Arthropod-Borne Bacteria Cause Nonmalarial Fever in Rural Ethiopia: A Cross-Sectional Study in 394 Patients

Abstract

Bacterial arthropod-borne pathogens are a common cause of fever in Africa, but their precise impact is unknown and usually underdiagnosed in the basic rural laboratories of low-resourced African countries. Our aim was to determine the prevalence of arthropod-borne bacterial diseases causing fever among malaria smear-negative patients in a rural hospital located in Ethiopia. The study population included patients aged 2 years or older; referred to Gambo Rural General Hospital (West Arsi, Ethiopia), between July and November 2013, for fever or report of fever in the previous 48 h; attending the outpatient department; and testing negative for malaria by Giemsa-stained thin blood smears. We extracted DNA from 394 whole blood samples, using reverse line blot assays of amplicons to look for bacteria from the genera: Anaplasma, Bartonella, Borrelia, Coxiella, Ehrlichia, Francisella, and Rickettsia. Thirteen patients showed presence of DNA for these pathogens: three each by Borrelia spp., the Francisella group (F. tularensis tularensis, F. tularensis holartica, and F. novicia), Rickettsia bellii, and Rickettsia Felis, and one by Bartonella rochalimae. Thus, in this rural area of Africa, febrile symptoms could be due to bacteria transmitted by arthropods. Further studies are needed to evaluate the pathogenic role of R. bellii.

Introduction

After malaria, rickettsial diseases are widely considered to be one of the most important causes of systemic febrile illness in sub-Saharan Africa; however, the precise disease burden is largely unquantified (Freedman et al. 2006, Jensenius et al. 2009). The incidence and prevalence of rickettsioses diagnosis by DNA amplification in nonmalaria febrile illness (NMFI) in indigenous populations in sub-Saharan Africa are poorly studied, with most investigations taking place in Western Africa and Kenya. The prevalence of Rickettsia spp. in NMFI ranges from 1.5% to 10.5%, with R. felis predominating (Mourembou et al. 2015, Sothmann et al. 2017).

Since reclassifying the genus Bartonella in 1993, the number of species has ballooned to the current 45 members (Okaro et al. 2017). Infection by Bartonella may be associated with a wide range of clinical presentations, including chronic bacteremia, trench fever, bacillary angiomatosis, and endocarditis. Despite the evident interest of bartonellosis, there is still a need to know its prevalence in sub-Saharan Africa. In Senegal and South Africa, it was present in 2.5% of the NMFIs studied (Trataris et al. 2012, Diatta et al. 2014).

Borrelia causes fever in sub-Saharan Africa and B. recurrentis, louse-borne relapsing fever, produces sporadic illness and outbreaks especially in Ethiopia, Eritrea, Somalia, and Sudan. This bacterium has also been described in Europe among immigrants from sub-Saharan Africa (Hoch et al. 2015, Zammarchi et al. 2016). The genus Francisella is a highly virulent intracellular Gram-negative bacterium, classified into different species, mainly F. turalensis and F. novicida. The associated infections are transmitted by direct contact with animals and their products. Also, the lice can transmit them. F. tularensis has been classified as a biothreat; of its four subspecies, F. tularensis subsp. tularensis and F. tularensis subsp. holarctica are responsible for tularemia in humans (Njeru et al. 2017).

The role of this microorganism in infection in Africa is poorly understood and usually underdiagnosed in the basic rural laboratories of low-resourced African countries. We performed this study to fill some of the large gaps in knowledge of arthropod-borne pathogens as a cause of nonmalarial fever in sub-Saharan Africa. We aimed to determine the bacterial species causing zoonoses and belonging to the genera Anaplasma, Bartonella, Borrelia, Coxiella, Ehrlichia, Francisella, and Rickettsia in NMFI in rural Ethiopia.

Materials and Methods

The study took place in Gambo Rural General Hospital (GRH), situated in the province of West Arsi, Ethiopia, 245 km southeast of the capital, Addis Ababa, at an altitude of 2250 meters (∼7382 feet) above sea level (7°18′22.4″N+38°48′54.7″E). GRH serves 11 municipalities (or “kebeles”), whose population is estimated to be 100,000 inhabitants. The temperature varies between 13°C and 30°C, with varying levels of precipitation, concentrated from June to October. Subsistence farming and animal husbandry are the residents' major occupations.

