Background

Bartonella quintana is a vector-borne, intracellular, Gram-negative bacillus [1]. The bacterium has a tropism for erythrocytes and endothelial cells, causing clinical disease in the form of infective endocarditis, bacillary angiomatosis, and chronic bacteremia [2,3,4,5]. Because of B. quintana’s intracellular localization and sluggish replication, the pathogen is not identified by blood culture with routine 5-day incubation [6]. Bartonella quintana is thus described as a major cause of culture-negative endocarditis and predominantly requires molecular techniques for species-level identification [6, 7].

Bartonella quintana was discovered in 1915 as the cause of trench fever, a relapsing febrile illness afflicting 1 million soldiers during World War I (WWI) [8, 9]. Soon after the bacterium’s detection, the War Office Committee on Trench Fever determined in 1917 that B. quintana was transmitted via body lice: “A small quantity of the excreta of infected lice rubbed into scratches will almost invariably reproduce the disease in a healthy man” [10]. Experimental infections of laboratory-raised body lice fed with blood from patients with trench fever produced “rickettsial bodies” in the lice [10]. As described in these early experiments, transmission of B. quintana entails the inoculation of B. quintana-infected body louse feces into skin abrasions and mucous membranes [1, 2]. Once in the human host, the bacillus infects erythrocytes, causing chronic bacteremia [3, 11]. The longest documented duration of B. quintana bacteremia is 8 years, but most recent studies describe periods of up to 1 year [3, 11]. As B. quintana predominantly affects individuals with body louse infestation (pediculosis corporis), the bacterial infection is associated with poverty, overcrowding, and barriers to maintaining personal hygiene such as insufficient access to running water [12,13,14]. Contemporary outbreaks of B. quintana infection have occurred among populations experiencing homelessness, refugees in low-income countries, and Canadian Indigenous communities with limited access to adequate housing and running water [1, 13].

For over 100 years, dogma has maintained that the human body louse (Pediculus humanus humanus) is the arthropod vector for B. quintana despite increasing reports of B. quintana DNA detection from human head lice (P. humanus capitis) and other arthropods, including cat fleas, pigeon mites, bedbugs, and ticks of various species [2, 15,16,17,18,19,20]. While head and body lice belong to the same species (P. humanus) and are morphologically identical, they belong to two separate ecotypes, inhabiting different ecological niches [21]. Head lice live on head hair, and body lice live in clothing seams [21]. Some experts refer to the latter more accurately as clothing lice [21]. Both ecotypes of lice feed intermittently on human blood [21]. Beyond B. quintana, body lice transmit epidemic typhus (Rickettsia prowazekii) and louse-borne relapsing fever (Borrelia recurrentis) [1]. Head lice are not known to transmit pathogens and are thus not viewed as a major health hazard [21, 22].

For non-Pediculus arthropods, individual studies of macaque lice, bedbugs, pigeon mites, ticks, and fleas have detected B. quintana DNA using molecular methods, but no studies have exhaustively documented the detection of B. quintana among different arthropods and analyzed the results according to arthropod species, region, and host characteristics [15, 19, 23, 24].

This systematic review aimed to describe B. quintana detection among different arthropod species and their hosts.

Methods

Systematic literature search strategy

We searched databases in PubMed Central/MEDLINE, Scopus, Embase, and Web of Science from January 1, 1915 (the year of B. quintana discovery) to January 1, 2024, to identify publications containing specific search terms relating to B. quintana detection among arthropods. We searched for titles and abstracts using the following search string, with associated Medical Subject Headings (MeSH) terms and Boolean operators for each database: {(Bartonella quintana OR Rochalimaea quintana OR Rickettsia quintana OR Trench fever) AND (Arthropod OR Insect OR Vector OR Ectoparasite OR Lice OR Flea OR Mite OR Fly OR Bedbug OR Tick)}. Moreover, we searched reference lists of selected publications to identify other articles. No language restrictions were placed, though search terms were run in English. This review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for systematic literature reviews and was registered in the International Prospective Register of Systematic Reviews (PROSPERO; identifier CRD42024503951) [25, 26].

Study selection

Laboratory confirmation of B. quintana at the species level was required for inclusion (Additional file 1). Studies that tested arthropods for B. quintana but lacked arthropod identification at the genus level were excluded. In vitro studies where arthropods were raised in a laboratory (not collected in the environment) were excluded, as were review articles describing previously published data. There was no spatial limitation.

