Prevalence, probability, and outcomes of typhoidal/non-typhoidal Salmonella and malaria co-infection among febrile patients: a systematic review and meta-analysis

The geographical overlaps of malaria parasites and Salmonella spp. can lead to co-infection of these two pathogens, especially in the tropics where malaria is endemic. Moreover, few literatures suggested that malaria infection was associated with Salmonella bacteremia. Therefore, this study quantified pooled prevalence of typhoidal/non-typhoidal Salmonella (NTS) and probability of typhoidal/NTS and malaria co-infection among febrile patients. The systematic review protocol was registered at PROSPERO (CRD42021252322). Studies on co-infection of typhoidal/NTS and malaria were searched in PubMed, Scopus, and Web of Science. The risk of bias of the included studies was assessed using the checklist for analytical cross-sectional studies developed by the Joanna Briggs Institute. Meta-analyses on the following criteria were performed: (1) pooled prevalence of typhoidal/NTS and malaria co-infection among febrile patients, (2) pooled prevalence of typhoidal/NTS among malaria patients, (3) pooled prevalence of malaria infections among patients with Salmonella spp. infection, and (4) probability of typhoidal/NTS and malaria co-infection among febrile patients. Additionally, the case fatality rate and mean difference of malarial parasitemia between typhoidal/NTS and malaria co-infection and Plasmodium monoinfection were also determined. The subgroup analyses of typhoidal/NTS, regions (Africa and Asia), countries, time (publication year), characteristics of participants, and diagnostic tests for identifying Salmonella spp. were also conducted. A sensitivity test was performed to determine the robustness of the study outcomes. Publication bias among the included studies was evaluated using the funnel plot and Egger’s test. All analyses were performed using Stata version 15 (StataCorp LLC, Texas, USA) with a p-value < 0.05 indicating statistical significance. Eighty-one studies that met the eligibility criteria were included in the analyses. Of the 73,775 study participants, 4523 had typhoidal/NTS and malaria co-infections. The pooled prevalence rates of typhoidal/NTS and malaria co-infection among febrile patients were 14% (95% confidence interval [CI], 9–19%; I2, 99.4%; 2971/17,720 cases) and 1% (95% CI 1–1%; I2, 89.9%; 252/29,081 cases) using the Widal test and culture methods for identifying Salmonella spp., respectively. The pooled prevalence rates of typhoidal/NTS infection among patients with malaria were 31% (95% CI 23–39%; I2, 99.5%; 3202/19,208 cases) and 3% (95% CI 2–3%; I2, 86.8%; 407/40,426 cases) using the Widal test and culture methods for identifying Salmonella spp., respectively. The pooled prevalence rates of malaria infection among patients with typhoidal/NTS were 17% (95% CI 6–29%; I2, 33.3%; 13/75 cases) and 43% (95% CI 32–53%; I2, 89.1%; 287/736 cases), respectively. Malaria infection was associated with typhoidal/NTS in children aged < 15 years (p < 0.0001; odds ratio, 0.36; 95% CI 0.23–0.58; I2, 73.9%; 3188/43,212 cases). The case fatality rate in patients with malaria and NTS co-infections was 16% (95% CI 9–24%; I2, 89.1%; 18/103 cases). From the view of the present study, the inappropriate use of the Widal test for Salmonella spp. diagnosis can overestimate the prevalence of typhoidal/NTS and malaria co-infections. Malaria infection associated with typhoidal/NTS in children and the high case fatality rates among few patients with co-infections were highlighted. Future prospective longitudinal studies using the appropriate and confirmatory dsiagnosis for Salmonella spp. infections are highly recommended to ensure the real prevalence of co-infection and highlight the outcome of co-infection for providing adequate treatment in febrile patients who live in areas where malaria is endemic, such as tropical Africa and India.

Malaria is a major public health problem in tropical and subtropical countries. The World Health Organization (WHO) reported 229 million cases of malaria and 409,000 deaths in 2019 1 . Most of these cases (51% globally) were found in Nigeria (27%), the Democratic Republic of the Congo (12%), Uganda (5%), Mozambique (4%), and Niger (3%), whereas 51% of malaria deaths occurred in Nigeria (23%), the Democratic Republic of the Congo (11%), the United Republic of Tanzania (5%), Mozambique (4%), Niger (4%), and Burkina Faso (4%) 1 . Although the incidence of malaria declined from 2000 to 2019, its diagnosis among febrile patients in malariaendemic settings remains challenging; malaria and typhoid and non-typhoid fever are co-endemic and have similar clinical signs and symptoms.