Clinical sample collection

The study was prospective and clinical sample collection was performed between July 1 and November 30, 2013. The study population included patients aged 2 years or older; referred to the hospital laboratory for fever (axillary temperature >37.5°C) or report of fever in the previous 48 h; attending the outpatient department; and testing negative for malaria by Giemsa-stained thin blood smears, with no malaria trophozoites or gametocytes identified by microscopic analysis. The differential diagnosis in study participants with microscopy-negative malaria considered short febrile illnesses, including a relapsing fever, typhus, typhoid fever, and meningitis. These were treated with tetracycline hydrochloride or chloramphenicol in accordance with Ethiopian national guidelines. We collected the main symptoms, signs, and analytical variables from patients with NMFI.

A medical attendant examined each individual with fever, collecting whole blood (three to four drops) from the fingertip by lancet stick and preserving samples on filter paper (Whatman 3 MM) for molecular assay.

DNA extraction

Samples included in the study were received by the Laboratory of Special Pathogens of the National Microbiology Center of Spain (Institute of Health Carlos III, Madrid, Spain) for diagnostic purposes. DNA was extracted from blood using the QIAamp Tissue Kit (QIAGEN, Hilden, Germany), according to the manufacturer's instructions. A previous overnight treatment with a proteinase K solution (20 mg/mL) and a final step of 15 min at 100°C for inactivation of the protease were included. The DNA concentration was measured using the NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE).

Molecular detection

Around 200 ng of DNA from each sample was analyzed with routine in-house single PCRs with specific biotinylated primers followed by a RLB for the individual detection of Anaplasma spp., Bartonella spp., Borrelia spp., Coxiella burnetii, Francisella spp., and Rickettsia spp.

In brief, PCR amplifications for each bacteria genus were carried out individually using biotinylated primers in the MJ Research PCT-200 (Ecogen, Barcelona, Spain) as described previously (Jado et al. 2006). Subsequently, an RLB was performed hybridizing PCR products to specific probes with a C6 amino-link modification and detected using immunoenzymatic chemiluminescence (Super Signal West Dura Extended Duration Substrate; Pierce Biotechnology, Rockford, IL) as previously described (Jado et al. 2006). The sensitivity of the PCR-RLBs was assessed by amplifying 10², 10, and 1 genome equivalents for Bartonella, Borrelia, C. burnetii, and Francisella or plasmid copies with inserts of synthetic DNA for the detection of Anaplasma and Rickettsia in spiked distilled water (Jado et al. 2006, Toledo et al. 2009). After multiplex PCR combined with RLB, a conventional nested PCR and sequencing of at least one informative standard amplicon for each genus-specific were performed. These amplicons can be found in previous studies performed by Garcia-Esteban et al. (2008) and Jado et al. (2006). In summary, for the detection of Bartonella spp., a multiplex PCR to amplify the 16S rRNA and the hypervariable intergenic transcribed spacer 16S-23S rRNA (ITS) was carried out. The probe designed for the 16S rRNA is generic and allows the detection of Bartonella spp., whereas for species identification, 17 species-specific probes were designed for the 16S-23S rRNA (Garcia-Esteban et al. 2008).

For the detection of Borrelia spp., a generic probe and primers were designed to target the 16S rRNA. Subsequently, positive samples were subjected to a second PCR to amplify the 5S (rrf)–23S (rrl) intergenic spacer, genospecies were determined by using specific probes (Rijpkema et al. 1995). For the identification of C. burnetii, the transposase IS1111 was targeted (Toledo et al. 2009). To identify Francisella spp., the lpnA gene was amplified (Escudero et al. 2008). Lastly, Rickettsia was detected by amplifying a fragment of the 23S-5S rRNA intergenic spacer that allocates generic and specific probes, allowing the differentiation of the spotted fever and typhus groups, as well as the identification of Rickettsia at the species level (Jado et al. 2006).