Article review

Article titles and abstracts were screened by two individuals (CB, NG) to determine eligibility for full-text review. Full texts of the articles included after title/abstract screening were reviewed by two independent reviewers (CB, NG). Reviewer discrepancies were resolved mutually and by discussion with a third reviewer.

Quality assessment for included studies

Two reviewers (CB, NG) assessed articles for quality using the modified JBI critical appraisal checklist for methodological quality and potential bias (Additional file 1) [27]. Studies that failed to meet modified JBI criteria were excluded from the primary analysis.

Data extraction

Data were manually extracted from the included articles by one author (CB) using Microsoft Excel (2019, version 16.72) and corroborated by a second author (NG) who signaled inconsistencies, which were resolved through discussion. For each included reference, we extracted the following:

  • Data relating to the publication: last name of study’s first author, year of publication, country, and continent where data were acquired.

  • Data relating to arthropod analysis: arthropod genus and species (and louse ecotype and clade), method of arthropod identification, co-infestation of body lice with head lice (for studies that analyzed both body and head lice), total number of arthropods tested, number of body and head lice, and number of arthropod pools (for studies that pooled arthropods).

  • Data relating to B. quintana detection: number and percentage of arthropods with B. quintana detection, method of B. quintana identification, and associated molecular targets.

  • Data relating to co-pathogens: results of testing of other pathogens from arthropod samples.

  • Data relating to host: host species, number of hosts, host age category, and key population (e.g., homelessness, for human studies), number and percentage of hosts with B. quintana-infected arthropods, host bacteremia/presence of B. quintana DNA in host blood, and evidence of host clinical disease through symptom reporting.

Duplicate entries were prevented by consolidating identical data summarized in different articles into one record, prioritizing the first publication. Studies that reported data from multiple countries were divided by country to facilitate geographical analysis. Whenever applicable, data analyses were separated by studies that tested individual arthropods versus those that tested arthropod pools.

Statistical analysis

Descriptive statistics and meta-analysis of prevalence using random-effects models were performed post hoc using R version 4.2.2 software (2022-10-31). The random-effects model included an inverse variance method with a restricted maximum-likelihood estimator, logit transformation, a normal approximation confidence interval (CI) for individual studies, and a continuity correction of 0.5 for studies with zero cell frequencies. The global and continent-specific pooled prevalence of B. quintana detection among arthropods was performed as well as the pooled prevalence for body and head lice. Heterogeneity among eligible articles was performed using Cochran’s Q statistic (P-value < 0.10 for statistical significance) and I2 index. The Chi-square test with Yates correction was used to analyze categorical variables, specifically to compare B. quintana detection between head and body lice (P ≤ 0.05 considered statistically significant).

Results

Search results

We identified 1265 articles through the database search (Fig. 1). After 701 duplicate articles were removed, 564 articles remained for title and abstract screening. After review of titles/abstracts and full texts, 62 publications met inclusion criteria, describing 15,493 arthropods tested for B. quintana. Fifty-four articles reported data on 14,088 individual arthropods, and eight reported data on arthropod pools, adding 1405 arthropods [17, 28]. The included articles were published between 1961 and 2023. In some cases, multiple different studies reported data from arthropods collected from the same country [29, 30]. Twenty-one studies tested multiple types of arthropods (e.g., body and head lice, body lice, and non-louse arthropods).

Fig. 1
figure 1

PRISMA 2020 flow diagram for this new systematic review which included searches of databases and registers

Full size image

Host species and characteristics

This review included 8735 hosts, of which 7673 (0.8784, 95% CI [0.8714; 0.8852]) were human and 1062 were non-human. All but eight studies tested arthropods from human hosts. Nineteen studies included children: 12 studies tested arthropods from adults and children, and seven studies exclusively tested children. Seventeen studies describe human hosts as having experienced homelessness [17]. Five studies describe participants as being refugees [31, 32]. Two studies describe hosts as being incarcerated and two as coming from a rural Indigenous community [33]. Regarding non-human hosts, two studies tested arthropods from cats, two from non-human primates (one from rhesus macaques and one from Cercopithecus cephus monkeys), two from rodents, one from boars, and one from birds. Two hundred ninety-three hosts were infested with arthropods with evidence of B. quintana infection, of which all but four hosts were human.

Arthropod and pathogen identification

Arthropods were molecularly identified using polymerase chain reaction (PCR) in 46 of 62 studies (74.19%). Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry was used in one study, and the remaining 15 studies identified arthropods morphologically using taxonomic keys. Twenty-seven studies ascertained louse clade using molecular analysis of the mitochondrial cytochrome b (cytB) gene. Clade A was identified in all but three studies, and 10 studies identified clade A in addition to at least one other.

