Matthew P. Rubach

Background: Cryptococcus causes life-threatening infections; however, immune programs driving the pathogenesis of these infections in the human host remain incompletely described. We leveraged whole-blood transcriptomic analysis to define distinct, conserved elements of the host immune response in participants with cryptococcal infections. Methods: RNA sequencing was performed on whole blood from adult febrile inpatients admitted to 2 hospitals in Moshi, Tanzania, to compare gene expression between participants with cryptococcal, bacterial, or viral diseases, and adult community controls. Differential expression and gene set enrichment analyses were used to characterize expression differences between pathogen-class groups. Penalized logistic regression was used to identify gene expression signatures capable of identifying Cryptococcus disease. Results: Out of 118 participants, 13 (11%) had cryptococcal disease, and 41 (34.7%) were persons with HIV (PWH). Genes upregulated in response to cryptococcosis compared to nonacutely infected community controls reflected complement activation, neutrophil activation, type I interferon signaling, and myeloid dendritic cell differentiation. When compared to combined bacterial or viral illnesses, cryptococcal disease demonstrated hyperactivation of both innate and humoral immune pathways. A series of classifiers were developed that distinguished between cryptococcal and combined bacterial or viral diseases with an AUC of 0.946. Conclusions: Overall, the host response to cryptococcosis in individuals with advanced HIV, as manifested in transcriptional profiles, informs our understanding of the pathophysiology of and susceptibility to cryptococcal infection and may support the development of novel diagnostic and therapeutic approaches.

Background: Febrile illness is a common reason for hospital admission in sub-Saharan Africa, yet the illness etiology often remains unknown. To address these knowledge gaps, we conducted respiratory pathogen testing in a cohort of febrile inpatients in northern Tanzania. Methods: We enrolled a prospective cohort of febrile pediatric and adult inpatients within 24 hours after admission to 2 hospitals in northern Tanzania. Nasopharyngeal/oropharyngeal specimens were collected for detection of 19 pathogens with polymerase chain reaction. Participants were stratified by age (<10 or and ≥10 years) and categorized into the following 3 groups: severe acute respiratory infection (SARI), non-SARI respiratory illness, or no respiratory illness. Results: Among the 1132 participants enrolled, 930 (82.1%) had nasopharyngeal/oropharyngeal specimens tested. Their median (interquartile range) age was 22.3 (2.1–44.0) years, 432 (46.5%) were female, and 378 (40.7%) had SARIs. We observed 410 pathogen detections in 359 participants (38.6%), in 256 (67.2%) of 381 in the <10-year age group and 103 (18.8%) of 549 in the older group (≥10 years old). Of the 301 pathogen detections among participants <10 years old, the most common were rhinovirus (RV)/enterovirus (EV) (n = 80 [26.6%]), influenza (n = 60 [19.9%]) and adenovirus (n = 45 [15.0%]). Of the 109 pathogen detections in participants ≥10 years old, the most common were RV/EV (n = 41 [37.6%]), influenza (n = 37 [33.9%]), and respiratory syncytial virus (RSV) (n = 6 [5.5%]). Among SARIs, the most common pathogens were influenza (n = 71 [18.8%]), RV/EV (n = 66 [17.5%]), and RSV (n = 37 [9.8%]). Conclusions: RV/EV, influenza and RSV were commonly detected among febrile inpatients in Tanzania, including in participants with SARIs. These findings have implications for vaccine development and implementation in low- and middle-income countries.

Background: Timely and accurate diagnosis of spotted fever rickettsioses (SFR) is difficult due to nonspecific symptoms and testing challenges. We evaluated serum antibodies to the tick salivary proteins AV422 and calreticulin as potential markers of recent tick exposure in febrile patients from northern Tanzania. Methods: We recruited febrile patients at 2 hospitals in Moshi, Tanzania, from February 2012 through May 2014. Acute and convalescent sera from enrolled participants were tested against Rickettsia africae antigen using indirect immunofluorescence assays (IFAs). Participants with ≥4-fold rise in R africae IFA immunoglobulin G reciprocal titer between sera were classified as cases (n = 70), those with paired sera with reciprocal titer ≥64 but no ≥4-fold rise as exposed (n = 40), and participants with reciprocal titer <32 in either sera as unexposed (n = 38). Acute and convalescent sera were tested for calreticulin and AV422 antibodies using enzyme-linked immunosorbent assay. Results: We found significantly higher calreticulin antibody levels in acute sera of cases compared to exposed, unexposed, and the combined comparator of exposed and unexposed (adjusted mean difference [95% confidence interval {CI}], 17.49 [6.37–28.61]; 17.61 [6.55–28.67]; and 18.11 [9.13–27.10], respectively). Calreticulin distinguished cases from exposed and unexposed among persons 0–15 years old with positive predictive value of 0.77 (95% CI, .61–.88) and sensitivity of 0.94 (95% CI, .81–.99). Conclusions: In the acute phase of febrile illness, measurement of calreticulin antibody, a proposed marker for tick exposure, was higher in patients with confirmed tick-borne illness and had reasonable diagnostic test characteristics.