Salmonella species are Gram-negative bacilli members of Enterobacteriaceae and associated with human infection 2,3 . Salmonella is comprised of two major species, namely, Salmonella enterica and Salmonella bongori. S. enterica are classified into six serotypes, which are differentiated based on their antigenicity 3 . Several Salmonella enterica serotypes cause typhoid fever. S. typhi and S. paratyphi, collectively referred to as typhoidal Salmonella, are the most common species that cause enteric fever (typhoid fever and paratyphoid fever). While both fevers share clinical symptoms, paratyphoid fever tends to be more benign 4,5 . Paratyphoid fever is most commonly acquired by ingesting contaminated food or water 5 . Humans are key reservoir hosts for typhoidal Salmonella and contribute to disease transmission and dissemination 5 . Typhoid fever is characterized by gastroenteritis and presents with nonspecific clinical symptoms, such as high fever, fatigue, headache, malaise, abdominal pain, nausea, vomiting, constipation, and diarrhea 5,6 . These symptoms are indistinguishable from other causes of fever, such as malaria 5 . The complications of typhoid fever include septicemia, meningitidis, and immunological symptoms 7 . Typhoid fever may be mild or severe, and complications may contribute to typhoid-related deaths 7,8 . Non-typhoid fever is a febrile illness caused by non-typhoidal Salmonella (NTS), including S. enteritidis and S. typhimurium 9 . NTS infections most often cause mild gastroenteritis, which is usually self-limiting 7 . Recently, NTS infections have been associated with septicemia and high mortality rates in immunocompromised patients in sub-Saharan Africa 10 .
The Salmonella infection rate is high in low-and middle-income countries with > 100 per 100,000 infected people annually 11,12 . Typhoid fever is an important cause of morbidity and mortality worldwide, with an estimated 16-33 million cases and 500,000 to 600,000 deaths annually 13 . Typhoid is endemic in developing countries, especially Africa, whereas developed countries have a much lower incidence. The majority of patients in developed countries are travelers returning from endemic areas 14 . In developing countries, especially in Southeast Asia and Africa, NTS is endemic and is a global burden, contrary to typhoidal Salmonella 12,14 . Poor water quality, poor handwashing habits, and consumption of untreated drinking water or unsafe food are the main causes of typhoid fever 15 . Therefore, people with low socioeconomic status and poor or improper hygiene have high risk of fecal-oral enteric infections, including typhoidal and NTS. The most recent meta-analysis revealed that household behaviors, including poor hygiene and consumption of unsafe food and untreated water, increase the risk of typhoid transmission 16 . Among children in Africa, NTS is a leading cause of bacteremia, whereas typhoid fever has a relatively low burden [17][18][19] . Another study demonstrated that typhoid fever is more common in older children with a period of fever, whereas non-typhoidal bacteremia frequently develops in younger children of poorly educated women or women with low socioeconomic status 20 .

Outcomes of malaria and typhoidal/NTS co-infections.
A limited number of studies reported clinical outcomes of patients with co-infections (Table S2)  www.nature.com/scientificreports/ reported outcomes of malaria and typhoid co-infections. The case fatality rate in patients with malaria and NTS co-infections was 16% (95% CI 9-24%; I 2 , 89.1%; three studies), while one study 63 reported the case fatality rate in patients with malaria and typhoidal Salmonella co-infections at 33% (95% CI 6-79%) ( Supplementary  Fig. S1). The difference in malarial parasitemia between co-infections and Plasmodium spp. monoinfection was estimated by two studies 79,80 . Results showed a higher mean of malarial parasitemia in patients with co-infections than those with Plasmodium spp. monoinfection (p, 0.023; WMD, 7926.7 parasites⁄µL of blood (95% CI 1091-14,762.3 parasites⁄µL of blood; I 2 , 0%, two studies) ( Supplementary Fig. S2). The study by Bassat et al. 79 showed a lower rate of respiratory distress in patients with co-infections ( Fig. S3). However, the use of a fixed-effects model in the meta-analysis indicated that Plasmodium spp. and Salmonella spp. co-infection did not occur by chance (p < 0.0001; odds ratio, 0.82; 95% CI 0.76-0.88; I 2 , 93.6%) (Supplementary Fig. S4).   Fig. S6), indicating that the funnel plot asymmetry was likely due to factors such as heterogeneity, selection bias, and quality of the included studies rather than publication bias.