PCR was used to determine the presence of pathogens, including B. quintana, in all but one study. The most common PCR targets for B. quintana were the internal transcribed spacer gene (ITS, often used as an initial screening for Bartonella genus), the putative targeted effector protein gene (yopP), the citrate synthase gene (gltA), and the 3-oxoacyl-[acyl-carrier-protein] synthase gene (fabB). Thirty-seven studies used multiple molecular targets to confirm B. quintana. Twenty-four studies combined ITS for Bartonella genus with a second target for B. quintana species. Bartonella quintana was identified using culture from arthropod samples in one French study that identified bacterial colonies with 16S ribosomal RNA (rRNA) sequencing [34].

Geographical distribution

Arthropods were collected from 37 countries, of which 28 had arthropods with evidence of B. quintana infection (Fig. 2). Arthropods with B. quintana were reported from every continent except Oceania and Antarctica (Table 1). Arthropods from nine countries (Australia, South Korea, Thailand, Malaysia, Italy, Tanzania, Guinea, St. Kitts, CorsicaFootnote 1) were tested for B. quintana without reported detection (Additional file 2). Among articles that reported individual arthropods, B. quintana was detected among 1445 of 14,088 (0.1026, 95% CI [0.0976; 0.1077]) arthropods, generating a random-effects model global prevalence of 0.0666 (95% CI [0.0426; 0.1026]). High heterogeneity was found among studies (I2 = 93.3% [92.1%; 94.3%], Cochran’s Q, P < 0.0001) (Additional file 3) [35]. Africa had the most country-specific reports, with 43 reports describing 7040 arthropods. Of these, 703 (0.1000, 95% CI [0.0929; 0.1071]) tested positive for the presence of B. quintana DNA, representing a random-effects model prevalence of 0.0615 (95% CI [0.0311; 0.1180]), with elevated heterogeneity (I2 = 94.6% [93.5%; 95.5%], Cochran’s Q, P < 0.0001). The continent with the fewest publications was Oceania, which had only two reports describing five arthropods and no arthropods with B. quintana DNA.

Fig. 2
figure 2

Map of countries reporting arthropods testing positive for B. quintana. Interactive map with data linked to original publication (author, year), arthropod species/ecotype, and positivity for B. quintana is available here: https://www.google.com/maps/d/u/0/edit?mid=1NRbIN72IDnbgxjdytjWVycd7_LmJx9k&usp=sharing

Full size image
Table 1 Global and regional arthropod prevalence for Bartonella quintana
Full size table

Arthropod genera, species, and ecotypes and associated B. quintana detection

Fifty-six studies tested 8839 body lice, of which 1679 harbored B. quintana DNA (0.1899, 95% CI [0.1818; 0.1983]) (Table 2), generating a random-effects model pooled prevalence of 0.2312 (95% CI [0.1784; 0.2843]), with high heterogeneity (I2 = 98.7%, Cochran’s Q, P < 0.0001). Forty-two studies tested 4962 head lice, of which 390 head lice from 20 studies originating from 11 different countries harbored B. quintana DNA (0.0786, 95% CI [0.0713; 0.0864]). This generated a random-effects model pooled prevalence for head lice of 0.0301 (95% CI [0.0221; 0.0372]), again with high heterogeneity (I2 = 90.09%, Cochran’s Q, P < 0.001). Fourteen studies tested both body and head lice, of which nine confirmed louse ecotype using PCR (Additional file 4), and five documented greater detection among head lice than body lice (Table 3) [17, 28, 30, 36, 37]. Co-infestation of body lice with head lice was documented in eight studies among a minority of hosts (Appendix 4).

Table 2 Arthropod type, number of studies, and B. quintana detection
Full size table
Table 3 Studies where B. quintana detection in head lice exceeded that of body lice
Full size table

Overall, the presence of B. quintana DNA was greater among body lice than head lice (χ2 = 308.3, df = 1, two-tailed P < 0.0001). However, nine studies that exclusively tested head lice detected B. quintana DNA (Table 4). All nine of these studies collected arthropods from low-income contexts: seven were from low- and middle-income countries (LMICs), and of the two from high-income countries, one was from rural Georgia, USA (an area known for poverty), and one from an individual experiencing homelessness in France (Table 4) [18, 38]. Head lice collected from schoolchildren in high-income countries such as France, Portugal, and Australia had no evidence of B. quintana DNA (Additional file 5).