Statistical analysis

We entered data into a spreadsheet using Microsoft Excel 2011 and analyzed the data using IBM SPSS statistical software, version 22.0 (SPSS, Inc., Chicago, IL). To analyze the association of categorical variables and the presence of pathogen DNA, we used either the chi-square test with Yates' correction or Fisher's exact test, where appropriate. To analyze continuous variables (age and white blood cell counts), we used the Mann–Whitney U-test, as data did not follow a normal distribution. We considered p values of <0.05 to be statistically significant. We obtained the estimates of prevalence along with their 95% CIs.

Ethics approval and consent to participate

The Research and Publication Committee of the GRH, and the Health Unit and Ethical Review Committee of the Ethiopian Catholic Secretary (GH/MSMHF/709), approved the study protocol, which conformed to the ethical guidelines of the Declaration of Helsinki. Given the high rates of illiteracy in the population, we deemed written consent to be unfeasible. Therefore, we obtained oral informed consent from each participant, who took part on a voluntary basis and freely provided information to complete a comprehensive epidemiological and clinical questionnaire. The next of kin, caretaker, or guardians provided oral consent on behalf of participating minors/children enrolled in the study, and this was recorded on the questionnaire as well. Participants received no economic compensation for their role in the study. All the data were treated confidentially and anonymized.

Availability of data and material

The data sets used and/or analyzed during this study are available from the corresponding author on reasonable request.

Results

Among the 394 patient samples studied, 13 (3.3%, 95% confidence interval [CI]: 1.8–5.7) were positive for arthropod-borne disease. We isolated three cases each of Borrelia spp., the Francisella group (F. turalensis tularensis, F. tularensis holartica, and F. novicia), Rickettsia bellii, and R. felis (0.8%; 95% CI: 0.02%–2.4%), and one case of Bartonella rochalimae (0.3%; 95% CI: 0.01%–1.6%). The sequence of the positive Bartonella was 100% identical to B. rochalimae (GenBank accession no. MK693114). The sequence of R. bellii was 98% identical for two patients and 97% identical for one patient (GenBank accession nos. MK693110 and MK693111, respectively). The sequence of R. felis was 100% identical for two patients (GenBank accession nos. MK693112 and MK693113).

Table 1 gives the epidemiological, clinical, and analytical characteristics of patients with and without positive amplification of DNA in blood. In patients with amplification of DNA, the following symptoms are significantly more common: generalized pain (100% vs. 18.6%; p < 0.001), arthralgia–myalgia (69.2% vs. 18.4%; p < 0.001), skin rash (38.5% vs. 3.9%; p < 0.001), headache (92.3% vs. 51.4%; p = 0.004), vomiting (30.8% vs. 7.1%; p = 0.014), and sweating (38.5% vs. 13.6%; p = 0.027).

Table 1.

Epidemiological, Clinical, and Analytical Characteristics of Patients

	DNA amplification
	Negative (n = 381*)	Positive (n = 13†)	p
Age
Median (IQR)	21 (10–35)	35 (23.5–39)	0.083
Range	2–70	2–50
<15 years old	194 (66.2)	9 (81.8)	0.348
Gender, female	208 (54.6)	9 (69.2)	0.297
Symptoms
Headache	196 (51.4)	12 (92.3)	0.004
Nausea	113 (29.7)	3 (23.1)	0.763
Shivers	35 (9.2)	3 (32.1)	0.120
Generalized pain	71 (18.6)	13 (100)	<0.001
Chest pain	48 (12.6)	2 (15.4)	0.674
Abdominal pain	149 (39.1)	5 (38.5)	0.963
Stomach burning	70 (18.4)	4 (30.8)	0.260
Vomiting	27 (7.1)	4 (30.8)	0.014
Difficulty of moving	21 (5.5)	2 (15.4)	0.172
Dyspnea	6 (1.6)	1 (7.7)	0.211
Sweating	52 (13.6)	5 (38.5)	0.027
Coughing	123 (32.3)	7 (53.8)	0.133
Arthralgia–myalgia	70 (18.4)	9 (69.2)	<0.001
Diarrhea	7 (12.3)	2 (15.49	0.669
Astenia	70 (18.4)	1 (7.7)	0.478
Skin rash	15 (3.9)	5 (38.5)	<0.001
Signs
Hepatomegaly	5 (1.3)	0 (0.0)	1.0
Splenomegaly	2 (0.5)	0 (0.0)	1.0
Laboratory
Hb, median (IQR)	13 (11.4–14.3)	12.4 (11.5–13.7)	0.615
WBC, median (IQR)	7.4 (5.7–10.4)	7.9 (6.1–9.0)	0.675
Outcome
Hospitalization	51 (13.4)	1 (7.7)	1.0

All values are presented as n (%) unless stated otherwise.

