using the Cardiff Teleform system (Cardiff, Inc, Vista, California) into an Access database (Microsoft Corporation, Redmond, Washington). Descriptive statistics are presented as proportions, medians, ranges, and interquartile ranges. The Pearson χ2 test was used to compare categorical data; Fisher exact test was used when any cell contained <10 observations. Wilcoxon rank-sum was used to compare categorical and nonparametric continuous data. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated when appropriate. All P values are 2 sided and evaluated for statistical significance at the .05 significance level.
Backward multivariable logistic regression was performed to assess risk factors for NTS and Salmonella Typhi bacteremia. Populations from Teule and KCMC were combined for analysis. Variable selection for logistic regression was decided by the authors on the basis of theoretical and empirical findings. Collinearity testing was performed for selected variables. Likelihood ratios were used to compare iterations of the models. Presented P values are based on Wald tests.
World Health Organization (WHO) Anthro version 3.2.2 macros for Stata were used to calculate nutritional z scores using WHO Child Growth Standards [15] for children aged 6 months to <60 months and WHO Reference 2007 [16] for children aged ≥60 months. Participants with z score outliers (less than −5 or >5) were excluded from analysis. All analyses were performed using Stata software, version 12 (StataCorp, College Station, Texas).
Research Ethics
This study was approved by the KCMC Research Ethics Committee, the Tanzania National Institutes for Medical Research National Research Ethics Coordinating Committee, and the institutional review boards of Duke University Medical Center and the London School of Hygiene and Tropical Medicine.
RESULTS
At Teule, 4334 pediatric admissions were screened for enrollment over the study period, 3766 were eligible, and 3639 were enrolled and had sufficient data for study inclusion. At KCMC, 1154 pediatric inpatients were screened, 644 were eligible, and 467 were enrolled. Participant demographics for each site are displayed in Table 1. Median age and sex were not statistically different between the sites. At Teule, more participants reported rural residence, participants' mothers had completed fewer overall years of formal education, and fewer were referred from another inpatient facility compared to those at KCMC. Antibacterial use in the 48 hours before admission was reported by 450 of 3600 (12.5%) children at Teule compared to 210 of 398 (52.8%) at KCMC (P<.001). Antimalarial use before admission was reported by 1308 of 3166 (41.3%) participants at Teule compared to 143 of 408 (35.1%) at KCMC (P = .015).
Table 1.
Demographic Characteristics of Participants, Teule Hospital and Kilimanjaro Christian Medical Centre, Tanzania
| Characteristic | Teule (n = 3639), no./No. (%) | KCMC (n = 467), no./No. (%) | P Value |
|---|---|---|---|
| Age, y, median (range) | 1.57 (0.2–13.0) | 1.43 (0.2–13.5) | .319 |
| Female | 1669/3639 (45.9) | 200/467 (42.8) | .215 |
| Mother's highest level of education | |||
| None | 511/3582 (14.3) | 29/451 (6.4) | <.001 |
| Less than Standard 7 | 573/3582 (16.0) | 34/451 (7.5) | <.001 |
| Standard 7 | 2395/3582 (66.9) | 250/451 (55.4) | <.001 |
| Higher | 103/3582 (2.9) | 138/451 (30.6) | <.001 |
| Referred from another inpatient facility | 151/3635 (4.2) | 178/464 (38.4) | <.001 |
| Residencea | |||
| Rural | 2805/3633 (77.2) | 168/392 (42.9) | <.001 |
| Urban | 433/3633 (11.9) | 200/392 (51.0) | <.001 |
| Mixed urban/rural | 395/3633 (10.9) | 24/392 (6.1) | .003 |
Abbreviation: KCMC, Kilimanjaro Christian Medical Centre.
a Residence was classified as rural, urban, or mixed based on the 2002 Tanzania Population and Housing Census [18].
Clinical Presentation
Clinical and laboratory features of participants by site are shown in Table 2. Compared with those at Teule, on admission, KCMC participants had a longer median duration of illness, and more frequently had lethargy, prostration, delayed capillary refill, prolonged skin pinch, and acute malnutrition. Compared with those at KCMC, on admission Teule participants were significantly more likely to have severe pallor, splenomegaly, hypoglycemia, and severe anemia.
Table 2.
