Nasopharyngeal colonisation dynamics of bacterial pathogens in patients with fever in rural Burkina Faso: an observational study

This study was designed as a side project to the PALUBAC study, which was set up to test and optimise an algorithm to distinguish malaria from bacterial infections in acutely ill febrile patients. Inclusion and exclusion criteria of the PALUBAC study and a complete report on which data were collected have been published previously [21‐23]. Briefly, participants were enrolled between March 2016 and June 2017 at the Centre Médical avec Antenne Chirurgicale (CMA) Saint Camille de Nanoro, Burkina Faso. Nanoro is located in the central-western part of the country. It has a rainy season from June to October, with an average rainfall of 450–700 mm per year, and a dry season from November to May [24]. The Nanoro department is a rural area with an average population density of 139/km². The majority (90.13%) of the population lives in rural conditions. The principal activity of the inhabitants is agriculture, mainly subsistence farming. 50.57% of inhabitants are children under 15 years of age [25]. The national immunisation program has included the conjugate H. influenzae type B vaccine since 2006 and the 13-valent conjugate S. pneumoniae vaccine since 2013. Children under 5 years of age receive seasonal malaria chemoprophylaxis from July to October. Participants were eligible for inclusion if they had a fever (temperature ≥ 38.0 °C), hypothermia (temperature ≤ 35.5 °C), a reported history of fever or hypothermia up to 48 h prior to presentation at the health facility or a suspicion of severe infection. Patients under 3 months of age and patients with fever or hypothermia lasting longer than 7 days were excluded. From March until November 2016, only hospitalised patients were enrolled. Thereafter, until the end of the study in June 2017, non-hospitalised patients were also enrolled because of insufficient study recruitment [23].

Basic demographic and clinical data were recorded on a standardized case report form by study nurses. Upon inclusion, a blood sample was taken, as well as a nasopharyngeal swab in UTM medium. The blood sample was used for malaria diagnostics (rapid diagnostic tests, thin and thick film microscopy), blood culture and basic hemocytometry. Cultures of urine, stool, pus or cerebrospinal fluid and imaging diagnostics (chest X-ray or abdominal echography) were performed only when indicated, as was serology for specific infectious diseases (e.g. HIV). Residual blood and the nasopharyngeal swabs were stored at − 20 °C and shipped to the Radboud university medical center in Nijmegen, the Netherlands. There, all blood samples were tested for malaria using quantitative polymerase chain reaction (qPCR). qPCR for Salmonella spp., S. pneumoniae, H. influenzae and S. aureus was performed on blood samples of patients with negative blood cultures. qPCR methods for the detection of bacterial pathogens in blood samples were described in a previous publication [23]. The detection limit was 75 copies of DNA per ml of blood.

For the purpose of data analysis, ‘bacteraemia’ was defined as the presence of a pathogenic bacterium either in blood culture or in blood qPCR. Upon discharge, a final diagnosis made by a local physician was recorded on the case report form. For children below 5 years of age, pneumonia was defined according to the WHO criteria [26]. For older children and adults, pneumonia was defined as patients having fever and one or more signs of dyspnoea (rapid breathing, nasal flaring, chest indrawing, saturation below 95%, subjective feeling of dyspnoea), combined with either an X-ray showing pulmonary infiltrate or chest auscultation suspect for pneumonia. The criteria for a diagnosis of tuberculosis were cough lasting ≥ 14 days and weight loss, in combination with (1) a positive Mycobacterium tuberculosis PCR (GeneXpert® MTB/RIF) on sputum, (2) acid-fast bacteria in Ziehl–Neelsen stain on sputum or (3) typical caseating granulomas on X-ray.

