Aetiological agents of pneumonia among HIV and non-HIV infected children in Ghana: A case-control study.

Pneumonia is the leading cause of death in children, however, the microbial aetiology of pneumonia is not well elucidated in low- and middle-income countries. Our study was aimed at determining the microbial aetiologies of childhood pneumonia and associated risk factors in HIV and non-HIV infected children. We conducted a case-control study that enrolled children with pneumonia as cases and non-pneumonia as controls from July 2017 to May 2020. Induced sputum and blood samples were investigated for microbial organisms using standard microbiological techniques. DNA/RNA was extracted from sputum samples and tested for viral and bacterial agents. Four hundred and four (404) subjects consisting of 231 (57.2%) cases and 173 (42.8%) controls were enrolled. We identified a significant (p = 0.011) proportion of viruses in cases (125; 54.1%, 95%CI: 47.4–60.7) than controls (71; 33.6%, 95%CI: 33.6–48.8) and these were mostly contributed to by Respiratory Syncytial Virus. Staphylococcus aureus (16; 4.0%), Klebsiella spp. (15, 3.7%) and Streptococcus pneumoniae (8, 2.0%) were the main bacterial agents identified in sputum or induced sputum samples. HIV infected children with viral-bacterial co-detection were found to have very severe pneumonia compared to those with only viral or bacterial infection. Indoor cooking (OR = 2.36; 95%CI:1.41–3.96) was found to be associated with pneumonia risk in patients. This study demonstrates the importance of various microbial pathogens, particularly RSV, in contributing to pneumonia in HIV and non-HIV paediatric populations. There is a need to accelerate clinical trials of RSV vaccines in African populations to support improvement of patient care.

Lower respiratory tract infections (LRTIs) refer to infections which affect the airways below the epiglottis and could result in severe disease such as pneumonia. Pneumonia is the leading cause of morbidity and mortality in children and adolescents. Despite significant reductions in child mortality over the past two decades, pneumonia still remains among the top five causes of under-five mortality in the sub-Saharan region [[1]]. The microbial aetiology of pneumonia is variable throughout different population structure, and this is influenced by pathogen endemicity, seasonal climate, geographic location and patient specific characteristics such as human immunodeficiency virus [[2]].

Pulmonary infections remain the commonest immunodeficiency virus associated illness with complication in the era of antiretroviral therapy (ART) [[4]]. A systematic review of hospital admissions in HIV-infected patients reported that community acquired pneumonia and tuberculosis collectively accounted for 57% of in-patient deaths globally [[5]]. Although tuberculosis is often implicated in HIV associated pneumonia, information on the role of other bacterial and viral pathogens is still limited in many developing countries including Ghana. The risk of mortality in hospitalized patients is believed to increase due to limited diagnosis of microbial aetiologies of pneumonia [[6]]. Effective treatment and management of patients with pneumonia heavily depends on the knowledge of the microbial aetiologies but clinically differentiating between these aetiologies is difficult as they have overlapping presentations. Apart from tuberculosis, it is unclear whether other viral or bacterial organisms found in patients with pnuemonia exist as commensals or pathogens. It is also not known if the presence of these microbial organisms is influenced by HIV infection.

To help improve on this, there have been numerous efforts made especially in developed countries to ascertain the scope of infectious pathogens occurring in HIV-infected patients as compared to non-HIV patients [[7]]. However, information on the microbial aetiologies of pneumonia in sub-Saharan African countries is still limited. The expansion of molecular diagnostic tests in Africa has improved the identification of microbial pathogens in pneumonia irrespective of the HIV status. The aim of this study was to determine the microbial agents associated with pneumonia in HIV and non-HIV populations and identify other associated risk factors.

This was a prospective, observational case-control study that was conducted at the Komfo Anokye Teaching Hospital from July 2017 to May 2020. The hospital is the second largest in Ghana with 1200 beds. It also serves as the main referral hospital for the Ashanti, Brong-Ahafo and most of the northern parts of Ghana. Recruitment of study participants were done at the Paediatric Emergency Unit (PEU), Paediatric Intensive Care unit (PICU) and Paediatric HIV Clinic (PHC). The PEU takes care of children presenting with medical emergencies. The annual admissions at PEU is about 10,000 with about 21% being pneumonia (Komfo Anokye Teaching Hospital, unpublished). PICU is an 8-bed unit that takes care of children up to 16 years and admits about 102 patients every year. The PHC takes care of confirmed HIV positive patients from birth up to 20 years of age on outpatient basis. In the year 2018, there were approximately 3,025 registered patients in the PHC. The PHC attends to approximately 100 patients a week (Komfo Anokye Teaching Hospital, unpublished).

