STRENGTHS AND LIMITATIONS OF THIS STUDY

The strengths of this study were the recruitment of study participants from different institutions, for which there is little evidence, and the clear indication of what was already known prior to this study.

One limitation was that the study did not include sociodemographic and economic characteristics of the parents.

Another limitation was that the death reports were not cause specific, which may lead to misreporting.

WHAT IS ALREADY KNOWN ON THIS TOPIC

Pneumonia is the major cause of morbidity and mortality in children responsible for around 14.9 million hospital admissions every year.

Most pneumonia-related deaths are caused by severe community-acquired pneumonia (SCAP).

These deaths are preventable by early detection, timely referral and access to high-quality treatment.

Prolonged hospitalisation due to pneumonia leads to medical complications in children and increased healthcare cost and stress in the family and the society at large.

WHAT THIS STUDY ADDS

Younger age and early presentation to the healthcare institution were proven to hasten recovery time from SCAP.

On the other hand, presence of comorbidity and being stunted were proven to prolong recovery time from SCAP.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Interventions that increase recovery rate and hasten recovery time such as primary prevention to eradicate malnutrition and to escalate healthcare seeking behaviour of the society should be strengthened, and the health workforce should focus on early identification and management of comorbidities.

Future researchers might use this study as a baseline to instigate this title using prospective cohort design with inclusion of some missed variables of interest.

Introduction

Pneumonia is an acute respiratory infection that primarily affects the parenchymal tissue of the lungs.1 It is a major cause of morbidity and mortality in children worldwide, leading to approximately 14.9 million hospitalisations each year.2 The WHO estimates that around 2 million children under the age of 5 years die from pneumonia each year, with most of these deaths occurring in developing countries.3

In Africa, thousands of children die from pneumonia every year.4 In sub-Saharan Africa, it causes around 750 000 child deaths each year.4 Although Ethiopia has reduced the under-5 mortality rate by 71% in recent decades, pneumonia still causes 15 deaths per 1000 live births and the country is ranked 6th in the world among the 15 countries with the highest morbidity and mortality due to pneumonia.5 According to WHO estimates from 2016, pneumonia is responsible for 16.4% of under-5 mortality in Ethiopia.6 These alarming death rates have been reported in Ethiopia despite the increasing availability of preventive vaccines such as pneumococcal conjugate vaccine and Haemophilus influenzae type b conjugate vaccine. In Ethiopia, the Federal Ministry of Health recommends the administration of three doses of these two vaccines at 4, 10 and 14 weeks of age.7 8

Community-acquired pneumonia (CAP) is an infection that begins before admission to hospital and is diagnosed within the first 48 hours of admission in a person who has not previously been hospitalised for 14 days or more. In contrast, hospital-acquired pneumonia occurs after 48 hours of admission with no evidence of infection at the time of admission.9 Although pneumonia is easily treatable and curable in most patients, there comes a time when it becomes severe and causes a life-threatening (complicated) medical condition that requires a long hospital stay, especially in children.3 10

Severe CAP (SCAP) is a clinical condition characterised by a temperature of >38.5°C, nasal flaring, cyanosis, intermittent apnoea, grunting respirations, failure to feed, tachycardia, a capillary refill time of ≥2 s and a respiratory rate>70 and 50 breaths/min in infants and older children, respectively.11 Pneumonia is considered complicated when the patient develops a parapneumonic effusion, empyema, necrotising pneumonia, lung abscess, etc, and these complications were associated with the length of hospital stay, the time elapsed since the onset of pneumonia symptoms, the duration of fever before admission, asymmetric chest pain, treatment with ibuprofen and paracetamol before hospitalisation, heart disease, older age and low O2 saturation.12–15 It is well recognised that most pneumonia-related deaths are caused by SCAP, and treatment of these conditions requires early recognition, timely referral and access to high-quality care. However, in many resource-limited contexts, referral is difficult and often delayed.3 10 In countries such as Ethiopia, where little is spent on healthcare and hard priorities must be set, pneumonia has killed thousands of children and remains the leading cause of death in children.5 6

Studies have shown that pneumonia accounts for 38.6% of all emergency admissions and is the leading cause of death in Ethiopian referral hospitals.16 These deaths are easily preventable and treatable by simple and cost-effective measures if the duration of hospitalisation is estimated.16 The total medical cost of CAP is directly related to the cost of hospitalisation and length of stay. Hospitalisation for pneumonia has enormous economic disadvantages, either directly through medical costs or indirectly through the loss of work hours of parents of sick children.17

The disadvantages of prolonged hospitalisation of children due to pneumonia affect not only the children themselves but also the mothers and caregivers, the family, society and the country as a whole. In addition to the financial burden, families who stay longer in hospitals face many problems, such as reduced responsibility for the household and poorer school performance of household members.18 The impact of inadequate treatment of pneumonia becomes even more pronounced when the length of stay is prolonged and the time of recovery is unpredictable.19

Although several studies have been conducted in different countries, these studies were conducted specifically in children under 5 years of age to assess the extent, determinants and associated factors of pneumonia. These studies found that the following factors were associated with the outcome: pallor; abnormal white blood cell count; age; excessive breast feeding; emaciation; oedematous malnutrition; granting; and oral antibiotic intake.6 7 10 12 13 20

In addition, there are some studies in Ethiopia that investigated the estimated recovery time and predictors of SCAP in rural hospitals. In these studies, the estimated median recovery time was 3 and 4 days, respectively, and age, stunting, underweight, danger signs on admission, presence of comorbidities, delayed hospitalisation, type of infant feeding and presence of severe signs and symptoms on admission were found to be associated with length of hospital stay.2 21

Moreover, the socioeconomic and educational status of the population studied in the rural towns was not comparable to that of the population of Addis Ababa; consequently, the health level of the population in the rural towns is very low compared with the population of Addis Ababa, the capital of Ethiopia (the setting of this study). This in turn leads to delayed healthcare utilisation by society, complicated diseases and a longer duration of hospitalisation. Moreover, the findings of these two studies were variable in terms of recovery rate and time and identified associated factors which indicated that the need for further study. Therefore, the aim of this study was to determine the median recovery time after SCAP and the associated factors.

