Sample size estimation.

To determine the prevalence of pneumonia and associated factors, we calculated the sample size basing on Ugandan studies done in Mwanamugimu-Kampala, Mbarara and Jinja on prevalence of pneumonia in children with SAM at an average of 26% [8–10]. Using Kish Leslie formula for cross sectional studies [20] the minimum sample size required to determine the prevalence of pneumonia was 301 children. Adjusting for 10% non-response rate, the minimum sample size was 332 children. However, in the current paper we utilized all the 400 study participants from the randomized trial, therefore our study was powered enough to make an inference. To determine the incidence of pneumonia and its associated factors, we estimated a sample size using formula for independent cohort studies [21]. We computed sample size basing on a study by Olivier et al who reported the prevalence of hospital acquired pneumonia to be 33% [22]. Considering the power at 80%, 5% level of significance, RR=1.5 and 1 control per exposed subject, the required sample was 215 per group. The study utilized the single-arm retrospective cohort (only exposed group), therefore the minimum sample size was 215 subjects. Since we utilized the single-arm retrospective cohort (only exposed group), the number of children that were at risk of acquiring pneumonia during hospitalization were 289.Therefore our study was powered enough to make an inference.

Study variables.

We extracted data from the ProbiSAM trial data base. The trial enrolled 400 children aged between 6 to 59 months hospitalized with SAM. The children were randomized into a treatment arm to receive a daily dose of probiotics (Lactobacillus and Bifidobacterium) and placebo arm to receive a dummy sachet of similar appearance and taste like in the other arm, from admission up to discharge. Each arm had 200 children. Data were collected using a validated assessment tool at the time of admission and during follow-up.

At admission.

At admission, trained research assistants that included pediatrician, medical officer, nurses and nutritionist collected data according to their area of specialization.The pediatrician or medical officer collected data on demographics that included age, sex and clinical characteristics of cough or difficulty in breathing, respiratory rate, chest-indrawing, auscultatory findings in the chest, oxygen saturation, grunting, and general danger signs (inability to breastfeed or drink, lethargy or unconscious and convulsions). Diagnosis of pneumonia was made according to the WHO clinical criterion for pneumonia [23]. The clinicians also collected additional clinical characteristics of presence of oedema, fever, temperature, diarrhoea, vomiting, history of any illness in the 2 weeks prior to enrolment and loss of appetite. Anthropometric measurements were done by nutritionists and diagnosis of SAM made according to WHO criteria [24]. Length/height was measured using an infant length board (infant/child shorr-board, Maryland, USA) and MUAC was assessed using a measuring tape, both to the nearest 0.1cm. Bodyweight was measured using a digital scale (Seca 813, Hamburg, Germany) to the nearest 0.1 kg. Z-scores were calculated using the WHO child growth standards [25]. The nurses collected information on the social characteristics that included: house hold size, breastfeeding status, immunisation status and caregiver’s level of education. A visual analogue scale (VAS) assessing the perceived severity of the child’s illness by the caretaker was filled in at admission and daily during inpatient therapeutic care (ITC). The VAS had faces with facial expressions of children and scores ranging between 0 and 10 according to how sick the caretaker thought the child was. Blood samples were taken off to do laboratory measurements that included: HIV serostatus, C-reactive protein (CRP), white blood cell count and a blood smear for malaria parasites. The mother’s HIV status was also done to determine HIV-exposure status of their children. Chest radiographs were done for children who had pneumonia at admission or had presumptive TB. The details of how the different procedures were done in the ProbiSAM trial have been reported elsewhere [18]. We extracted all the variables collected at admission as indicated above from the ProbiSAM data set and analysed them to determine the prevalence and associated factors of pneumonia which is the first part of the objective of our study.

During hospitalization.