Background: Extraintestinal pathogenic Escherichia coli (ExPEC) are a leading cause of human bloodstream infections (BSIs) in sub-Saharan Africa, yet few studies have characterized African strains implicated in BSI or explored their potential reservoirs. Methods: We enrolled febrile patients at 2 hospitals in Moshi, Tanzania, 2007–2019, and performed blood cultures. Whole-genome sequencing was conducted on E. coli originating from the bloodstream to characterize sequence types (STs), serotypes, and theoretical coverage of a 9-valent ExPEC polysaccharide conjugate vaccine (ExPEC9V). Separately, we evaluated 601 E. coli whole-genome sequences from humans, animals, and environmental sources in nearby communities. We assessed genetic relatedness between bloodstream and community isolates based on single-nucleotide polymorphisms and allele differences. Findings: Of 3046 participants receiving blood culture, 48 (0.2%) had BSI yielding 48 E. coli isolates. The median (range) age of participants with E. coli BSI was 40.7 (0.3–89.0) years, and 32 (68.1%) were female. We identified 16 STs including ST131 (n = 16, 33.3%), ST73 (n = 10, 20.8%), and ST69 (n = 6, 12.5%) and 19 O groups including O25 (n = 13, 27.1%), O6 (n = 10, 20.3%), O17 (n = 4, 8.3%), and O18 (n = 4, 8.3%). Theoretical coverage for an ExPEC9V was 72.9%. None of the bloodstream and community E. coli pairs were closely related. Conclusions: We found a high diversity of STs among E. coli human bloodstream isolates in Tanzania. Despite this diversity, we observed that an EXPEC9V in development would provide good coverage. Reservoir-attribution studies at finer spatial and temporal scales may better identify transmission networks and reservoirs of ExPECs.

Background: The presence of multiple medical chronic conditions is a growing burden across Africa, where people experience a double burden of communicable and non-communicable diseases (NCD). Integrated chronic disease care at the primary healthcare level is a means to improve person-centred care for people living with multimorbidity. This study aimed to assess public primary healthcare facilities’ service availability and readiness to manage multimorbidity in Malawi, through the lens of HIV, type 2 diabetes, hypertension and chronic kidney disease. We explored healthcare workers’ perceptions of their capacity to provide care for individuals with multimorbidity and the availability of services and resources at their facility. Methods: We conducted a mixed-methods health facility cross-sectional study in primary healthcare facilities in Blantyre and Chiradzulu districts, Malawi. We used the modified WHO Service Availability and Readiness Assessment Tool to assess 12 facilities, and conducted in-depth interviews with 12 purposively selected health facility leads. Availability is presented as the proportion of facilities offering each service. Facility readiness is defined as the mean(percentage) availability of 44 tracer items with a cutoff point of ≥70%, classified as ‘ready’ to manage multimorbidity. Readiness score for individual conditions was also calculated based on the availability of tracer items of that condition with a cutoff point of ≥70%. Results: Of the surveyed facilities, 83%(10/12) offered services for all four selected conditions. Only 30%(3/10) of all facilities had a readiness score of ≥70%, indicative of minimum service delivery requirements for multimorbidity were met. The mean readiness score for type 2 diabetes (65.7%) was the highest, followed by HIV (63.8%), hypertension (60.4%), and chronic kidney disease had the lowest readiness score (22.9%). Barriers to service readiness included nonfunctional diagnostic equipment, frequent drug stockouts and insufficient staffing and supervision, and a lack of clinical guidelines or protocols. Conclusion: The assessed sites were inadequately prepared to provide integrated care for chronic kidney disease, type 2 diabetes, HIV and hypertension. The ability to provide integrated screening and management will require functional diagnostic equipment and consistent drugs. Periodic quality assurance and combined multimorbidity audits could further highlight the gaps in service delivery and give clues to improvement.