Discussion
The present meta-analysis revealed a high prevalence of malaria and typhoidal/NTS co-infections among febrile patients detected using the Widal test (14%) and a low prevalence of malaria and typhoidal/NTS co-infections among febrile patients detected using blood cultures (1%). Moreover, the meta-analysis demonstrated that the prevalence of typhoidal/NTS infection among patients with malaria using the Widal test was high (31%), whereas the prevalence of typhoid/non-typhoid using blood culture was low (3%). A high prevalence of malaria infections among patients with typhoidal Salmonella spp. infections (17%) and NTS (43%) was also detected. The highest prevalence of co-infections detected using the Widal test was observed in Cameroon 40 , followed by Nigeria 36,51,59,75,90,103 and Sierra Leone 87 , compared with Ghana, India, Ethiopia, Tanzania, and Pakistan. In using blood cultures, the gold standard method for the identification of Salmonella spp., the results indicated that the highest prevalence of co-infection was reported in Nigeria 37,57,83 compared with India, Tanzania, Ghana, and Kenya. Based on these results, typhoid/non-typhoid and malaria co-infection among febrile patients frequently occurred in Nigeria. In 2020, Nigeria accounted for the most malaria cases (27%) and malaria-related deaths (23%) worldwide 1 . Moreover, typhoid fever is a major disease in Nigeria due to increased urbanization, insufficient water supply, movement of immigrant workers, poor processing of human waste, and overuse of antibiotics 109 . Due to the co-endemicity of these two pathogens, the possibility of co-infection might increase in this country.
Using the data from studies perfoming blood culture to identify typhoidal/NTS infection, the subgroup analysis of typhoidal/NTS infection demonstrated low prevalence of malaria and typhoid co-infections among febrile patients (1%) and low prevalence of typhoid among patients with malaria (6%). Moreover, the low prevalence of malaria and NTS co-infections among febrile patients (1%) and NTS infection among patients with malaria (2%) www.nature.com/scientificreports/ was observed. The highest prevalence of malaria and typhoid co-infections among febrile patients was reported in Nigeria, suggesting that malaria and typhoid are indeed halo-endemic in this area 83 . In the meta-analysis of typhoid among patients with malaria, the highest prevalence of typhoid among patients with malaria was noted in Nigeria 38,57 . These results suggested an increasing episode of persistent fever among patients with S. typhi and P. falciparum infections in Nigeria. For NTS infection among patients with malaria, the prevalence was higest in Kenya, and NTS infection was the most common bacteremia in children with malaria 28 . The high rate of bacteremia in patients with malaria in Nigeria might be due to the high prevalence of NTS infections and malnutrition 28 .
Using the data from studies performing blood culture to identify typhoidal/NTS infection, the subgroup analysis of regions demonstrated that the prevalence of malaria and typhoidal/NTS co-infections were 1% in both Africa and Asia. However, the prevalence of typhoidal/NTS among patients with malaria was higher in Asia (6%) than those of Africa (2%). The difference in the prevalence of typhoidal/NTS co-infections between two regions might be caused by the heterogeneity of the prevalence estimates between two regions or real difference caused by environmental factors. For example, studies in India suggested that malaria and typhoid are endemic because of poor hygiene and environmental factors 104,108 . In Africa, although the pooled prevalence of typhoidal/NTS infection among patients with malaria was lower than those in Asia; the results of individual studies were heterogenous. For example, the high prevalence of typhoidal/NTS infection among patients with malaria were reported by four studies conducted in Nigeria 37,38,57,83 , while a lower prevalence was reported by other studies included in the meta-analysis. www.nature.com/scientificreports/ Using the data from studies performing blood culture to identify typhoidal/NTS infection, the subgroup analysis of time (year of publication) showed that the prevalence of malaria and typhoid co-infections among febrile patients, and typhoidal Salmonella infections among patients with malaria was highest in 2016, while lower prevalence was reported in before and after 2016. In 2016, three studies conducted in Nigeria and India 66,83,104 reported the highest prevalence rates of typhoid among patients with malaria. The peak of typhoid among patients with malaria in 2016 was different from those of NTS infections among patients with malaria. The subgroup analysis showed that the peak prevalence rate of NTS infection among patients with malaria was highest in 2011, lower in 2012-2016, and 2001. These results indicated that the prevalence of NTS might decreased with time in 2011-2016, while the prevalence of typhoid among patients with malaria might not depend on time, which are needed to be further investigated.