Cases were defined as children 3–12 months presenting with clinical symptoms and signs of pneumonia including fever (>37.5°C) and cough with or without difficulty in breathing as well as at least one of the following signs: high respiratory rate (>50 cpm), chest in-drawing, intercostal recession, stridor, pulmonary crackles, lethargy, oxygen saturation < 90% and radiological evidence of pneumonia if available [[8]]. Cases were sub-classified as severe pneumonia and very severe pneumonia based on the World Health Organization (WHO) classification. Severe pneumonia were patients presenting with cough and/or difficulty in breathing plus fast breathing and chest in-drawings. Very severe pneumonia was defined as cough and/or difficulty in breathing, with one or more of the following signs: central cyanosis or oxygen saturation < 90% on pulse oximetry, severe respiratory distress (e.g., grunting, stridor, severe chest in-drawing) and danger signs of pneumonia (inability to drink or breast feed, head nodding, lethargy or unconsciousness and convulsions) [[9]].

Controls were children with comparable age (+/- 3 months) to the recruited cases without history of chronic underlying disease, pneumonia, or respiratory ailment over the past one month prior to screening. Controls were randomly drawn from the same catchment area of the cases and invited to the healthcare facility for sample collection.

Cases and controls who refused sampling and those outside the catchment area were excluded from the study. Patients positive for tuberculosis and those not on anti-retroviral therapy were also excluded from the study. Controls with history of febrile or respiratory tract illness up to 1 month prior to recruitment were also excluded.

All recruitment processes were carried out by paediatricians and senior resident paediatricians with the support from nurses at PEU, PICU and PHC. For study cases, on arrival at the units, patients with clinical suspicion of pneumonia were considered for the study. A pre-recruitment screening checklist was used to determine eligibility and those who satisfied the criteria were recruited. The study procedure was explained to the parents/guardians and written informed consent was obtained if they agreed to participate. Chest X-rays were done for subjects who had been treated for 48 hours but showed no signs of improvement. The decision to request for chest-x-ray for patients was solely based on the judgement of the paediatric specialist. Chest X-rays were interpreted by a radiologist as normal or abnormal based on standard reporting format. Patients who satisfied the screening procedure were counselled and tested for Human Immunodeficiency Virus (HIV) as part of their routine care. Those positive were referred to the HIV Clinic for subsequent assessment. For every case that was recruited, a healthy control of comparable age was also selected from the neighbouring community of the case. Neighbours and relatives of the cases assisted with this exercise. Controls were selected within 2–4 weeks of recruiting cases.

Biodata collected from the study participants were socio-demographic variables (e.g., age, sex, occupation and educational level of caregivers), clinical information (e.g., weight, height/length, signs and symptoms of pneumonia, presence of complications, radiological features and outcomes of cases) and risk factors (e.g., method of cooking, marital status, mother's status of breastfeeding and cigarette exposure). Data were collected from the participants using structured paper-based questionnaire.

Sputum or induced sputum was collected from both cases and controls. Sputum induction was performed according to guidelines and procedures described by Hammit et al and Grant et al [[11]]. Briefly, 2.5 mg/ml of inhaled salbutamol was administered for 10–15 minutes and later followed by 2ml of inhaled hypertonic saline using nebulizer. A minimum volume of about 1ml was collected from patients. The entire sampling procedure was carried out under aseptic conditions.

Sputum and induced sputum samples were transported via cold chain (4°C) to the Kumasi Centre for Collaborative Research in Tropical Medicine (KCCR). At KCCR laboratories, 1ml of the sputum samples were aliquoted into 1ml of RNA later and stored for viral analysis. The leftover was subjected to microbial culture. Blood culture was performed on patients for whom the doctor suspected sepsis in addition to respiratory illness. About 1–3 ml of blood was directly collected into BACTEC bottles from children weighing below 13 kg and 5–8 ml from those 14 kg and above and transported at room temperature to KCCR. Volumes of blood cultures were marked on the bottles before addition of blood and this was observed until the expected mark was reached.