Clinical and Laboratory Characteristics of Participants on Admission, Teule Hospital and Kilimanjaro Christian Medical Centre, Tanzania
| Characteristic | Teule (n = 3639), no./No. (%) | KCMC (n = 467), no./No. (%) | OR (95% CI) | P Value |
|---|---|---|---|---|
| Reported history | ||||
| Antibacterial use in past 48 h | 450/3600 (12.5) | 210/398 (52.8) | 0.13 (.10–.16) | <.001 |
| Antimalarial (oral or parenteral) use in past 48 h | 1308/3166 (41.3) | 143/408 (35.1) | 1.30 (1.05–1.63) | .015 |
| Duration of illness, d, median (range) | 3 (0–92) | 4 (1–90) | NA | <.001 |
| Clinical findings | ||||
| Temperature, °C, median (IQR) | 37.9 (37.1–38.7) | 38.2 (37.8–39) | NA | <.001 |
| Prostrationa | 410/3625 (11.3) | 73/459 (15.9) | 0.67 (.51–.90) | .004 |
| Lethargy | 508/3637 (14.0) | 88/466 (18.9) | 0.70 (.54–.91) | .005 |
| Severe pallor | 812/3638 (22.3) | 33/466 (7.1) | 3.77 (2.62–5.59) | <.001 |
| Splenomegalyb | 703/3597 (19.5) | 6/463 (1.3) | 18.50 (8.38–50.88) | <.001 |
| Capillary refill >3 s | 41/3638 (1.1) | 16/467 (3.4) | 0.32 (.17–.62) | <.001 |
| Skin pinch >2 s | 43/3638 (1.2) | 29/465 (6.2) | 0.18 (.11–.30) | <.001 |
| Core to peripheral temperature difference | 146/3630 (4.0) | 25/449 (5.6) | 0.71 (.46–1.15) | .123 |
| Hypoxiac | 78/3574 (2.2) | 59/448 (13.2) | 0.15 (.10–.21) | <.001 |
| Blantyre coma score ≤2 | 116/3606 (3.2) | 9/461 (2.0) | 1.67 (.84–3.77) | .153 |
| Severe acute malnutritiond | 146/3193 (4.6) | 69/370 (18.7) | 0.21 (.15–.29) | <.001 |
| Aged 6 mo to <5 y | 123/2921 (4.2) | 54/297 (18.2) | 0.20 (.14–.29) | <.001 |
| Aged 5–13 y | 23/272 (8.5) | 15/73 (20.6) | 0.36 (.17–.079) | .003 |
| Laboratory findings | ||||
| Glucose, mg/dL, median (IQR) | 109.9 (88.3, 129.7) | 115.3 (100.9, 131.5) | NA | <.001 |
| Hypoglycemia, glucose <45 mg/dL | 113/3631 (3.1) | 3/463 (0.6) | 4.93 (1.63–24.33) | .001 |
| Lactate, mmol/L, median (IQR) | 2.4 (1.8–3.6) | 2.3 (1.8–3.2) | NA | .133 |
| Hyperlactatemia, lactate >5 mmol/L | 429/3248 (13.2) | 35/459 (7.6) | 1.84 (1.28–2.72) | .001 |
| Hemoglobin, g/dL, median (IQR) | 8.1 (5.9–9.8) | 9.6 (7.9–11.2) | NA | <.001 |
| Severe anemia, hemoglobin <5 g/dL | 595/3632 (16.4) | 20/459 (4.4) | 4.30 (2.72–7.17) | <.001 |
Abbreviations: CI, confidence interval; IQR, interquartile range; KCMC, Kilimanjaro Christian Medical Centre; NA, not applicable; OR, odds ratio.
a Prostration defined as inability to sit unsupported if >9 months or, if age ≥9 months, inability to drink.
b Splenomegaly defined as palpable spleen >2 cm below the costal margin.
c Hypoxia defined as oxygen saturation <90% as measured by pulse oximetry.
d Participants aged <6 months and those with z score outliers excluded from analysis.
Malaria, Bacteremia, and HIV Infection
Smear-positive P. falciparum malaria was detected in 2195 of 3639 (60.3%) children enrolled at Teule compared to 11 of 460 (2.4%) at KCMC (OR, 62.05; 95% CI, 34.16–125.56); evidence of recent malaria was found in 501 of 3639 (13.8%) children at Teule compared to 3 of 458 (0.6%) at KCMC (OR, 24.21; 95% CI, 8.18–118.26) (Table 3). Bacteremia was present in 336 of 3639 (9.2%) children at Teule compared to 20 of 463 (4.3%) children at KCMC (OR, 2.25; 95% CI, 1.42–3.77). There was no difference in bacteremia prevalence between patients who reported or denied prior antibacterial use (OR, 0.95; 95% CI, .69–1.29).
Table 3.
Malaria, HIV Infection, Bacteremia, and Associated Mortality, Teule Hospital and Kilimanjaro Christian Medical Centre, Tanzania
| Infection | Teule (n = 3639), no./No. (%) | KCMC (n = 467), no./No. (%) | OR (95% CI) | P Value |
|---|---|---|---|---|
| Malaria | 2195/3639 (60.3) | 11/460 (2.4) | 62.05 (34.16–125.56) | <.001 |
| Recent malaria | 501/3639 (13.8) | 3/458 (0.6) | 24.21 (8.18–118.26) | <.001 |
| HIV | 142/3639 (3.9) | 60/445 (13.2) | 0.27 (.19–.37) | <.001 |
| Bacteremia | 336/3639 (9.2) | 20/463 (4.3) | 2.25 (1.42–3.77) | <.001 |
| Nontyphoidal Salmonella | 162/3639 (4.5) | 1/463 (0.2) | 21.53 (3.79–857.09) | <.001 |
| NTS associated with malaria | 53/162 (37.7) | 0/1 (0) | NA | NA |
| NTS associated with recent malaria | 67/162 (41.4) | 0/1 (0) | NA | NA |
| NTS associated with HIV | 8/162 (4.9) | 1/1 (100.0) | NA | NA |
| Bacteremia if NTS excludeda | 175/3478 (5.0) | 19/462 (4.1) | 1.24 (.76–2.12) | .391 |
| Bacteremia if NTS associated with malaria or recent malaria excluded | 216/3519 (6.1) | 20/463 (4.3) | 1.45 (.90–2.44) | .119 |
| Salmonella Typhi | 11/3639 (0.3) | 6/463 (1.3) | 0.23 (.08–.76) | .008 |
| Escherichia coli | 23/3639 (0.6) | 3/463 (0.7) | 0.98 (.29–5.09) | 1.000 |
| Streptococcus pneumoniae | 54/3639 (1.5) | 6/463 (1.3) | 1.15 (.49–3.28) | 1.000 |
| Blood culture contaminantsb | 252/3639 (6.9) | 27/463 (5.8) | 1.20 (.79–1.88) | .379 |
Abbreviations: CI, confidence interval; HIV, human immunodeficiency virus; KCMC, Kilimanjaro Christian Medical Centre; NTS, nontyphoidal Salmonella; NA, not applicable; OR, odds ratio.