Using the data from studies performing blood culture to identify typhoidal/NTS infection, the subgroup analysis of age of patients demonstrated that the prevalence rate of typhoidal/NTS infection among patients with malaria was higher in adults (3%) compared to that in children (2%). The previous study showed that peaks of NTS infection occurred in children aged < 2 years and adults aged 25-40 years 110 , while the lower rate of NTS infection occurred in children aged less than 12 years old, and the proportion of hospitalization was decreased with age 111 . These age groups were supported by the subgroup analysis of age that the prevalence of typhoidal/ NTS was higher in adults than in children. Nevertheless, as the limitation of age information in studies reported typhoidal/NTS co-infections among febrile patients, the subgroups analysis of age might not represent the exact difference in the prevalence of typhoidal/NTS co-infections between adults and children.
The present meta-analysis demonstrated a wide gap in prevalence of malaria and typhoid/non-typhoid coinfections among febrile patients as measured by the Widal test and blood culture in analysis. The high rate of typhoid/non-typhoid and malaria co-infections detected using the Widal test and low rate of co-infections detected using blood cultures might be due to the lack of differentiation between Salmonella species/serotypes by the Widal test and cross-reactivity with other Enterobacteriaceae 5,7 . Moreover, false-positive Widal tests have been reported in patients with malaria and other infections 5 . The malaria Plasmodium may share similar strong immunogenic antigens with the typhoidal Salmonella (S. typhi); thus, Plasmodium infections could induce the generation of antibodies against S. typhi antigens, leading to cross-reactivity and false-positive results 47 . Furthermore, malaria loading strongly correlated with Salmonella antibody titers in numerous studies 47 . This cross-reaction of typhoidal/NTS antibodies with malarial antigen leads to overdiagnosis of typhoid fever 74,112 . The Widal test also generates false-negative results if patients are tested during the early phase of typhoid fever 5 .  www.nature.com/scientificreports/ The high prevalence of typhoid fever may also be due to poor interpretation of the Widal test when diagnosing typhoid fever 113 . Nevertheless, in Africa and other territories, the Widal test is the most common diagnostic tool used for typhoid fever; owing to its low cost, ease of performance, and minimal training and equipment requirements. Of note, false-positive results of the Widal tests in febrile patients suspected of having Salmonella spp. infection may lead to incorrect treatment for malaria parasites. Thus, careful interpretation of the Widal test for the diagnoses of typhoid fever in resource-poor countries is required, as the overdiagnosis of typhoid fever can lead to unnecessary treatment of patients with antibiotics, microbial resistance, and poor outcome. The use of Widal test alone for the diagnosis of typhoid fever will cause misdiagnoses.
Using blood cultures alone to identify Salmonella spp. infection may underestimate Salmonella spp. infections, as blood culture has a lower sensitivity compared with the Widal test. Negative blood culture test results may be noted in patients with acute disease before the antibody response 5 . Based on the results of this study, the Widal test should not be used alone but in combination with blood/stool cultures. Therefore, a combination of the Widal test and blood and stool cultures is an excellent choice for diagnosing Salmonella spp. infection among febrile patients or patients with malaria. Although the high laboratory expenses for combination testing are difficult to overcome, the use of more than one diagnostic method to identify Salmonella spp. infections among patients with malaria is important to prevent incorrect treatment and misdiagnoses of malaria and other acute febrile illnesses. Infections caused by typhoidal Salmonella, including S. typhi and S. paratyphi, and the associated serious complications require treatment with antibiotics, including chloramphenicol, cefixime, amoxicillin, trimethoprim/sulfamethoxazole, azithromycin, aztreonam, and cefotaxime, to prevent severe illness and death 3,114 . NTS infections do not usually require treatment with antibiotics. However, complications, such as septicemia and meningitis, require treatment with ciprofloxacin, ceftriaxone, and ampicillin, according to the WHO 3,114 . Presently, antibiotic resistance of Salmonella species is an emerging threat, so reliable diagnostic test and appropriate treatments for typhoid/non-typhoid fever are important.