During hospitalisation, the child was evaluated for clinical symptoms and signs of pneumonia on daily basis until discharge from the study. The clinical signs included cough or difficulty in breathing, respiratory rate, chest-in drawing and oxygen saturation. Diagnosis of pneumonia was made as per the WHO clinical criterion. Date of onset and stop of these clinical signs of pneumonia were indicated in the questionnaire. If the pneumonia episode reappeared some days after admission in the same child, it was considered part of the previous pneumonia episode. Other clinical characteristics that were collected during hospitalization include degree of oedema, pulse rate, dehydration, fever, temperature, use of a nasogastric tube for feeding, use of Resomal for dehydration, choice of antibiotics given, diagnosis of malaria and date of discharge or death to determine duration of hospitalization. The nutritionist took daily weight and other anthropometric measurements (height, sitting length, MUAC) were taken weekly during hospitalization. Children were treated according to the Uganda national protocol “Integrated Management of Acute Malnutrition (IMAM) [19].

Children who were diagnosed with pneumonia at admission or during hospitalization received intravenous antibiotics of ampicillin and gentamycin as the first line treatment. Those that failed to respond within 48 to 72 hours and those with severe pneumonia were given intravenous ceftriaxone. All children received antibiotics over a period ranging from 5 to 10 days. Re-feeding was started with F-75 formula (Nutriset, Malaunay, France) every 2 hours at 100 -130 ml/kg/day depending on presence or absence of oedema according to the National Protocol for management of malnutrition [19]. When children were unable to feed orally, feeds were given via a nasogastric tube. When oedema plus medical complications resolved and appetite returned, children had an acceptance test performed to evaluate if they can consume a predefined amount of RUTF. When they passed the acceptance test, they were gradually transitioned to RUTF over 2-3 days. We extracted additional variables collected during hospitalization from the data set and analysed them to determine the incidence and risk factors of pneumonia which is the second part of the objective of our study.

Characteristics of severely malnourished children at admission.

There were 400 children in this study, 58% were males. Most (87%) of the children were aged 24 months and below. Seventy-five percent (75%) had received the required immunization doses for that age (Table 1). Two thirds of the children in this study had oedema. About half of the children had severe wasting and half had severe stunting. Only 15% of the children were found to still be breastfeeding among those less than 2 years. Eleven percent (11%) of the children had confirmed HIV infection and 18% were HIV-exposed. About half of the children had mothers/caregivers with no formal education.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline characteristics of severely malnourished children at admission.

https://doi.org/10.1371/journal.pone.0331698.t001

Chest radiographs were done for 129 children who had pneumonia at admission or had presumptive TB. Of these 70% (90/129) had pathological findings (infiltrates or consolidation or pleural effusion). Among those who demonstrated clinical signs of pneumonia at admission, 85% (51/60) had radiological findings confirming presence of pneumonia (Details of these results are not shown in the Table 1).

Prevalence of pneumonia among children admitted with SAM.

The prevalence of pneumonia at admission was found to be 28% (95% CI: 24–33) among children admitted with SAM.

Factors associated with prevalence of pneumonia.

Female Sex, presence of bipedal oedema, WHZ, HAZ, MUAC, CRP, presumptive TB and HIV status of the mother were factors associated with presence of pneumonia at admission on bivariate analysis (Table 2). The prevalence of pneumonia at admission was 65% higher among female children compared to males (cPR: 1.65; 95% CI: 1.20–2.27). Children with oedema, the prevalence of pneumonia at admission was 35% lower compared to children without oedema at admission (cPR: 0.65; 95% CI: 0.48–0.89).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Factors associated with p revalence of p neumonia among children with SAM .

https://doi.org/10.1371/journal.pone.0331698.t002

In the multivariable regression model, female Sex, HAZ and presumptive TB were factors associated with pneumonia at admission (Table 2). The prevalence of pneumonia at admission was 72% higher among female children compared to males (Adjusted PR: 1.72, 95% CI: 1.20–2.45). Further, the prevalence of pneumonia at admission was 60% higher among children with a HAZ score of < -3SD compared to those with a Z score of ≥ -3 SD (Adjusted PR: 1.60, 95% CI: 1.12–2.30). Children with presumptive TB, the prevalence of pneumonia at admission was 95% higher compared to those with no presumptive TB (Adjusted PR: 1.95, 95% CI: 1.34–2.82).

Characteristics of severely malnourished children at risk of acquiring pneumonia during hospitalization.