a One of 162 patients from Teule with NTS also had a second pathogenic organism isolated.
b Any of the following isolates in the absence of a concurrent pathogenic organism: coagulase-negative Staphylococcus, Micrococcus, Corynebacterium, Bacillus species, viridans group Streptococcus, and non-Cryptococcus and non-Candida yeasts.
NTS was the isolated pathogen in 162 of 336 (48.2%) bacteremia cases at Teule, whereas at KCMC 1 of 20 (5.0%) bacteremia cases was due to NTS (OR, 17.69; 95% CI, 2.74–739.32) (Figure 2A). The overall prevalence of NTS bacteremia at Teule was 4.5% compared to 0.2% at KCMC (OR, 21.53; 95% CI, 3.79–857.09). When NTS was excluded from the bacteremia figures, bacteremia prevalence at Teule was 175 of 3478 (5.0%) and at KCMC 19 of 462 (4.1%; OR, 1.24; 95% CI, .76–2.12). Salmonella Typhi was isolated in 11 of 3639 (0.3%) patients at Teule compared to 6 of 463 (1.3%) at KCMC (OR, 0.23; 95% CI, .08–.76). The ratio of NTS to Salmonella Typhi was 162:11 (14.7:1) at Teule compared to 1:6 (0.2:1) at KCMC (Figure 2B).There was no significant difference in Streptococcus pneumoniae or Escherichia coli bacteremia prevalence between the sites (OR, 1.15; 95% CI, .49–3.28 and OR, 0.98; 95% CI, .29–5.09, respectively). HIV prevalence among children at Teule was 142 of 3639 (3.9%) compared to 60 of 445 (13.2%) at KCMC (OR, 0.27; 95% CI, .19–.37). Of NTS cases at Teule, 119 of 160 (74.4%) were associated with malaria or recent malaria and 8 of 160 (5.0%) were associated with HIV infection. The sole case of NTS at KCMC was associated with HIV infection.
Figure 2.
A, Bloodstream isolates, as proportion of total, among participants at Teule Hospital and Kilimanjaro Christian Medical Centre (KCMC). B, Nontyphoidal Salmonella and Salmonella Typhi bloodstream isolates, as proportion of total isolates, among participants at Teule Hospital and KCMC.
Risk Factors for NTS and Salmonella Typhi Bacteremia
Results of univariate and multivariate analysis of risk factors for NTS and Salmonella Typhi bacteremia are presented in Tables 4 and 5, respectively. In multivariate analysis, iNTS was associated with younger age (adjusted OR [AOR], 0.79; 95% CI, .70–.90; per each 1-year increase in age), recent malaria (AOR, 4.13; 95% CI, 2.66–6.44), severe anemia (AOR, 2.19; 95% CI, 1.48–3.23), and severe acute malnutrition (AOR, 2.01; 95% CI, 1.15–3.52). A negative association with current P. falciparum parasitemia that did not reach statistical significance was observed (AOR, 0.68; 95% CI, .43–1.08).
Table 4.
Univariate and Multivariate Analysis for Risk Factors Associated With Nontyphoidal Salmonella Bacteremia, Combined Data From Kilimanjaro Medical Centre and Teule Hospital, Tanzania
| Risk Factor | Univariate Analysis |
Multivariate Analysis |
||||
|---|---|---|---|---|---|---|
| NTS (n = 163), no./No. (%) | No NTS (n = 3939), no./No. (%) | OR (95% CI) | P Value | AOR (95% CI)a | P Value | |
| Age, y, median (range)b | 1.25 (0.3–9.0) | 1.57 (0.16–13.5) | 0.84 (.76–.94) | .001 | 0.79 (.70–.90) | <.001 |
| Rural ward residence | 133/146 (91.1) | 2836/3453 (82.1) | 2.23 (1.25–3.96) | .006 | NA | NA |
| Antibacterial preexposurec | 19/157 (12.1) | 639/3838 (16.7) | 0.69 (.42–1.12) | .134 | NA | NA |
| Antimalarial preexposure, oralc | 56/146 (39.0) | 1126/3430 (32.8) | 1.31 (.93–1.84) | .118 | NA | NA |
| Antimalarial preexposure, parenteralc | 19/147 (12.9) | 285/3413 (8.4) | 1.63 (.99–2.68) | .052 | NA | NA |
| Current malaria | 53 (32.5) | 2153/3933 (54.7) | 0.40 (.29–.56) | <.001 | 0.68 (.43–1.08) | .100 |
| Recent malaria | 67 (41.1) | 437/3930 (11.1) | 5.58 (4.02–7.74) | <.001 | 4.13 (2.66–6.44) | <.001 |
| Severe acute malnutrition | 17/149 (11.4) | 198/3410 (5.8) | 2.09 (1.24–3.53) | .005 | 2.01 (1.15–3.52) | .014 |
| HIV infection | 9 (5.5) | 193/3927 (4.9) | 1.13 (.57–2.25) | .711 | NA | NA |
| Severe anemia | 47 (28.8) | 568/3925 (14.5) | 2.39 (1.67–3.40) | <.001 | 2.19 (1.48–3.23) | <.001 |
Abbreviations: AOR, adjusted odds ratio; CI, confidence interval; HIV, human immunodeficiency virus; NA, not applicable; NTS, nontyphoidal Salmonella; OR, odds ratio.