Table 4 Studies that exclusively tested head lice and reported B. quintana detection
Full size table

Regarding non-Pediculus arthropods, one of the two studies that tested 67 pubic lice (Pthirus pubis) reported the presence of B. quintana DNA among two lice from one individual. One study tested macaque lice (Pedicinus obtusus) from 10 non-human primates; arthropods were separated into two pools per monkey, and B. quintana was identified in all lice pools [39]. Two studies tested bedbugs (Cimidae); B. quintana DNA was detected in one arthropod in one instance [40]. Nine studies tested 820 fleas (Siphonaptera) of various species. While B. quintana DNA was not detected in dog fleas, three human fleas (Pulex irritans), 14 cat fleas (Ctenocephalides felis), and nine rodent fleas (Xenopsylla cheopis) tested positive for B. quintana using molecular methods, with associated proportions of 0.075 (95% CI [0.0157; 0.2039]), 0.0729 (95% CI [0.0404; 0.1193]), and 0.0154 (95% CI [0.0071; 0.0291]), respectively. Three studies tested ticks, with two testing hard ticks (multiple species) and one testing soft ticks (Orthodoros sawaii); none were associated with the presence of B. quintana DNA. Two studies tested mites, one testing Demodex species and the other testing Dermanyssus species, with B. quintana detected molecularly in three and seven mites, respectively [15, 41].

Bartonella quintana-related disease among hosts

Host clinical disease and the presence of B. quintana DNA in blood samples were reported in seven studies. Clinical disease was most commonly described as fever [3]. All hosts with B. quintana DNA in blood samples were human. Five studies described infestations with B. quintana DNA-positive body lice, two studies described infestation with B. quintana DNA-positive head lice without body lice co-infestation, and one described infestation with B. quintana DNA-positive Dermanyssus mites [15, 16, 42,43,44].

Co-infection with other pathogens

Among the 31 studies that tested arthropods for pathogens other than B. quintana, various Acinetobacter species were most commonly identified in 18 studies. Despite their shared transmission with body lice, R. prowazekii and B. recurrentis were rarely detected in arthropods. Rickettsia prowazekii was identified in only three studies testing arthropods from refugees in Turkey and Burundi and homeless populations in Colombia [29, 32, 45]. Borrelia recurrentis was identified in only one study of head lice from the Republic of the Congo [46].

Quality assessment

Quality assessment using the modified JBI critical appraisal checklist for cross-sectional studies revealed that 62 publications had sufficient information to be included in the full-text analysis (Additional file 6) [27]. Thirteen publications were excluded predominantly due to insufficient diagnostic information.

Discussion

The detection of B. quintana DNA among arthropods in 37 countries across most continents suggests that B. quintana is a vector-borne disease with a global distribution. The elevated number of B. quintana DNA-positive arthropods from certain African and Asian countries suggests a considerable burden in LMICs [30, 37]. While B. quintana is often described as a rare pathogen, our description of B. quintana among 1445 arthropods from 293 hosts indicates that B. quintana may be more common than previously believed. However, additional data are needed to fully elucidate the epidemiology of B. quintana. This review highlights how entomologic studies may provide a non-invasive method of facilitating B. quintana surveillance among key populations, such as those with a history of homelessness, incarceration, or forced displacement/immigration from certain LMICs. Although data were reported according to country, B. quintana detection among arthropods is associated with poverty: schoolchildren with head lice from high-income environments are unlikely to harbor B. quintana and wealthy adults from LMICs are unlikely to have pediculosis. Entomologic surveillance of B. quintana would have the highest yield when applied only to specific key populations. Certain countries, such as Brazil, Colombia, Madagascar, Mali, Nigeria, and Peru, reported B. quintana DNA-positive arthropods but have no published cases of B. quintana endocarditis, indicating that arthropod studies may identify a hidden burden of disease in LMICs [13, 31, 43, 45, 47, 48]. Many countries had a limited number of studies or no published data, suggesting that arthropod testing for B. quintana is an underutilized surveillance tool [41, 49].

While this review summarized B. quintana detection among many different arthropod species, not all of these arthropods meet criteria to be considered vectors for B. quintana. To be considered a vector, transmission between hosts needs to be demonstrated [50]. Detection of B. quintana DNA is insufficient to deem an arthropod a vector, as the arthropod may have acquired B. quintana through a blood meal but remain incapable of transmitting the bacterium to a new host. This “dead-end” acquisition of B. quintana may be enhanced by the known chronicity of B. quintana bacteremia [1, 3]. Transmission of B. quintana from body lice to human hosts involves the inoculation of infected body louse feces into skin abrasions and mucous membranes [1, 51]. Arthropods that do not defecate at the time of feeding and do not live on or close to the host are unlikely to transmit B. quintana. These arthropod species may be more likely to demonstrate dead-end DNA acquisition without vector competence. Furthermore, B. quintana transmission requires bacterial multiplication in the arthropod gut. Arthropods that consume and digest a blood meal with B. quintana may have fragments of B. quintana identified by PCR without any live bacteria, precluding transmission to the next host.