Of the 289 children that did not have pneumonia at admission, 36% (103/289) developed signs and symptoms of pneumonia (cough or difficulty in breathing, fast breathing, chest in drawing) (Table 3). Most (87%) of the children were aged 24 months and below. About half of the children had axillary temperature recording of more than 37.5⁰C during hospitalization. A third (36%) had difficulty in feeding and were fed through a nasogastric tube (NGT). Two thirds of the children (65%) spent 15 days or more during hospitalization. All other characteristics of the study participants were similar both at admission and during hospitalization.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Characteristics of severely malnourished children who were at risk of acquiring pneumonia during hospitalization (N = 289).

https://doi.org/10.1371/journal.pone.0331698.t003

Sixty-nine (69) children had chest radiographs done and out of these 57% (39/69) had pathological findings (infiltrates or consolidation or pleural effusion) suggestive of pneumonia. Among those who demonstrated clinical signs of pneumonia during hospitalization,70% (35/50) had radiological findings confirming presence of pneumonia (details of the results not shown in the Table 3).

Incidence of pneumonia among children hospitalized with SAM.

The incidence of pneumonia was estimated at 356 (95% CI: 305–416) per 1000 hospital admissions of children with SAM. In the stratified analysis, the incidence of pneumonia was higher among male children 361 (95% CI: 297–439) per 1000 hospital admissions with SAM compared to females 349 (95% CI: 269–451) per 1000. However, the difference was not statistically significant (p value 0.831).

Prevalence of pneumonia.

The prevalence of pneumonia was 28%. The findings in the current study are consistent with previous studies done in Uganda and other developing countries where the prevalence of pneumonia among children with SAM ranged from 20 to 40% [9,10,26,27]. However, the prevalence of 28% is still very high and similar to the current findings of pneumonia among malnourished children in our settings in Jinja (37.6%) and Mbarara (26.8%) [9,10]. And yet to effectively prevent, protect and treat against pneumonia by the year 2025, a multifaceted approach was needed with special focus on vaccination, improved hygiene, adequate nutrition, and access to timely and appropriate medical care [28]. Efforts have been made to address some of this approaches however with severe acute malnutrition we still have a big burden of 3.2 % according UDHS 2022 [4]. Therefore, there is urgent need to address malnutrition to contribute to reducing the high burden of pneumonia in children. In contrast, the prevalence of 28% in the current study is much lower than the one reported in another study done in Bangladesh where they found a prevalence of 68% [29]. The difference in prevalence between these two studies could be explained by the method used in making diagnosis of pneumonia. In the current study, we based on WHO clinical criteria to make a diagnosis of pneumonia compared to Bangladesh where they used both clinical and radiological criteria [29]. In the current study, CXR were done to confirm diagnosis for only children who presented with clinical symptoms of pneumonia. And yet, evidence shows that clinical signs are poor predictors of pneumonia in severely malnourished children [30]. There is a possibility that some children with pneumonia were missed at admission making the prevalence lower than the previous study. Therefore, to improve on clinical diagnosis of pneumonia among children with SAM, it is advisable to do a CXR for children presenting with respiratory symptoms.

Notably, the prevalence of pneumonia in the current study is still high and similar to evidence before and after COVID 19. COVID 19 caused economic disruptions which led to increased food insecurity in households, disrupted food supply and breast feeding practices contributing to increased levels of malnutrition [31] However, after COVID 19 the economic disruptions reduced and the situation normalized. The current study was done prior to COVID 19 among children with severe acute malnutrition which is an important risk factor for pneumonia among children under 5 years [5]. And yet, the burden of malnutrition in our setting is still high (3.2%) according to the recent UDHS 2022 [4]. Therefore, findings in this study reflect the current situation of pneumonia among children in this age group and conclusions have significance in daily clinical practice.

There observations that have reported a decline in pneumonia admissions following introduction of PCV vaccination in Africa [32,33] and Uganda [34]. However, pneumonia is still reported to be the leading cause of death in children under 5 years in this age group [35]. In Uganda, PCV was introduced during the data collection period of the current study. However, hospital based pneumonia studies conducted among severely malnourished children after introduction of PCV vaccine in Uganda have reported a prevalence of 26.8% (Mbarara) and 37.6% (Jinja) [9,10] which is similar with prevalence of 28% in the current study. In addition, the decline reported in pneumonia admissions may not be the same in malnourished children due to presence of other pneumonia-causative organisms that PCV does not protect against such as klebsiella and viral infections like CMV. Furthermore, the PCV vaccination (PCV3) vaccination coverage in Uganda is 72% leaving many children under 5 at risk of pneumonia [4].