Standard microbiological techniques were performed on sputum and blood culture samples to isolate and identify bacteria from the clinical specimens collected. Sputum samples were pre-treated with sputasol and inoculated by streaking onto chocolate agar (CA), blood agar (BA) and MacConkey agar using sterile disposable loop (10μl). MacConkey and BA were incubated overnight under aerobic conditions at 35–37°C while CA plates were incubated in candle jar and placed in incubator at 35–37°C overnight. Blood cultures were incubated at 35°C using automated BACTEC 9050 system (BD, USA) for a maximum of 7 days. Samples with a positive signal were further processed by inoculating on MacConkey and blood agar plates for further incubation under aerobic conditions (35–37°C).

Morphological characteristics, lactose fermentation, haemolysis on BA, Gram staining and biochemical reactions were used to describe presumptive identification. All Gram-negative bacteria were biochemically confirmed using analytical profile index (API) 20E or 20NE (Biomerieux, France) following all standard microbiological procedures. All Gram-positive bacteria were identified using colony morphology, haemolysis on BA, optochin sensitivity, catalase, and coagulase tests. Streptococci bacteria were speciated using Streptococcal latex grouping kits (Oxoid, UK). Escherichia coli ATCC 25922, Salmonella Typhimurium ATCC 14028, Staphylococcus aureus ATCC 25923 and Streptococcus pneumoniae ATCC 254697 were set up together with the test organisms to control media and biochemical tests.

Nucleic acids were purified from the sputum samples using Qiagen Viral RNA Mini Kit according to the manufacturer's protocol (Qiagen, Hilden, Germany). Purified nucleic acids were tested for Respiratory Syncytial Virus (RSV), Influenza A and B (INF A, B), Parainfluenza 1, 2, 3 (PIV 1, 2, 3), Adenoviruses, Human Coronaviruses (OC43, HKU1, NL63, 229E), Human metapneumovirus (hMPV) and Rhinoviruses (HRV). SARS-CoV-2 test was performed on RNA extracts in retrospect during the outbreak of COVID-19. The Qiagen One Step RT-PCR kit was used for testing following manufacturer's protocol. Briefly, a 25μl reaction mixture consisting of optimized volumes of deoxynucleotide triphosphate, magnesium chloride, Qiagen One Step Buffer and enzyme mix was reconstituted for virus detection. Samples were pooled together and singleplex RT-PCR was performed with primers and probes highly specific to the target regions of the viruses. SARS-CoV-2 testing was performed on the samples using DaAnGene Detection Kit for 2019 n-CoV (Guangdong, China). The SARS-CoV-2 PCR kit was designed to target the ORF1ab and nucleocapsid regions of the virus. Test results for each virus was interpreted as positive based on cut-off values of the cycling threshold stated in the manufacturer's protocol. The PCR cycling conditions for the other viruses were done as described by other authors (Table 1).

Graph

Table 1 Primers and probes used for the molecular detection of respiratory viruses.

Ethical approval was sought and obtained from the Committee on Human Research Publication and Ethics (CHRPE) of the School of Medicine and Dentistry, Kwame Nkrumah University of Science and Technology, Kumasi (Approval numbers: CHRPE/AP/389/17; CHRPE/AP/537/18; CHRPE/AP/530/19) and permission sought from the Komfo Anokye Teaching Hospital before commencement of the study. Confidentiality was ensured by issuing participants with unique identification numbers, with the patients' names omitted. Written informed consent were provided by parents or guardians prior to recruitment of their children.

Prior to undertaking this study, we estimated a samples size of 179 individuals each for cases and controls. This estimate was based on previous RSV detections of 21.3% being the most predominant virus among children and estimated 10% exposure among controls (Kwofie et al., 2018) [[21]]. The estimated sample size determined was based on a power of 80% and a two-sided alpha error of 5%.

Data were entered into Microsoft Excel and exported to R statistical software (version 3.6.1) for analysis. Descriptive statistics were computed for continuous and categorical variables. Statistical comparisons between subgroups of categorical variables were analyzed using the Fischer's exact test or Chi-square test where appropriate. A multivariate logistic regression was used to determine the association between microorganisms detected and pneumonia risk. Multivariate models were adjusted for microorganisms detected in induced sputum separately for HIV and non-HIV patients. Results were expressed as odds ratio with 95% confidence interval. For all analysis, a p-value of less than 0.05 was considered statistically significant.