Endemic zoonoses have a dual burden, impacting human and animal health, as well as livelihoods due to livestock losses. One such disease is anthrax, caused by Bacillus anthracis . Data on the burden of anthrax on human and animal health are extremely limited, perpetuating neglect at the policy level in endemic areas and internationally. To address these data inadequacies, we quantified the widespread impacts of anthrax on highly affected communities, focusing on a hyper-endemic area of eastern Africa, the Ngorongoro Conservation Area (NCA) of northern Tanzania. Data were collected between 2016 and 2019 through cross-sectional household surveys, active surveillance, compilation of available hospital and human disease surveillance records, and willingness-to-pay choice experiments conducted within wider socioeconomic surveys. Human anthrax was reported by 34 (16%) of 209 households surveyed. Cases of anthrax in livestock were reported in 63 households (30%), and in all 20 households within Ngoile, the most affected area. The majority of sudden deaths in livestock (368 of 500 cases in 30 months of active surveillance) were confirmed to be caused by anthrax using microscopy or qPCR. Monetary losses due to livestock deaths were significant and during outbreaks commonly exceeded average monthly household income. Despite the dual burden, willingness to adopt anthrax mitigation measures was low. Likely due to resource limitations, households would only consider such measures if they had high efficacy, be it for animals or humans, but even then, price was still an important factor. We discuss the potential benefits that anthrax prevention could have on affected communities, and factors that might influence the success of possible interventions. This study demonstrates that in endemic settings where dependence on livestock productivity is high, the impacts of neglected zoonoses can be diverse, persistent and severe and highlights the challenge of identifying locally acceptable mitigation strategies.

: Acute Q fever diagnosis via paired serology is problematic because it requires follow-up for convalescent sample collection; as such, it cannot provide a diagnosis to inform a treatment decision at the time of acute presentation. Real-time polymerase chain reaction (PCR) may be a useful approach for the diagnosis of acute Q fever in endemic settings. Among febrile patients enrolled in a sentinel surveillance study for Q fever at two referral hospitals in Moshi, Tanzania, from 2012 to 2014, we analyzed those with paired sera for IgG to Coxiella burnetii (C. burnetii) phase II antigens using immunofluorescent antibody (IFA) testing, and acute serum was tested for C. burnetii with PCR. Acute Q fever was defined as a fourfold or greater rise from the acute to convalescent sample in IFA reciprocal titer or PCR detection that was confirmed through repeat testing. Test characteristics were tabulated. Among 496 participants tested using both paired IFA and PCR testing, 463 (93.3%) tested negative on both IFA and PCR, five (1.0%) tested positive for Q fever on both IFA and PCR, and 28 (5.6%) tested positive for Q fever on IFA alone. The sensitivity of PCR testing using paired IFA testing as an index was 0.15 (5/33), and the specificity was 1 (463/463). C. burnetii PCR testing provides a clinically specific method that may aid in timely diagnosis in settings in which acute Q fever is a common cause of febrile illness. However, we found a low clinical sensitivity of PCR testing on serum when compared with paired IFA serology.

Background: Crimean‐Congo hemorrhagic fever is a tick‐borne zoonotic disease that may be severe and is present in many African countries. We aimed to understand the seroprevalence and risk for Crimean‐Congo hemorrhagic fever virus in Tanzania by testing archived serum samples from patients enrolled in a prospective cohort study.Methods: We prospectively enrolled febrile inpatients and outpatients from 2012 through 2014 at two referral hospitals in northern Tanzania. Archived serum samples were tested for Crimean‐Congo hemorrhagic fever virus antibodies initially by a Luminex assay screen followed by confirmation with immunofluorescence assay. Evidence of exposure to Crimean‐Congo hemorrhagic fever virus was defined as antibody detection by Luminex and confirmed by immunofluorescence assay. Questionnaire data were used to construct logistic regression models to understand factors associated with prior exposure to Crimean‐Congo hemorrhagic fever virus. Domains of predictor variables included sociodemographics, livestock‐rearing activities, and environmental factors.Results: Of 735 participants included, antibodies to Crimean‐Congo hemorrhagic fever virus nucleocapsid protein were detected by Luminex assay in 23 (3.1%) and confirmed by immunofluorescence assay in 13 (1.8%). In multivariable logistic regression, prior exposure to Crimean‐Congo hemorrhagic fever virus was associated with self‐report of milking livestock in the past month (adjusted OR [aOR]: 12.6, 95% CI 1.6–99.8) and natural log increase in goat density (head/km²; aOR: 1.7, 95% CI: 1.1–2.7).Conclusions: We show serologic evidence of prior exposure to Crimean‐Congo hemorrhagic fever virus among humans in northern Tanzania. Similar to other settings, our results suggest that exposure is closely linked to livestock activities. Additional research is warranted to understand reservoirs and modes of transmission of Crimean‐Congo hemorrhagic fever virus to humans in northern Tanzania.