Factors associated with prevalence of pneumonia.

We found that female sex, HAZ scores of < -3SD (stunting), and presumptive TB were factors associated with pneumonia at admission in our analysis.

In the current study, the prevalence of pneumonia was higher among female children compared to males. Our findings are different from other studies which have found that male children are more susceptible to getting infections like pneumonia compared to females [36]. The possible reasons behind the high susceptibility of pneumonia in females than in males could arise from severe morbidity and socioeconomic factors. The female children may come with severe forms of SAM hence severe immunosuppression making them more prone to infections than males. Severe forms of SAM among females could be due to gender bias in food and health care allocation among children by the parents [37]. Parents particularly mothers may give higher attention to male children by giving them greater quantities of higher quality nutritious food and promptly taking them to a health facility when they are sick compared to female children [37,38]. Therefore, we recommend further studies to examine the sex differences in prevalence of pneumonia to broaden our understanding of correlates of pneumonia among children with SAM.

Notably, HAZ scores of < -3SD (stunting) was an important factor associated with both prevalence and incident pneumonia in the current study. From our study, children who had stunting, had a higher prevalence and incident pneumonia during hospitalization compared to those without. Our observations are similar to findings from other studies where stunting is associated with pneumonia [36,38]. In the study by Kumdin et al the odds of stunting were 3.6 times higher among children with pneumonia [38]. Stunting as a severe form of malnutrition has long term sequels on lung growth and development predisposing the child to getting pneumonia [38]. In addition, stunting is reported to increase risk of treatment failure and prolong course of recovery in children with pneumonia [39]. Therefore, to broaden our understanding of how stunting relates with pneumonia, we recommend further studies in these area.

In the current study, prevalence and incident pneumonia was higher among children with features of presumptive TB. The findings in the current study are similar to other studies where pneumonia is reported to be a strong correlate of TB in children with SAM [40,41]. In a study by Atalell et al, the odds of being infected with TB were 2.8 times higher among children with SAM and pneumonia as a comorbidity [40]. Children with SAM are prone to infections like Mycobacteria Tuberculosis causing pneumonia because of the impaired cell mediated immunity which is the principal host defense against TB.

Notably, some children with presumptive TB in the current study did not present with pneumonia at admission however they developed incident pneumonia during hospitalization. The possible explanation is some patients with presumptive TB could have responded to the initial antibiotics given due to bacterial co-infection while others failed to mount an immune response against TB due to severe immune suppression. However, as the immune system recovered during hospitalization they mounted an immune response causing a clinical manifestation of TB presenting as pneumonia. Therefore, malnourished children with presumptive TB could be monitored and evaluated for incident pneumonia as a potential comorbidity during hospitalization.

Other factors associated with pneumonia in previous studies including breastfeeding were not reported in this study. This may be attributed to how data on these variable was collected so it could not be correctly interpreted contrary to what we know.

Incident pneumonia.

Our results show incident pneumonia among children hospitalized with SAM was 356 per 1000 hospital admissions of children with SAM. To our knowledge, there is hardly any documentation on incident pneumonia among children hospitalized with SAM. However, previous studies have documented incident rate of pneumonia among general pediatric populations to range from 7% to 84% [22,42,43]. The differences in patient population, study design and case definition of incident pneumonia may explain this variability with the current study. Previous studies were conducted among general pediatric populations of children less than 5 years which includes both the well and poorly nourished children [22,42]. And yet our study focuses on children with SAM which is an important risk factor for incident pneumonia during hospitalization. Furthermore, incident pneumonia in the current study varied by study designs with the previous studies. In one previous study, incident pneumonia was determined by routine point prevalence survey for infection prevention in hospital settings [22] which underestimates incident pneumonia. And yet, the current study utilized data from the ProbiSAM trial database which was collected prospectively with children being monitored daily to identify incident cases of pneumonia. Other previous studies were retrospective case control studies where charts were reviewed causing over estimation of incident pneumonia due to recall bias [42].