The study recruited 404 individuals with males being 232 (58.3%). Of the 404 recruited, 148 (36.6%) were HIV infected persons and among them 57.4% (85/148) presented with pneumonia. A comparison of the socio-demographic variables and other risk variables among non-HIV populations showed history of breastfeeding, indoor cooking, marital status and mode of cooking methods were associated with pneumonia. Among HIV populations we observed, guardians' highest educational level, exposure to cigarette and mode of cooking were associated with pneumonia. Table 2 gives a breakdown of the socio-demographic variables their association with the pneumonia.

Graph

Table 2 Socio-demographic characteristics of HIV and non-HIV cases and controls.

Of the 404 patients, blood cultures were performed on 194 pneumonia cases who presented with signs and symptoms suggestive of sepsis as determined by the attending physician. The prevalence of pathogenic bacteria isolation in blood culture was 5.7% (11/194). The organisms isolated were Acinetobacter spp. (1; 0.5%), E. coli (2; 1.0%), Pseudomonas spp. (2; 1.0%), Salmonella spp. (3, 1.5%), Staphylococcus aureus (2; 1.0%) and Group F streptococci (1; 1.0%). There was no difference between the proportion of bacteria isolated in the blood culture of HIV patients (4, 5.1%) compared to non-HIV (7, 6%).

Of the 404 patients recruited, bacteria organisms were isolated in 62 subjects (15.3%). The commonest bacteria isolated was Klebsiella spp. (15, 3.7%), followed by Staphylococcus aureus (16; 4.0%) and Streptococcus pneumoniae (8, 2.0%). The overall proportion of bacteria in pneumonia cases (46, 20%) was significantly different (p = 0.005) than that of controls (16, 9.2%). There was no significant difference in the overall rate of bacteria identified in the HIV cases (19, 22.4%) compared to non-HIV cases (25, 17.1%). Subgroup analysis however identified Pseudomonas aeruginosa as being predominant in HIV cases (5, 5.9%) compared to non-HIV cases (0) but the difference was not significant. On the other hand, Staphylococcus aureus was more common in non-HIV cases (9, 6.2%) compared to HIV cases (0) while Klebsiella spp. was predominant in cases (14, 6.1%) compared to controls (1, 0.6%). Klebsiella spp. was independently associated with pneumonia risk. Of all patients tested, we observed only one patient had Staphylococcus aureus isolated in both blood and sputum. Table 3 gives details of pathogens identified among study subjects.

Graph

Table 3 Bacteria distribution in sputum/induced sputum samples.

1 N/A: Could not estimate values due to low numbers.

• 2 aOdds ratio adjusted for other viral organisms predicting pneumonia in non-HIV patients compared to HIV patients
• 3 bOdds ratio adjusted for other viral agents predicting for pneumonia cases compared to controls.

We tested for 14 different viruses among 404 patients consisting of 231 pneumonia cases and 173 controls. Of all patients tested, the most common virus identified was HRV (89, 22.1%), followed by RSV (52, 12.9%), Adenovirus (49, 12.2%), Influenza A (24, 6%), HCoV-HKU1 (10, 2.5%), HCoV-NL63 (8, 2%), HCoV-OC43 (5, 1.2%), PIV-1 (2, 0.5%) and Influenza B (5, 1.2%). PIV-3 and SARS-CoV-2 were not detected in the samples. Viral co-infections occurred for Adenovirus and HRV (20, 5%); Adenovirus and Influenza A (2, 0.5%); HRV and RSV (11, 2.7%) and Influenza A and RSV (6, 1.5%). The total number of viruses identified in the cases (125, 54.3%) was higher than controls (71, 41%) and the difference was statistically significant (p<0.01). RSV and Influenza viruses were the most common in cases compared to controls. RSV was found to be independently associated with the odds of developing pneumonia (6.85, 95%CI = 2.81–16.71), especially in non-HIV cases with higher odds (7.49, 95%CI:2.55–21.98) compared to HIV patients. Table 4 gives description of viral pathogens identified.

Graph

Table 4 Virus distribution in cases and controls.