a AOR and P values presented for multivariable analysis are the values of the most simplified/best fit model. AOR and P values for variables dropped from the final model due to nonsignificance are not presented.
b Modeled as continuous variable (OR represents the reduced odds of infection per year of age).
c Exposure to antibacterial or antimalarial in the 48 hours preceding hospital admission.
Table 5.
Univariate and Multivariate Analysis for Risk Factors Associated With Salmonella Typhi Bacteremia, Combined Data From Kilimanjaro Medical Centre and Teule Hospital, Tanzania
| Risk Factor | Univariate Analysis |
Multivariate Analysis |
||||
|---|---|---|---|---|---|---|
| Salmonella Typhi (n = 17), no./No. (%) | No Salmonella Typhi (n = 4085), no./No. (%) | OR (95% CI) | P Value | AOR (95% CI)a | P Value | |
| Age, y, median (range)b | 5.83 (0.3–13.5) | 1.56 (0.2–13.4) | 1.41 (1.26–1.58) | <.001 | 1.35 (1.21–1.51) | <.001 |
| Rural ward residence | 10/12 (83.3) | 2963/3592 (82.5) | 1.06 (.23–9.86) | .939 | NA | NA |
| Antibacterial preexposurec | 6/17 (35.3) | 652/3978 (16.4) | 2.78 (1.03–7.55) | .045 | NA | NA |
| Antimalarial preexposure, oralc | 9/16 (56.3) | 1174/3560 (33.0) | 2.61 (.97–7.03) | .057 | NA | NA |
| Antimalarial preexposure, parenteralc | 3/15 (20.0) | 301/3545 (8.5) | 2.69 (.76–9.60) | .126 | NA | NA |
| Current malaria | 2/17 (11.8) | 2204/4079 (54.0) | 0.11 (.03–.50) | .004 | 0.14 (.03–.64) | .011 |
| Recent malaria | 0/17 (0.0) | 504/4076 (12.4) | … | .254 | NA | NA |
| Severe acute malnutrition | 4/15 (26.7) | 211/3544 (5.6) | 5.74 (1.81–18.19) | .003 | NA | NA |
| HIV infection | 0/17 (0.0) | 202/4073 (5.0) | … | 1.000 | NA | NA |
| Severe anemia | 1/17 (5.9) | 614/4071 (15.1) | 0.35 (.05–2.66) | .311 | NA | NA |
Abbreviations: AOR, adjusted odds ratio; CI, confidence interval; HIV, human immunodeficiency virus; NA, not applicable; OR, odds ratio.
a AOR and P values presented for multivariable analysis are the values of the most simplified/best fit model. AOR and P values for variables dropped from the final model due to nonsignificance are not presented.
b Modeled as continuous variable (OR represents the increased odds of infection per year of age).
c Exposure to antibacterial or antimalarial in the 48 hours preceding hospital admission.
To further explore the relationship between current malaria and iNTS, malaria parasitemia at Teule was categorized by mean parasite density as low (1 to <5000 parasites/µL) or high (≥5000 to >50 000 parasites/µL). Low mean parasite density was present in 405 of 2195 (18.5%) of those with current malaria, whereas high mean parasite density was present in 1790 of 2195 (81.5%). Those with low mean parasite density had 3.53 times the odds of developing NTS bacteremia compared to those with high mean parasite density (95% CI, 1.93–6.36). Among the entire Teule study population, the odds of NTS bacteremia in those with low mean parasite density were 1.34 (95% CI, .81–2.13), whereas the odds of NTS bacteremia in those with high mean parasite density were 0.22 (95% CI, .14–.33).
Salmonella Typhi was associated, in multivariate analysis, with older age (AOR, 1.35; 95% CI, 1.21–1.51 per each 1-year increase in age) and negatively associated with current malaria (AOR, 0.14; 95% CI, .03–.64). No patients with Salmonella Typhi had HIV infection or recent malaria, so these variables were not included in the logistic regression model.