qPCR for S. pneumoniae, H. influenzae, M. catarrhalis, S. aureus and K. pneumoniae was performed on all nasopharyngeal swabs. Stored swab samples were thawed on ice and vortexed. From each sample, 200 µl was aliquoted into a 96-wells plate. The plate was incubated for 15 min at 93 °C to lyse the bacteria [27]. The qPCR was performed in five monoplex reactions, using the Bio-rad CFX96 Touch Real-Time PCR Detection System. All reactions were performed in a 10 µl final volume containing 1 µl bacterial lysate, 5 µl SsoAdvanced™ Universal Probes Supermix (Bio-Rad), 400 nM of each primer and 200 nM probe. For details on primers and probes, see Additional file 1: Sheet 1. Every 96-wells plate contained, in duplicate, a no template control and a seven step tenfold serial dilution of a positive control, starting at approximately 10 ng/µl. The qPCR program consisted of 3 min of incubation at 95 °C followed by 50 cycles of 10 s at 95 °C and 20 s at 60 °C for all targets except for K. pneumoniae, for which the cycles were 10 s at 95 °C and 20 s at 65 °C. Fluorescence was measured after each cycle. In order to accurately compare results within targets, the baseline threshold was adjusted so that the log 3 dilutions of the positive control had the same Cq value per target for all 12 plates. Cq cutoff was determined per target using data on negative controls and samples that were tested in duplicate. The detection limit of the qPCR was calculated to be below ten copies of DNA per µl UTM medium for all targets. All qPCR results can be viewed online (Additional file 2).

Data were analysed using SPSS Statistics 25.0.0.1 (IBM, Armonk, New York, USA). Carriage prevalences were estimated from logistic regression models adjusted for by sex (male), age (5.0 years) and/or month of inclusion (January). Odds ratios and p-values were calculated after adjustment for age, sex and month of inclusion using logistic regression. p Values below 0.05 were considered statistically significant. The SPSS output file can be viewed online (Additional file 3).

The study protocol was approved by the national ethics committee of Burkina Faso, the institutional review board of IRSS, the ethical committee of the University of Antwerp and the internal review board of the Institute of Tropical Medicine Antwerp, Belgium. Written informed consent was obtained from all participants or their parents/legal guardians with additional assent from those aged 7 to 15 years.

A total of 984 patients met the inclusion criteria for the PALUBAC study, 54 of whom were not included: 27 patients refused participation, 15 patients were referred to another hospital before they could be included, 4 patients died before they could be included and 8 were not included for other reasons. A total of 930 patients were enrolled, 6 of whom were excluded from this analysis because they did not provide a nasopharyngeal swab, resulting in a final number of 924 patients.

More patients were included in the dry season, as compared to the rainy season (83 vs 68 patients per month) and more males than females were included (533 males vs 391 females) (Table 1 and Additional file 1: Sheet 2). The age of included patients ranged from 3 months to 90 years, with a median of 5.0 years. Out of 565 children below the age of 12 years, 62 (11.0%) were classified as malnourished, which was defined as a WHO Anthro weight-for-height z-score < 2.

After admittance, 713 patients (77.2%) were eventually discharged because they had improved clinically, 128 (13.9%) were referred to another hospital, 50 patients (5.4%) died and 33 (3.6%) left against medical advice.

Overall, the unadjusted prevalence of carriage in our study group was 0.44 (S. pneumoniae), 0.30 (H. influenzae), 0.47 (M. catarrhalis), 0.17 (S. aureus) and 0.04 (K. pneumoniae). K. pneumoniae carriage was consistently below 10% in all age groups and throughout the year (Figs. 1 and 2). After adjustment for age and month of inclusion, females were overall less likely to be colonised with S. pneumoniae (OR 0.71, p = 0.022, 95% CI 0.53–0.95) and M. catarrhalis (OR 0.73, p = 0.044, 95% CI 0.54–0.99) than males.

The prevalence of colonisation with S. pneumoniae, H. influenzae and M. catarrhalis was highest in the age groups < 1 year and 1–2 years of age and then decreased sharply with increasing age. S. aureus colonisation showed a peak in the 6–11 years age group but was otherwise constant. K. pneumoniae carriage did appear to be higher in the younger age groups, although no statistically significant linear correlation was found (Fig. 1).

The colonisation rates for S. pneumoniae, H. influenzae and M. catarrhalis showed a marked seasonal association, with rates 2.0–2.9 times higher in January (dry season) as compared to September (rainy season). S. aureus colonisation over the year was more erratic and did not follow a clear seasonal pattern. K. pneumoniae colonisation rates did not significantly vary over the months (Fig. 2).