• 4 aOdds ratio adjusted for other viral organisms predicting pneumonia in non-HIV compared to HIV
• 5 bOdds ratio adjusted for other viral agents predicting for pneumonia cases compared to controls. NA: Could not estimate values due to low numbers.

As part of this study, we collected chest X-ray information on patients and correlated this with selected viral or bacterial organisms. Of the 231 pneumonia cases enrolled, chest X-rays were done for 149 (64.5%). Of these, 72 (48.3%) were abnormal. The common abnormalities were alveolar (42, 28.2%) and interstitial (22, 15.4.%) infiltrates. Others were mainly pleural diseases including effusion, pneumothorax and hydropneumothorax. For all the pathogens identified, we observed influenza A was significantly associated with abnormal chest x-ray. Further stratification based on HIV status did not show any difference in x-ray abnormality for all pathogens except for Influenza A which was found to be associated with abnormal chest-X-ray in non-HIV patients (Table 5).

Graph

Table 5 Chest X-rays findings for various pathogens detected.

We explored other clinical variables and outcomes associated with viral detection, bacterial infections and viral-bacterial co-infections. We observed significant association between patients having viral-bacterial co-infections and lower chest-indrawing (p = 0.04) and very severe pneumonia (p = 0.017) (S1 Table). Among HIV cases, we observed significant variations between viral-bacterial co-infections and clinical manifestations including poor feeding, fast breathing, chest recession, lower chest in-drawings and very severe pneumonia. Pulmonary crackles were more associated with viral infections (S2 Table). We did not observe significant association between non-HIV populations and microbial detection (S3 Table). A sub-level analysis of the individual viruses (RSV, Influenza A and others) and bacteria did not also show any significant association with the clinical presentations.

In order to ascertain the seasonal variations of microbial detections over the study period, we plotted the top three microbial organisms (influenza A, RSV and Klebsiella spp.) by the date of enrolment of patients. We observed significant variations in RSV detections (p = 0.001) with most infections occurring in the month of June and July. Influenza A viral infections also occurred frequently in June and July however the variations across the various months was not significant. Klebsiella spp. was almost evenly distributed over the various months (Fig 1).

Graph: Fig 1 info:doi/10.1371/journal.pone.0299222.g001

Of the 231 pneumonia cases enrolled, 3 (0.74%) patients died at the end of the study. All 3 cases of death had a median age of 4 months (IQR = 3.5–6) and needed oxygen within the first 48 hours of admission. The deaths occurred among non-HIV subjects with pneumonia. One of the 3 patients who died had only Staphylococcus aureus isolated in the induced sputum, and another had both Staphylococcus aureus and influenza A co-infection.

Of the various infectious agents of pneumonia, viral associated pneumonia is a major public health problem due to their ease of transmission, substantial morbidity, and wide occurrence. Our study identified common viral agents including RSV, Adenoviruses, influenza A and human coronaviruses in the study population. We did not record any case of SARS-CoV-2 because the samples were collected before the outbreak of SARS-CoV-2 in Ghana.

Higher proportion (54.3%) of viruses was observed among patients with pneumonia compared to controls with RSV being most predominant. This prevalence is higher than what was reported from our previous research conducted 10 years ago at the same study site, in which there was virus isolation in 25.8% of pneumonia patients [[21]]. A possible explanation for this discrepancy could be the utilization of induced sputum samples which some studies have reported to provide better microbial yield than nasopharyngeal samples [[22]]. We observed control subjects had frequent detections of HRV and Adenoviruses contributing mostly to 41% of controls. This observation has been reported by other authors where HRV and Adenoviruses are deemed to be more common in health individuals and may not be of any clinical value causing disease. The younger age group in this current study could also explain the increased detections of RSV compared to our previous study [[21]]. The previous study enrolled children less than 5 years of age while this current study enrolled children less than 12 months.

RSV was observed as the main virus associated with pneumonia in both HIV and non-HIV patients. The implications of RSV in pneumonia has been reported by other authors [[24]]. Our study did not however record significant association between RSV infections and clinical manifestations perhaps due to the low number of case groups enrolled. Other studies have however recorded more severe and prolonged symptoms with increased risk of complications especially in very young children [[26]–[29]]. In both developed and developing countries, wheezing has also been reported [[30]].