DISCUSSION
Among pediatric inpatients recruited under similar protocols during comparable enrollment periods in Tanzania, we observed at a site of high malaria endemicity that iNTS was a common cause of bacteremia and Salmonella Typhi was uncommon, whereas the inverse was true in an area of low malaria endemicity. A similar observation has been made in Kenya, where in a rural area holoendemic for malaria, there was a high ratio of iNTS to Salmonella Typhi infection among adults and children, whereas in an urban informal settlement where malaria is sporadic, Salmonella Typhi bacteremia predominated [19]. Other studies of bacteremia in malaria endemic areas of sub-Saharan Africa have noted that Salmonella Typhi may be rare or absent in settings where iNTS and malaria are prevalent [5, 6, 20, 21]. A recent study comparing bacteremia incidence over time at Teule showed that as malaria and iNTS incidence declined, incidence of Salmonella Typhi rose [22]. Additionally, recently presented data from Malawi showed that, during the past 2 years, following a decade-long decline in iNTS, a sustained outbreak of Salmonella Typhi has occurred [23].
The relationship between malaria and bacteremia has been well documented in sub-Saharan Africa [7, 24–26]. We found a higher overall prevalence of bacteremia among febrile hospitalized children at a site with high malaria transmission compared to a site with low malaria transmission. The difference in the prevalence of bacteremia between the 2 sites was driven largely by NTS. The prevalence of Streptococcus pneumoniae and Escherichia coli, the other most common bacterial isolates in children in sub-Saharan Africa [2], was no different between the sites.
The association between iNTS and malaria in sub-Saharan Africa was observed initially as a temporal association between NTS bacteremia and the rainy season in a malaria-endemic area [7]. Subsequently, declining NTS bacteremia following reduced malaria prevalence has been observed at a number of African sites [22, 25, 26]. Among African children, iNTS has been described as being associated with malaria parasitemia [6, 21, 27], recent malaria [5], anemia [5], and severe malarial anemia [28]. In multivariable analysis, we demonstrated significant associations between iNTS and severe anemia and between iNTS and recent malaria, but not current malaria. In univariate analysis, iNTS was positively but not significantly associated with mean malaria parasite density <5000 parasites/µL, but was significantly negatively associated with mean malaria parasite density 5000 to >50 000 parasites/µL. Other studies in Africa have shown that, although malaria is a risk factor for bacteremia, the risk is inversely related to malaria parasite density [7, 20, 29]. Additionally, although other studies have shown an association between current malaria and NTS bacteremia [6, 21, 27] a study of Kenyan children also found a negative association with current malaria, but a positive association with recent malaria [5].
Recent evidence suggests that malaria may predispose to iNTS via to changes in iron storage metabolism as a result of malarial hemolysis, causing neutrophil dysfunction and increased susceptibility [30, 31]. Additionally, mouse models have suggested that malaria parasite–specific factors, such parasite-induced reductions in levels of interleukin 12, may increase susceptibility to iNTS [32]. Other potential mechanisms for the increased risk of iNTS in malaria include translocation of gut microflora across a disordered gut barrier, caused by malarial sequestration of parasitized red blood cells in the intestine [33–35], and increased free iron from hemolysis promoting survival of Salmonella [4].
Other reported risk factors for iNTS in children in sub-Saharan Africa include malnutrition [6] and HIV infection [1, 36]. We found an independent association between iNTS and severe acute malnutrition, but no significant association with HIV infection. The association between iNTS and HIV infection has been more frequently reported among adults than children [36, 37], and has been linked to level of immunosuppression [38]. We were unable to assess iNTS risk in relation to level of HIV-related immunosuppression as CD4-positive T-lymphocyte counts were unavailable for more than half of HIV-infected participants at Teule. Additionally, all iNTS cases but one were from Teule, where HIV prevalence was relatively low, compared to some studies in which an association between iNTS and HIV infection has been described [5]. This could have affected our ability to detect an association between iNTS and HIV. Although a significantly greater proportion of participants at the low malaria-endemicity site had severe acute malnutrition or HIV infection, iNTS was rare at this site. This finding adds weight to the role of malaria and anemia in host susceptibility to iNTS infection, and raises the possibility that, in this particular setting in sub-Saharan Africa, these host risk factors may be among the most important contributors to iNTS infection.
Case numbers for Salmonella Typhi were small, but multivariate analysis showed that children of older age were at increased risk. Similar to iNTS, current malaria was negatively associated with Salmonella Typhi. However, in contrast to iNTS, no child with Salmonella Typhi had HIV infection or recent malaria, and only one had severe anemia. It is well described that iNTS and Salmonella Typhi have different risk factors and pathogenesis, so these findings are not surprising. Past studies, including a study of adult inpatients at KCMC, have noted that HIV-infected persons appear to have lower odds for clinical typhoid fever than HIV-uninfected persons [2, 39]. One proposed mechanism for this finding is a dampened cell-mediated immune response in patients with HIV infection affecting the clinical manifestations of Salmonella Typhi bacteremia [40]. We did not measure variables such as sanitation and hygiene, consumption of unsafe food and water, and crowding, that are commonly associated with typhoid fever. Although more participants reported urban residence at the site where Salmonella Typhi was more common, urban residence was not independently associated with infection with typhoidal Salmonella. Overall, although malaria endemicity likely accounts for differences in iNTS prevalence between the sites, our data cannot account for the differences in Salmonella Typhi prevalence. Other, unmeasured factors, including differences in environmental reservoirs and exposures, the occurrence of geographically localized epidemics, and potential interactions between the pathogens, may play a role in the differing epidemiology.