While WWI xenodiagnostic studies have long established B. quintana transmission via body lice, transmission via head lice has been demonstrated more recently and less frequently [16]. Only two studies describe the presence of B. quintana DNA in human blood samples linked to head lice infestation [16, 42]. In one recent study describing a Senegalese outbreak, B. quintana transmission via head lice (without concomitant body lice infestation) was supported by genomic analyses, with 99.98% similarity between B. quintana acquired from blood and head lice samples [16]. For non-Pediculus arthropods, only a single study of macaque lice (P. obtusus) and of Dermanyssus pigeon mites have shown a link to host infection [15, 39]. However, these findings have not been replicated, and additional data are needed to explore whether arthropods beyond P. humanus may transmit B. quintana [15, 39] While bedbugs (Cimidae) have demonstrated vector competence for B. quintana in vitro, no studies have demonstrated B. quintana transmission in vivo [23].

Despite the infrequency of proven transmission via head lice, the presence of B. quintana DNA among head lice is not rare, but appears to depend on the socioeconomic context [17, 37, 38]. While all studies of head lice collected from schoolchildren in high-resource countries are B. quintana-negative [29], B. quintana DNA may be detected among head lice from adults and children in low-resource settings [28, 30, 33, 37]. The reason that head lice among schoolchildren in high-income contexts test negative for B. quintana is unknown but may relate less to the absence of vector competence and more to epidemiology: if not belonging to the key populations mentioned above, schoolchildren in high-income contexts are unlikely to be in close contact with B. quintana-bacteremic individuals. In high-income countries, B. quintana among head lice is limited to low-resource contexts such as urban homelessness or rural impoverishment [17, 18, 47]. In the USA, B. quintana DNA has been detected among head lice collected from homeless populations in San Francisco and housed populations in Georgia, an area known for rural poverty [17, 47].

The presence of B. quintana in head lice may not require co-infestation of body lice with head lice in a single individual. Among the 13 studies that analyzed both body and head lice, co-infestation of body and head lice was documented in eight studies with rates as low as 3% among a homeless population in the USA [17]. The presence of B. quintana in head lice from different environments suggests that future introductions into high-income populations of school-age children may be possible should these children share close contact with individuals with ongoing B. quintana bacteremia and pediculosis.

While R. prowazekii and B. recurrentis are frequently described as louse-borne, very few studies have detected DNA of these pathogens, indicating that B. quintana may be the most common louse-borne disease [29, 32, 46, 45]. Certain high-income jurisdictions have described imported cases of B. recurrentis but have not reported cases of B. quintana [13, 52]. This relative predominance of B. recurrentis may reflect diagnostic bias: B. recurrentis is visible on Giemsa stains performed for malaria while B. quintana is not [52]. The preponderance of B. quintana in arthropods implies that any cases of B. recurrentis and R. prowazekii should also be tested for B. quintana infection.

This systematic review is subject to several limitations. All included articles necessitated confirmation of B. quintana to species level, which inherently created a bias towards recent studies that used molecular diagnostics. Significant heterogeneity was found between studies. The data may have been influenced by publication bias and other forms of bias in the original studies. It is possible that some head lice studies included patients with undocumented concomitant or previous body lice infestation. Heterogeneity in study methodology prevented the inclusion of a minority of articles in the random-effects models of prevalence. Statistical analyses were applied post hoc, increasing the risk of false discovery.

Conclusions

This systematic review reveals that B. quintana is a louse-borne disease with a global distribution and a disproportionate burden in low-resource settings. While less commonly infected than body lice, a limited number of studies suggest that head lice may also act as vectors for B. quintana in specific low-resource contexts. Prospective studies that simultaneously test arthropods for B. quintana carriage and human hosts for B. quintana infection/disease are needed to elucidate transmission rates and proportions of severe illness, such as endocarditis. These studies are necessary to determine whether individuals with B. quintana DNA-positive ectoparasites should be treated for B. quintana infection, especially in areas where molecular testing and echocardiography are difficult to obtain. We encourage public health laboratories to include B. quintana testing of body lice (and head lice originating from low-resource contexts) as a surveillance modality to improve our understanding of the epidemiology of this neglected disease.