^*Missing values for participants testing negative: age, n = 88; hemoglobin, n = 44; white blood count, n = 110.

^†Missing values for participants testing positive: age, n = 2; hemoglobin, n = 1; white blood count, n = 7.

Hb, hemoglobin; IQR, interquartile range; WBC, white blood cells.

The presence of DNA from Borrelia spp. was statistically associated with generalized pain (100%; p = 0.007) and arthralgia–myalgia (100%; p = 0.007). Likewise, amplification of R. bellii was significantly associated with generalized pain (100%; p = 0.007), arthralgia–myalgia (100%; p = 0.007), and skin rash (66.7%; p = 0.05). Positive PCR of R. felis was associated with generalized pain (100%; p = 0.007) and difficulty moving (66.7% vs. 5.5%; p = 0.010). And finally, positive PCR results for the Francisella group were statistically associated with generalized pain (100%; p = 0.007), shivers (66.7% vs. 9.2%; p = 0.026), chest pain (66.7% vs. 12.6%; p = 0.046), and sweating (66.7% vs. 13.6%; p = 0.05) (Table 2).

Table 2.

Symptoms Showing Significant Differences with Group Testing Negative for Bacterial Infection, by Bacterium of Infection

	Bacterium
Symptom	Borrelia spp. (n = 3)	p*	Rickettsia bellii (n = 3)	p*	Rickettsia felis (n = 3)	p*	Francisella group (n = 3)	p*
Shivers	1 (33.3)	0.256	0 (0.0)	1.0	0 (0.0)	1.0	2 (66.7)	0.026
Generalized pain	3 (100)	0.007	3 (100)	0.007	3 (100)	0.007	3 (100)	0.007
Chest pain	0 (0.0)	1.0	0 (0.0)	1.0	0 (0.0)	1.0	2 (66.7)	0.046
Difficulty moving	0 (0.0)	1.0	0 (0.0)	1.0	2 (66.7)	0.010	0 (0.0)	1.0
Dyspnea	0 (0.0)	1.0	0 (0.0)	1.0	1 (33.3)	0.054	0 (0.0)	1.0
Sweating	1 (33.3)	0.361	1 (33.3)	0.361	2 (66.7)	0.053	1 (33.3)	0.361
Arthralgia–myalgia	3 (100)	0.007	3 (100)	0.007	1 (33.3)	0.459	2 (66.7)	0.092
Skin rash	1 (33.3)	0.120	2 (66.7)	0.050	1 (33.3)	0.120	1 (33.3)	0.120

^*Comparison with samples negative for DNA amplification.

Italic values p < 0.05.

Discussion

In our study, we found DNA from Bartonella, Borrelia, Francisella, and Rickettsia in malaria-negative febrile patients attending a rural hospital of Ethiopia. These bacteria are transmitted by arthropod vectors such as lice, fleas, and ticks.

In Ethiopia, Bartonella has been found in body and head lice and reported as an infrequent cause of endocarditis in children (Cutler et al. 2012, Tasher et al. 2017). In our series, we found B. rochalimae in an outpatient with unknown clinical evolution. To our knowledge, this species has only been described in human clinical infection in one patient returning to the United States from Peru, after receiving multiple insect bites and showing fever, myalgia, nausea, insomnia, anemia, and splenomegaly (Eremeeva et al. 2007). B. rochalimae has also been detected in dogs, wild carnivores, rats, and fleas in different parts of the Americas and Europe (Chomel et al. 2009, Diniz et al. 2013).

Louse-borne relapsing fever is responsible for 2.5%–4.5% of febrile episodes in some areas of Ethiopia, and outbreaks incur a severe impact, with mortality rates of 30%–70% (Yimer et al. 2014a, 2014b, Hoch et al. 2015). The organisms belonging to Borrelia spp. and identified by 16S rRNA sequencing were associated with febrile episodes, generalized pain, and myalgia. However, there were no hemorrhagic symptoms, contrary to classical clinical manifestations described for this microorganism (Yimer et al. 2014a, 2014b, Hoch et al. 2015).