This study has a number of potential limitations. Although the study protocol was similar at both sites, different proportions of eligible children and different absolute numbers of children were enrolled at each site. Our analyses are hospital-based, and we were unable to compare community-level incidence, which may be a better way to describe the relative pathogen prevalence between communities. Additionally, there were differences in the referral levels and patient populations of the 2 hospitals. We collected only inpatient data, and it is possible that data could be biased by differences in admission practices between the sites. Many clinical features that have been shown to be predictors of pediatric mortality in low malaria endemicity areas [41] were more frequent at KCMC, a referral hospital. However, some predictors of pediatric mortality often associated with malaria, such as hypoglycemia and hyperlactatemia [42], were more common at Teule. Markers of disease severity and mortality among hospitalized children may differ between malarious vs nonmalarious areas, making comparisons of these features between the sites challenging. Children at Teule came from more rural areas and had mothers with fewer years of formal education than those at KCMC. Differences in socioeconomic status and in environmental exposures related to rurality, which we did not explore but may be risk factors for invasive salmonellosis, could have contributed to invasive salmonellosis differences between sites.
Self-reported use of antibacterials before admission was common, which could have resulted in reduced blood culture yield. Patients at KCMC were significantly more likely to have reported prior antibacterial use, which could have led to an underestimate of bacteremia at the low malaria transmission site. However, self-report reliability is questionable, as supported by a lack of statistical difference in bacteremia prevalence between those reporting and denying prior antibacterial use in this study.
CONCLUSIONS
Our comparisons between sites of high and low malaria endemicity in Tanzania add weight to the growing body of evidence that recent malaria infection and anemia are risk factors for iNTS and raise the possibility that NTS and Salmonella Typhi may compete in as yet unknown ways. It is possible that other undetermined factors such as Salmonella serovar–specific immune responses, interactions with coinfections, the gut microflora, and environmental reservoirs may provide venues for competition and replacement of these pathogens. The interaction between these pathogens, the environment, and the host requires further elucidation and is a compelling area for future research.
Notes
Acknowledgments. The authors thank the patients, staff, and leadership of Teule Hospital and Kilimanjaro Christian Medical Centre (KCMC) for their support and contributions to this research. We thank Ahaz T. Kulanga for providing administrative support at KCMC. Aikande Shoo, Halima Mohammed, Charles Mgaya, Selina Wycliffe, Emmanuel Swai, Edward Mtili, Christina Kemi, Stella Emmanuel, Rosalia Marwa, and Simphorosa Silaye were involved in clinical and laboratory data collection at Teule Hospital. Pilli M. Chambo, Beata V. Kyara, Beatus A. Massawe, Anna D. Mtei, Godfrey S. Mushi, Lillian E. Ngowi, Flora M. Nkya, and Winfrida H. Shirima reviewed and enrolled study participants at KCMC. Evaline M. Ndosi and Enock J. Kessy assisted with data entry and management at KCMC. Frank Mtei managed the data from Teule Hospital and provided the image for Figure 1. We thank Gino R. Micalizzi and John R. Bates, Queensland Health Forensic and Scientific Services, Brisbane, Australia, for assistance with serotyping nontyphoidal Salmonella strains. We acknowledge the Hubert-Yeargan Center for Global Health at Duke University for critical infrastructure support for the Kilimanjaro Christian Medical Centre–Duke University Collaboration.
Financial support. The work at Teule Hospital was supported by the European Commission (Europaid; grant code SANTE/2004/078-607). The work at Kilimanjaro Christian Medical Centre was supported by an International Studies on AIDS Associated Co-infections (ISAAC) award, a US National Institutes of Health (NIH)–funded program (U01 AI062563). This publication was made possible with help from the Duke University Center for AIDS Research, an NIH-funded program (2P30 AI064518). Authors received support from the NIH Fogarty International Center AIDS International Training and Research Program D43 PA-03-018 (J. A. C.), National Institute of Allergy and Infectious Diseases-AI007392 (H. M. B.), the Duke Clinical Trials Unit and Clinical Research Sites U01 AI069484 (J. A. C.), NIH grant R01TW009237 as part of the joint NIH-National Science Foundation Ecology of Infectious Disease program and the UK Economic and Social Research Council and Biotechnology and Biological Sciences Research Council (J. A. C.), Berkeley Fellowship (B. N.), Sir Halley Stewart Trust (B. N.), and Pfizer Pharmaceuticals (B. N.). At Teule Hospital, Pfizer Pharmaceuticals provided equipment and consumables for microbiology, and Abbott Pharmaceuticals provided reagents for HIV testing.