The current findings of incident pneumonia among children with SAM is high. And yet, these children are supposed to be on their way to recovery from infections like pneumonia since they are receiving antibiotic and dietary treatment. The high incidence of pneumonia could be explained by some reasons. The incident pneumonia may be high considering that the study was conducted among malnourished children which is a risk factor of pneumonia among children less than 5 years [12,15]. In addition, the causative organisms identified for incident pneumonia are reported to be resistant to the common antibiotics used in our set up [15]. Furthermore, therapeutic feeds given to malnourished children stimulate recovery of the immune system to mount an appropriate immune response against organisms that are initially silent due to severe immunosuppression to manifest as pneumonia. Notably, the risk of death among children with incident pneumonia is reported to be high compared to those who come in with pneumonia at admission [15]. Therefore, we recommend further studies to explore the possible causative organisms and outcomes including mortality for incident pneumonia. This will enable giving appropriate treatment and reduce mortality related to pneumonia among children with SAM. Furthermore, national guidelines could be revised to include criteria to monitor, detect and treat children with incident pneumonia to substantially reduce mortality among children with SAM.

Factors associated with incident pneumonia during hospitalization.

In the current study, age, stunting, HIV positive status, presumptive TB, use of a nasogastric tube for feeding and prolonged hospitalization were risk factors for developing incident pneumonia during hospitalization.

The findings in the current study show that young children aged between 6 to 12 months, were at a higher risk of incident pneumonia during hospitalization. And yet, there is limited data on how age relates with incident pneumonia in children with SAM. However, previous studies on prevalence of pneumonia in children have reported younger age as a risk factor for pneumonia [13,44,45]. The odds of developing pneumonia among younger children less than 12 months was reported to be 2.5 and 3.0 times higher in Ethiopian and Ugandan studies respectively [13,44]. These findings are similar to the current study where young children aged between 6 to 12 months, were at a higher risk of incident pneumonia during hospitalization. The young infants are prone to pneumonia because of their weak immune system that allows progression of the upper respiratory infection to the lungs causing pneumonia.

In the current study, HIV positive children, had a higher risk of incident pneumonia during hospitalization. Our findings are similar to previous studies that have reported HIV infection as a risk factor for pneumonia among malnourished children [46,47]. Both HIV and malnutrition elicit dysfunction in the immune system and promote increased vulnerability of the host to infections like incident pneumonia. Previous studies have also indicated that HIV positive children with pneumonia have poor outcomes including mortality and slow nutritional recovery [46]. Therefore, in the context of high burden of HIV and malnutrition in our set up, incident pneumonia should be recognized as priority for detailed investigation and treatment given the heightened risk of death associated with it.

From the current study, children who used a nasogastric tube (NGT) for feeding during hospitalization were at risk of incident pneumonia. Our findings are similar with a previous study where NGT feeding was associated with aspiration of gastric contents leading to a high incidence of Gram-negative pneumonia [48]. On the contrary, gastric tube feeding is supposed to protect against aspiration. However, aspiration may occur as a complication and it is associated with adverse outcomes like mortality [49]. The possible mechanism for aspiration in malnourished children bearing an NGT is muscle weakness and relaxation of oesophageal sphincters causing aspiration. This may be coupled with a standard practice in malnourished children of frequent feeding day and night which is sometimes forceful because the children may have a poor appetite.

In the current study, prolonged hospitalization was a risk factor for incident pneumonia during hospitalization. This finding is similar with previous studies that have reported incident pneumonia to be strongly associated with prolonged hospital stay [12,15]. During prolonged hospital stay, the normal barriers against infection are lost resulting in increased colonization of the respiratory tract by potential pathogens that cause incident pneumonia [12,15]. In addition, prolonged hospital stay exposes the child to nosocomial infection in overcrowded settings like the case with our inpatient care wards.The organisms identified to commonly cause incident pneumonia are usually resistant to the common antibiotics used in our set up leading to fatal outcomes [15]. These findings highlight the need for clinicians to closely monitor children hospitalized for SAM for incident pneumonia so that prompt treatment and prevention of infection is done to reduce mortality. Furthermore, the policy makers may need to revise and include guidelines for the management of these children.