Background: Acute febrile illness is a common reason for seeking healthcare in low‐ and middle‐income countries. We describe the diagnostic utility of a TaqMan Array Card (TAC) real‐time polymerase chain reaction (PCR) panel for pathogen detection in paediatric and adult inpatients admitted with febrile illness.Methods: In this prospective cohort study, we screened medical admissions for a tympanic temperature ≥38.0°C or reported fever within 72 h and used a PCR panel to detect pathogens, including bacteria, viruses, fungi and protozoa, in 697 participants. We compared PCR results to conventional diagnostic methods and considered PCR detections as the cause of fever, except for Plasmodium spp. and Schistosoma spp. Participants for PCR testing was consecutively selected from the end of enrolment.Results: Of 1132 participants enrolled in the cohort, 697 (61.6%) were tested by PCR. Median (IQR) age was 29.6 (4.6–46.4) years. Three hundred seventy‐eight (54.2%) were male. The PCR method improved illness identification, increasing diagnostic yield from 73 (10.5%) by conventional methods to 124 (17.8%) of 697 participants. PCR detections included four viral pathogens: dengue (n = 1), enterovirus (n = 7), measles (n = 1) and Rift Valley Fever Virus (RVFV) (n = 3). Forty‐six bacterial pathogens were detected in 44 (6.3%) participants, including fastidious bacteria such as Bartonella spp. (n = 2), Brucella spp. (n = 3), Coxiella burnetii (n = 2), Leptospira spp. (n = 1), M. tuberculosis (n = 7) and Rickettsia spp. (n = 9).Conclusion: The PCR panel improved pathogen detection in febrile inpatients, providing clinically actionable results for fastidious bacteria and epidemiologically relevant findings like RVFV detections, when combined with conventional methods.

Background: Despite a high burden of sepsis in Sub-Saharan Africa, clinical data for adolescent and adult sepsis in this setting are limited. We sought to describe clinical characteristics, management, and outcomes in adolescents and adults with sepsis in northern Tanzania. We also assessed for clinical associations with in-hospital mortality. Methods: We conducted a prospective observational cohort study at Kilimanjaro Christian Medical Centre in Moshi, Tanzania, 2019–2020. Data were collected on demographics, clinical characteristics, and management primarily from hours 0–6 after arrival at the emergency department. We calculated bivariable risk ratios (RRs) for associations between demographic and clinical factors and in-hospital death. A multivariable-adjusted analysis was performed for associations between antimicrobial and intravenous fluid administration and in-hospital death. Results: Of 86 participants with sepsis, 25 (29.1%) died in the hospital. Baseline characteristics associated with in-hospital mortality included inability to drink unassisted (RR, 3.15; 95% CI, 1.58–6.30), altered mentation (RR, 3.94; 95% CI, 2.12–7.33), quick Sequential Organ Failure Assessment (qSOFA) score ≥2 (RR, 2.86; 95% CI, 1.42–5.72), and Universal Vital Assessment score ≥5 (RR, 6.33; 95% CI, 2.36–17.02). Twenty-nine (33.7%) received an antimicrobial by hour 6. HIV antibody testing was performed for 4 (4.7%) participants by hour 6. On multivariable analysis, neither antimicrobial administration nor intravenous fluids >1L by hour 6 was associated with in-hospital mortality. Conclusions: Sepsis at our center in northern Tanzania carried a high risk of in-hospital mortality. Further research is needed to establish the highest-yield interventions suited to the unique characteristics of sepsis in Sub-Saharan Africa.