Potential conflicts of interest. All authors: No reported conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
References
- 1.Berkley JA, Lowe BS, Mwangi I, et al. Bacteremia among children admitted to a rural hospital in Kenya. N Engl J Med. 2005;352:39–47. doi: 10.1056/NEJMoa040275. [DOI] [PubMed] [Google Scholar]
- 2.Reddy EA, Shaw AV, Crump JA. Community acquired bloodstream infections in Africa: a systematic review and meta-analysis. Lancet Infect Dis. 2010;10:417–32. doi: 10.1016/S1473-3099(10)70072-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Crump JA, Mintz ED. Global trends in typhoid and paratyphoid fever. Clin Infect Dis. 2010;50:241–6. doi: 10.1086/649541. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Morpeth SC, Ramadhani HO, Crump JA. Invasive non-Typhi Salmonella disease in Africa. Clin Infect Dis. 2009;49:606–11. doi: 10.1086/603553. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Brent AJ, Oundo JO, Mwangi I, Ochola L, Lowe B, Berkley JA. Salmonella bacteremia in Kenyan children. Pediatr Infect Dis J. 2006;25:230–6. doi: 10.1097/01.inf.0000202066.02212.ff. [DOI] [PubMed] [Google Scholar]
- 6.Graham SM, Walsh AL, Molyneux EM, Phiri AJ, Molyneux ME. Clinical presentation of non-typhoidal Salmonella bacteraemia in Malawian children. Trans R Soc Trop Med Hyg. 2000;94:310–4. doi: 10.1016/s0035-9203(00)90337-7. [DOI] [PubMed] [Google Scholar]
- 7.Mabey DC, Brown A, Greenwood BM. Plasmodium falciparum malaria and Salmonella infections in Gambian children. J Infect Dis. 1987;155:1319–21. doi: 10.1093/infdis/155.6.1319. [DOI] [PubMed] [Google Scholar]
- 8.Maxwell CA, Chambo W, Mwaimu M, Magogo F, Carneiro IA, Curtis CF. Variation of malaria transmission and morbidity with altitude in Tanzania and with introduction of alphacypermethrin treated nets. Malar J. 2003;2:28. doi: 10.1186/1475-2875-2-28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Oesterholt MJ, Bousema JT, Mwerinde OK, et al. Spatial and temporal variation in malaria transmission in a low endemicity area in northern Tanzania. Malar J. 2006;5:98. doi: 10.1186/1475-2875-5-98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Crump JA, Ramadhani HO, Morrissey AB, et al. Invasive bacterial and fungal infections among hospitalized HIV-infected and HIV-uninfected children and infants in northern Tanzania. Trop Med Int Health. 2011;16:830–7. doi: 10.1111/j.1365-3156.2011.02774.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Nadjm B, Amos B, Mtove G, et al. WHO guidelines for antimicrobial treatment in children admitted to hospital in an area of intense Plasmodium falciparum transmission: prospective study. BMJ. 2010;340:c1350. doi: 10.1136/bmj.c1350. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Abba K, Deeks JJ, Olliaro P, et al. Rapid diagnostic tests for diagnosing uncomplicated P. falciparum malaria in endemic countries. Cochrane Database Syst Rev. 2011 doi: 10.1002/14651858.CD008122.pub2. CD008122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kyabayinze DJ, Tibenderana JK, Odong GW, Rwakimari JB, Counihan H. Operational accuracy and comparative persistent antigenicity of HRP2 rapid diagnostic tests for Plasmodium falciparum malaria in a hyperendemic region of Uganda. Malar J. 2008;7:221. doi: 10.1186/1475-2875-7-221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.World Health Organization. Management of the child with a serious infection or severe malnutrition: guidelines for care at the first referral level in developing countries. 2000.
- 15.World Health Organization and United Nations Childrens’ Fund. Geneva, Switzerland: WHO: 2009. WHO child growth standards and the identification of severe acute malnutrition in infants and children. [PubMed] [Google Scholar]
- 16.World Health Organization. Growth reference for 5-19 years: WHO reference 2007. Available at: http://www.who.int/growthref/who2007_bmi_for_age/en/index.html. Accessed 7 April 2013.
- 17.World Health Organization. Geneva, Switzerland: WHO; 2009. Guidelines for an integrated approach to the nutritional care of HIV-infected children (6 months–14 years) [PubMed] [Google Scholar]
- 18.Tanzania National Bureau of Statistics. Village statistics, 2002 population and housing census. Available at: http://www.nbs.go.tz/ Accessed 25 March 2010.