We found two patients infected by F. tularensis holarctica and one patient with a different species of Francisella, whose identification was totally compatible with F. novicida and F. hispaniensis. F. hispaniensis was first isolated from a human foot wound in Australia in 2003 (Sjodin et al. 2012). Molecular detection of the Francisella group was associated with generalized pain, shivers, chest pain, and sweating. In Kenya, Njeru et al. (2017) found antibodies to F. tularensis in 3.7% of their 730 febrile patients. In Sudan, a case report described bacteremia due to F. turalensis (Kugeler et al. 2008, Mohamed et al. 2012). In Ethiopia, Francisella-like endosymbionts and other atypical Francisella have been described in the Hyalomma rufipes tick and from human fluids, respectively (Kugeler et al. 2008, Szigeti et al. 2014).

Rickettsia felis has also been reported to be a cause of fever in sub-Saharan Africa, but this association has not been dealt with in depth in Ethiopia (Brown and Macaluso 2016). We found R. felis in <1% of the samples, a lower prevalence compared with other studies in Africa, which report a prevalence ranging from 1.5% to 10% of the NMFI (Socolovschi et al. 2010, Mourembou et al. 2015, Sothmann et al. 2017). In our study, R. felis was associated with generalized pain and headache. We also found three cases of Rickettsia bellii, a microorganism infecting argasid and ixodid tick species, aphids, and whiteflies throughout North and South America, although it has never been identified as a human pathogen (Hecht et al. 2016, Krawczak et al. 2018). There is little research on this microorganism in Africa, although Aarsland et al. (2012) did find DNA of Rickettsia spp. in 2.9% of 102 children with fever in Ethiopia.

Although we have identified R. bellii with 98% of sequence in two patients and 97% in the other, and B. rochalimae with sequence of 100% in one case by PCR-reverse line blot assay (RLB), a conventional PCR and sequencing of at least one informative standard amplicon for each genus-specific were performed (Jado et al. 2006, Garcia-Esteban et al. 2008), this finding should be investigated in other studies and in other places in Africa.

Arboviruses are another important source of NMFI. Although there is no literature on coinfection between arbovirus and NMFI, this cannot be excluded. Further studies analyzing arbovirus transmitted by arthropods in NMFI in Africa are warranted.

This study has two main limitations. The first is the absence of a control group of patients without fever, which would allow us to know whether the presence of DNA for these pathogens actually represents a symptomless infection. The use of controls could uncover carriers of well-known pathogens, so it is not easy to interpret data about the potential pathogenic role of these microorganisms. Findings of these arthropod-borne bacteria in febrile versus afebrile people are not well characterized. The second important limitation is the poor information about patient outcomes, which is evidence of the difficulty of collecting data in these rural settings.

Our results showed that nucleic acid amplification tests are highly sensitive and can potentially provide more accurate information on the epidemiology of arthropod-borne pathogens in sub-Saharan Africa, where the lack of molecular tools limits the management of all febrile illnesses in these areas. It also suggests that previously ignored organisms, such as Rickettsia, Bartonella, Francisella, and Borrelia, should be considered in empiric therapies. These tests could allow a shift in practice from empiric abuse of malarial and antibiotic treatment to more informed decision-making on treatment.

Consent for Publication

The publication of the article was accepted by all authors and by the responsible authorities where the work was carried out, and it will not be published elsewhere in the same form, in English or in any other language, including electronically without the written consent of the copyright holder.

Authors' Contributions

J.M.R., R.P.T., M.G., and I.J. were responsible for conception and design of the study. R.P.T., I.M.M., L.P.P., A.T., and G.T. collected the material and data. R.P.T., I.M.M., G.T., R.E., J.G.Z., H.G.G., and I.J. were responsible for the laboratory procedures. J.M.R., R.P.T., L.P.P., M.G., and I.J. were responsible for the logistics and organization of the network. J.M.R., R.P.T., and I.J. interpreted the data. J.M.R. and R.P.T. wrote the main part of the article. All authors read and approved the final version of the article.

Authors' Disclosure Statement

The authors declare that they have no conflict of interests.