The increased detections of RSV in children presenting with pneumonia emphasise the need for interventions to reduce the burden of disease in Africa. Currently there is no specific treatment for RSV. Only supportive care measures like oxygen and fluids exist. Vaccine to prevent RSV is also not widely available to the public. However, there are reports of Phase 3 trials of maternal RSV vaccines which have shown 82% effectiveness in preventing severe lower respiratory tract infections due to RSV in newborns up to 90 days after birth, and 69% effective for the first 6 months after birth, [[32]] giving hopes of possible implementation globally in the near future.

Assessment of sputum or induced sputum specimens using conventional cultures revealed a high number of bacterial organisms in pneumonia cases compared to controls. The most predominant microbial organism isolated was Klebsiella spp., followed by Streptococcus pneumoniae and Staphylococcus aureus. For both HIV and non-HIV populations, Klebsiella spp. was found to be significantly predominant in cases compared to controls. Klebsiella spp. is reported as one of the most common bacterial agents of community acquired pneumonia [[33]]. Klebsiella inclusively was associated with 63.4% of deaths from infectious diseases, including acute and secondary pneumonia, post-traumatic and post-operative complications and sepsis [[34]]. Klebsiella spp. strains were found to be three times more prevalent than pneumococci and more than six times prevalent than other types of bacteria [[34]]. We also observed other common microbial organisms including Streptococcus pneumoniae in pneumonia cases and Pseudomonas aeruginosa in HIV patients but these were not significant. Other studies have indicated the important role of these organisms in contributing to severe and very severe pneumonia [[35]].

Conversely to single viral or bacterial infection, viral-bacterial co-infection was found to be significantly associated with very severe pneumonia. This association was found to be more pronounced in HIV patients compared to non-HIV patients. Determining the contribution of mixed infection to disease severity is highly complex, particularly among the pediatric population. Despite these complexities, many clinical studies on co-infection seem to provide evidence of enhanced disease severity during polymicrobial acute respiratory tract infection in children [[36]]. Quah et al., recently reported viral-bacterial co-infections were associated with increased risk of hospital mortality (OR = 13.99; 95% CI = 1.3–151.05) [[38]]. The severity of clinical disease could be explained by the ability of mixed viral-bacterial co-interactions that are known to modulate the JAK-STAT signaling pathway which is believed to exacerbate IP-10 expression [[39]]. This therefore suggest the importance of managing these microbial organisms when isolated from patients who show clear signs of pneumonia.

In terms of socio-demographic characteristics of our study populations, we observed HIV infected children exposed to cigarette and non-HIV children involved in indoor cooking and other modes were more associated with pneumonia. Indoor air pollution is one of the leading risk factors of childhood pneumonia in developing countries. Exposure to environmental tobacco smoke is a major source of fine and respirable particles in indoor environment. Parental smoking, use of solid fuel and other indoor activities such as use of mosquito coils are reported to increase the risk of pneumonia. This phenomenon has been reported in other developing countries [[40]].

One major limitation of the study is our inability to use molecular techniques to investigate the full spectrum of non-viral microbial organisms. We did not also perform conventional cultures for fungal agents. The use of these advanced methods could have allowed for proper estimation of the full spectrum of other non-viral agents involved in the causal pathway of pneumonia. Nevertheless, the conventional techniques used provides baseline information that could guide future studies. The difficulty in enrolling healthy people to provide induced sputum limited our ability to recruit large number of persons for this study. An imbalance in the case-control ratio numbers could have impact on the results.

This study has emphasised the importance of microbial organisms particularly RSV in contributing to pneumonia among children with/without HIV. Identification of these pathogens has improved our knowledge on the burden of respiratory pathogens in Africa. The study has provided further evidence about the need for further advocacy for vaccines to improve the wellbeing of children in Africa.

S1 Checklist

STROBE statement—checklist of items that should be included in reports of observational studies.

(DOCX)

S1 Table

Clinical presentations associated with microbial detection for all cases.

(PDF)

S2 Table

Clinical presentation associated with microbial detection in HIV patients with pneumonia.

(PDF)

S3 Table

Clinical presentation associated with microbial detection in non-HIV patients.

(PDF)

The authors are grateful to the staff and management of Komfo Anokye Teaching Hospital, and to all participants, parents and guardians who participated in the study.