The present meta-analysis demonstrated that Salmonella spp. bacteremia developed in approximately 2% of patients with severe malaria. This occurrence was not much different from the Salmonella spp. bacteremia pooled prevalence of 3% in patients with non-severe malaria. Several mechanisms have been suggested to elucidate why patients with malaria may be predisposed to Salmonella spp. infection and bacteremia. First, immunosuppression occurs during malaria infection and treatment 115 . Second, malaria can lead to hemolysis, which may predispose patients to infection with Gram-negative bacteria, such as typhoidal Salmonella/NTS spp. 69 . Third, changes in iron storage metabolism from malaria-induced hemolysis cause neutrophil dysfunction and increased susceptibility [116][117][118] . Increased free iron from hemolysis may also promote the survival of Salmonella spp. 19 . Fourth, the sequestration of parasitized red blood cells in the intestine causes reduced blood flow in the mucosal gut barrier, which increases intestinal susceptibility to bacterial infection 119,120 . The high rate of NTS bacteremia  is well described in patients with malaria-related severe anemia 121 . Severe anemia and hemolysis increase the iron level in the blood and tissues; therefore, pathogens can be actively transported, and iron acquisition is easier 121 . Based on our results, the increased risk of typhoidal Salmonella bacteremia in patients with severe malaria might reflect the high rate of parasite sequestration and vital organ dysfunction. Moreover, bacteremia cannot be excluded from patients with severe malaria; severe malaria is difficult to distinguish from bacterial sepsis 56,85 . Therefore, the WHO guidelines for malaria recommend that children with severe falciparum malaria in high-transmission areas should receive empirical broad-spectrum antibacterial therapy. However, empirical antibiotics should not be administered to adults with severe malaria unless there is clear evidence of bacterial infection 122 . In the low-transmission areas, WHO suggests that physicians should determine whether patients should receive antibiotics depending on the patient's condition or parasitemia levels, but patients with severe malaria should not be routinely treated with antibiotics 122,123 . In addition to the WHO guidelines, two studies conducted in Myanmar 58,78 stated that "clinicians should have a lower threshold for commencing empirical antibacterial therapy in adults diagnosed with falciparum malaria in these locations than is presently recommended. " The present meta-analysis revealed that typhoidal/NTS and malaria co-infection occurred by chance when the random-effects model was employed to combine the effect estimates. However, the subgroup analysis demonstrated a decreased odds ratio of co-infection in children aged < 15 years, indicating that the current malaria infection was negatively associated with typhoidal/non-typhoidal Salmonella spp. infection in children 35,41,44,46,53,73 . Although the meta-analysis did not provide a significant outcome, individual studies demonstrated significantly higher odds ratios of co-infection in Nigeria 57,59,82,90,95,97,101   between typhoid/non-typhoid and malaria co-infection in some way. Further studies are required to investigate this association. The present meta-analysis of case fatality rate of patients with co-infection demonstrated the high rate of mortality (16%) without heterogeneity among the three included studies 60,80,92 . These three studies enrolled patients with severe malaria and co-infected with NTS and indicated that both diseases facilitate the higher fatality rate than those of the malaria or NTS infection alone. Moreover, the meta-analysis of two studies 79,80 showed a higher mean parasitemia level in patients with malaria and co-infected with NTS compared to those with malaria alone (without heterogeneity, 0%), but it is important to note the limitation in the number of included studies in the analysis. Therefore, there is a need to investigate if co-infection of malaria and NTS leads to poor outcome or demonstrated the association of both diseases.
This study had several limitations. First, most included studies were cross-sectional studies that determined the prevalence of typhoidal/non-typhoidal Salmonella spp. and malaria co-infection. Therefore, data were not available to determine the differences between co-infected patients and mono-infected patients. Second, the number of studies evaluating the occurrence of Salmonella spp. bacteremia in patients with severe malaria was limited; therefore, the pooled prevalence of Salmonella spp. bacteremia in patients with severe malaria might not represent all patients with severe malaria. Third, the heterogeneity among the included studies used to determine the probability of typhoidal/non-typhoidal Salmonella spp. and malaria co-infection was high; therefore, the association between typhoidal/non-typhoidal Salmonella spp. and malaria co-infection should be carefully interpreted with the results from the sensitivity test. Compared with the previous systematic review 32 , the present study excluded studies with recent malaria infection; most included studies used microscopy rather than RDTs for malaria detection; and there was no publication bias among the included studies. In conclusion, whether typhoidal/non-typhoidal Salmonella spp. and malaria co-infection occurred by chance or not, healthcare providers must provide support to patients with nonspecific clinical symptoms of malaria or typhoidal/non-typhoidal diseases. In the present study, malaria associated with typhoidal/NTS infection in children and the high case fatality rate among few co-infected patients were highlighted. Future prospective longitudinal studies using the appropriate and confirmatory diagnosis for Salmonella spp. infections are highly recommended to ensure the real prevalence of co-infection and highlight the outcome of co-infection for providing adequate treatment of co-infections in febrile patients who live in areas where malaria is endemic like tropical Africa or India.    Figure 24. Probability of Plasmodium spp. and Salmonella spp. co-infections. OR odds ratio, CI confidence interval, NS not specified.