Out of 924 patients who provided a nasopharyngeal swab, 155 were diagnosed with pneumonia. A total of 584 patients were diagnosed with one or more conditions other than pneumonia and the remaining 186 patients were discharged from the hospital with a diagnosis that neither in- or excluded pneumonia (e.g. ‘septicaemia’ without specification of the primary focus). We found higher odds for pneumonia in carriers of S. pneumoniae (OR 1.75, p = 0.008, 95% CI 1.16–2.63) and H. influenzae (OR 1.90, p = 0.004, 95% CI 1.23–2.92), but not M. catarrhalis (OR 1.13, p = 0.59, 95% CI 0.73–1.73) (Fig. 3a).

‘Bacteraemia’ was defined as the presence of a pathogenic bacterium either in blood culture or in blood qPCR. 88 patients had positive blood cultures, a further 40 patients had positive blood qPCR. In total, 128 patients (13.9%) were bacteraemic: 27 with S. pneumoniae, 20 with H. influenzae, 11 with S. aureus, 1 with K. pneumoniae and 72 with other bacteria (Additional file 1: Sheet 3). Three patients were bacteraemic with two different bacteria.

Colonisation with S. pneumoniae gave higher odds for S. pneumoniae bacteraemia (OR 9.63, p < 0.001, 95% CI 3.28–28.24). Nasopharyngeal colonisation, however, was not a prerequisite for bacteraemia: nasopharyngeal swabs for 5 out of 27 (18.5%) patients with pneumococcal bacteraemia tested negative for S. pneumoniae. In H. influenzae and S. aureus, the correlation between colonisation and bacteraemia was not statistically significant (Fig. 3b).

Remarkably, nasopharyngeal carriage of S. aureus was associated with bacteraemia and mortality, independent of disease caused by S. aureus itself. S. aureus carriers had similarly high odds ratios for bacteraemia with S. aureus (n = 11) (OR 2.44, p = 0.18, 95% CI 0.66–9.10, adjusted for age, sex and month of entry using a logistic regression model) and bacteraemia with other bacteria (n = 117) (OR 2.27, p = 0.001, 95% CI 1.42–3.63, adjusted for age, sex and month of entry using a logistic regression model) as compared to S. aureus noncarriers. Additionally, S. aureus colonisation was associated with lethal outcome (OR 4.01, p < 0.001, 95% CI 2.06–7.83) (Fig. 3c), even after adjustment for bacteraemia (OR 3.54, p < 0.001, 95% CI 1.79–7.02). We did not find this association in S. pneumoniae, H. influenzae or S. aureus. K. pneumoniae colonisation did appear to be correlated with mortality (OR 3.73, p = 0.009, 95% CI 1.38–10.07, adjusted for age, sex and month of entry using a logistic regression model), but with a very wide confidence interval due to low numbers.

HIV positive patients (n = 33) had somewhat higher odds for colonisation with S. pneumoniae, H. influenzae, M. catarrhalis and S. aureus, although none of these correlations reached statistical significance (Fig. 4a). Patients diagnosed with tuberculosis (n = 20), on the other hand, were less likely to be colonised with S. pneumoniae and M. catarrhalis (Fig. 4b). No patients with tuberculosis tested positive for K. pneumoniae. The impact of malaria (n = 343) on nasopharyngeal colonisation was less pronounced (Fig. 4c), although a positive correlation with M. catarrhalis colonisation did reach statistical significance (OR 1.58, p = 0.01, 95% CI 1.12–2.22).

643 patients (69.6%) tested positive for colonisation with at least one of the included bacteria. 416 of them (45.0% of the total) were colonised with multiple bacteria. The influence of nasopharyngeal colonisation with any of the five tested bacteria on colonisation with any of the four others was highly age-specific. In general, associations between carriage with S. pneumoniae, H. influenzae and M. catarrhalis were positive for children < 1 year and indifferent or negative in higher age groups, like for example the association between carriage of M. catarrhalis and S. pneumoniae (Fig. 5a and Additional file 1: Sheet 4). The odds for S. aureus colonisation generally rose for patients aged 12 years and older when colonised with S. pneumoniae, H. influenzae or M. catarrhalis, but in the younger age groups the opposite effect was visible (Fig. 5b and Additional file 1: Sheet 4).