- 19.Tabu C, Breiman RF, Ochieng B, et al. Differing burden and epidemiology of non-Typhi Salmonella bacteremia in rural and urban Kenya, 2006–2009. PLoS One. 2012;7:e31237. doi: 10.1371/journal.pone.0031237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Were T, Davenport GC, Hittner JB, et al. Bacteremia in Kenyan children presenting with malaria. J Clin Microbiol. 2011;49:671–6. doi: 10.1128/JCM.01864-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Walsh AL, Phiri AJ, Graham SM, Molyneux EM, Molyneux ME. Bacteremia in febrile Malawian children: clinical and microbiologic features. Pediatr Infect Dis J. 2000;19:312–8. doi: 10.1097/00006454-200004000-00010. [DOI] [PubMed] [Google Scholar]
- 22.Mtove G, Amos B, Nadjm B, et al. Decreasing incidence of severe malaria and community-acquired bacteraemia among hospitalized children in Muheza, north-eastern Tanzania, 2006-2010. Malar J. 2011;10:320. doi: 10.1186/1475-2875-10-320. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Feasey NA, Heyderman RS, Gordon MA. Salmonella blood-stream infection in Malawi: a fall in nontyphoidal Salmonella and an outbreak of typhoid. 8th International Conference on Typhoid Fever and Other Invasive Salmonelloses,; Dhaka, Bangladesh. 2013. Abstract 88. [Google Scholar]
- 24.Berkley J, Mwarumba S, Bramham K, Lowe B, Marsh K. Bacteraemia complicating severe malaria in children. Trans R Soc Trop Med Hyg. 1999;93:283–6. doi: 10.1016/s0035-9203(99)90024-x. [DOI] [PubMed] [Google Scholar]
- 25.Scott JA, Berkley JA, Mwangi I, et al. Relation between falciparum malaria and bacteraemia in Kenyan children: a population-based, case-control study and a longitudinal study. Lancet. 2011;378:1316–23. doi: 10.1016/S0140-6736(11)60888-X. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Mackenzie G, Ceesay SJ, Hill PC, et al. A decline in the incidence of invasive non-typhoidal Salmonella infection in The Gambia temporally associated with a decline in malaria infection. PLoS One. 2010;5:e10568. doi: 10.1371/journal.pone.0010568. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Green SD, Cheesbrough JS. Salmonella bacteraemia among young children at a rural hospital in western Zaire. Ann Trop Paediatr. 1993;13:45–53. doi: 10.1080/02724936.1993.11747624. [DOI] [PubMed] [Google Scholar]
- 28.Bronzan RN, Taylor TE, Mwenechanya J, et al. Bacteremia in Malawian children with severe malaria: prevalence, etiology, HIV coinfection, and outcome. J Infect Dis. 2007;195:895–904. doi: 10.1086/511437. [DOI] [PubMed] [Google Scholar]
- 29.Bassat Q, Guinovart C, Sigauque B, et al. Severe malaria and concomitant bacteraemia in children admitted to a rural Mozambican hospital. Trop Med Int Health. 2009;14:1011–9. doi: 10.1111/j.1365-3156.2009.02326.x. [DOI] [PubMed] [Google Scholar]
- 30.Cunnington AJ, de Souza JB, Walther M, Riley EM. Malaria impairs resistance to Salmonella through heme- and heme oxygenase-dependent dysfunctional granulocyte mobilization. Nat Med. 2012;18:120–7. doi: 10.1038/nm.2601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Cunnington AJ, Njie M, Correa S, Takem EN, Riley EM, Walther M. Prolonged neutrophil dysfunction after Plasmodium falciparum malaria is related to hemolysis and heme oxygenase-1 induction. J Immunol. 2012;189:5336–46. doi: 10.4049/jimmunol.1201028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Roux CM, Butler BP, Chau JY, et al. Both hemolytic anemia and malaria parasite-specific factors increase susceptibility to nontyphoidal Salmonella enterica serovar Typhimurium infection in mice. Infect Immun. 2010;78:1520–7. doi: 10.1128/IAI.00887-09. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Olsson RA, Johnston EH. Histopathologic changes and small-bowel absorption in falciparum malaria. Am J Trop Med Hyg. 1969;18:355–9. doi: 10.4269/ajtmh.1969.18.355. [DOI] [PubMed] [Google Scholar]
- 34.Davis TM, Pongponratan E, Supanaranond W, et al. Skeletal muscle involvement in falciparum malaria: biochemical and ultrastructural study. Clin Infect Dis. 1999;29:831–5. doi: 10.1086/520444. [DOI] [PubMed] [Google Scholar]
- 35.Seydel KB, Milner DA, Jr, Kamiza SB, Molyneux ME, Taylor TE. The distribution and intensity of parasite sequestration in comatose Malawian children. J Infect Dis. 2006;194:208–5. doi: 10.1086/505078. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Feasey NA, Dougan G, Kingsley RA, Heyderman RS, Gordon MA. Invasive non-typhoidal Salmonella disease: an emerging and neglected tropical disease in Africa. Lancet. 2012;379:2489–99. doi: 10.1016/S0140-6736(11)61752-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Arthur G, Nduba VN, Kariuki SM, Kimari J, Bhatt SM, Gilks CF. Trends in bloodstream infections among human immunodeficiency virus-infected adults admitted to a hospital in Nairobi, Kenya, during the last decade. Clin Infect Dis. 2001;33:248–56. doi: 10.1086/321820. [DOI] [PubMed] [Google Scholar]
- 38.Gordon MA. Salmonella infections in immunocompromised adults. J Infect. 2008;56:413–22. doi: 10.1016/j.jinf.2008.03.012. [DOI] [PubMed] [Google Scholar]
- 39.Crump JA, Ramadhani HO, Morrissey AB, et al. Invasive bacterial and fungal infections among hospitalized HIV-infected and HIV-uninfected adults and adolescents in northern Tanzania. Clin Infect Dis. 2011;52:341–8. doi: 10.1093/cid/ciq103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Levine MM, Farag TH. Invasive Salmonella infections and HIV in northern Tanzania. Clin Infect Dis. 2011;52:349–51. doi: 10.1093/cid/ciq109. [DOI] [PubMed] [Google Scholar]
- 41.Clifton DC, Ramadhani HO, Msuya LJ, et al. Predicting mortality for paediatric inpatients where malaria is uncommon. Arch Dis Child. 2012;97:889–94. doi: 10.1136/archdischild-2012-301812. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Jallow M, Casals-Pascual C, Ackerman H, et al. Clinical features of severe malaria associated with death: a 13-year observational study in the Gambia. PLoS One. 2012;7:e45645. doi: 10.1371/journal.pone.0045645. [DOI] [PMC free article] [PubMed] [Google Scholar]