MethodsStudy design, area and population

A facility-based retrospective cohort study was conducted in four randomly selected public hospitals in Addis Ababa, Ethiopia. The study was conducted at Tikur Anbessa Specialized Hospital (TASH), Menelik II Hospital (MIIH), Zewditu Memorial Hospital (ZMH) and Yekatit 12 Hospital (Y12H), which were selected by lot. According to the city administration, these hospitals serve more than 5 million inhabitants. In these hospitals, children under the age of 15 are treated in a separate paediatric department according to national guidelines adopted from the WHO. The total number of pneumonia cases over a 3-year period for each hospital was 1272 in TASH, 1583 in ZMH, 1017 in Y12H and 1368 in MIIH. In these hospitals, children with SCAP were admitted to the paediatric ward, and further diagnosis and treatment were provided by paediatricians, general practitioners and nurses. The study population consisted of randomly selected under-15 children with SCAP who were admitted to four selected public hospitals in Addis Ababa for treatment of SCAP between 1 January 2018 and 31 December 2020.

Illegibility criteria

Children aged 1 month to 15 years who were admitted to the paediatric ward with SCAP between 1 January 2018 and 31 December 2020 were included in this study. Pneumonia is considered severe if the oxygen saturation is <90% or if there is severe respiratory distress, central cyanosis, altered consciousness, convulsions or the inability to drink, breast feed or vomit anything. This study excluded children whose record was missed during data collection and children with incomplete records. The study also excluded debilitated patients with oncologic diseases, immunodeficiencies and rheumatologic diseases.

Patient and public involvement

Since the study was retrospective chart review, it was not possible to access and involve patients in the design, or conduct, or reporting, or dissemination plans of our research.

Sample size determination and sampling procedures

A double population proportion formula was used to determine the sample size, taking into account a study conducted in southwestern Ethiopia,22 using the Kelsey formula of the Epi Info V.7 statistical package, and considering vaccination status as an important predictor (see online supplemental file 1). The final sample size, including the non-response rate, was 407. According to the most recent data from the four hospitals, approximately 5240 children were admitted for treatment of SCAP, and this was used as the sampling frame. The total sample size for each hospital was proportionally divided based on the sampling frame (TASH, N=1272; ZMH, N=1583; Y12H, N=1017; and MIIH, N=1368) and sample size n (k=N/n). Therefore, 99 children were selected from the TASH, 123 children from the ZMH, 79 children from the MIIH and 106 children from the Y12H (see online supplemental file 2), using systematic random sampling at k=13 intervals to minimise selection bias.

Variables

The only dependent variable is time to recover from SCAP and independent variables included in this study were:

Sociodemographic variables like age of children, sex of children, etc.

Presence of comorbid diseases.

Nutritional status of children like stunting, wasting, underweight, etc.

Presence of danger sign during admission including loss of consciousness, abnormal body movement, vomiting everything, convulsions and inability to feed/drink.

Elapsed to seek care (duration before care seeking).

Immunisation status of children.

Operational definitions

Paediatrics: refers to children from 1 month to 15 years of age.2 23

Recovery: if the child had improved or cured from SCAP as confirmed by the physician and written on the patient’s discharge summary or HMIS. Recovery might not necessarily mean completion of the whole course of antibiotics.7

Survival time: refers to the time starting from admission date to recovery from SCAP.2

Event: refers to recovery from an illness during the study period.16

Censored: refers to children referred, died or discharged for any reason without recovery during the study period.24

Comorbidity: if the child had any disease condition either before or after admission in addition to SCAP which includes hyperactive airway disease, severe acute malnutrition, congenital heart disease, congestive heart failure, rickets, Down syndrome and others.2

Severe community-acquired pneumonia: the child acquired pneumonia before 48 hours of admission with oxygen saturation<90% or severe respiratory distress or central cyanosis or altered consciousness and convulsions or inability to drink or breast feed or vomiting everything.25

Danger sign: the child is considered as having danger sign if he/she has experienced loss of consciousness, abnormal body movement, vomiting everything, convulsions and inability to feed in addition to SCAP.23

Early presentation: refers to seeking medical care with in the first 5 days of illness.2 26

Late presentation: refers to seeking medical care after 5 days of illness.2 26

Data collection tool, procedure and data quality assurance

The format for data collection was adopted from another peer-reviewed article conducted in Ethiopia (see online supplemental file 3).2 All medical charts of children admitted with SCAP in four selected hospitals between 1 January 2018 and 31 December 2020 were retrieved from the Health Management Information System registration book. Two BSc nurses were recruited for supervision, and four BSc nurses were recruited for data collection. To ensure the quality of data, supervisors and data collectors were trained for 1 day before data collection; a standardised data extraction form was pretested to ensure that the format of data collection was consistent with the study requirements. During data management, storage, cleaning and verification, all completed data collection forms were checked for completeness and accuracy.

Data processing and analysis

Data were checked, coded, entered into EpiData V.4.2 and exported to Stata V.16.0 for analysis. Descriptive statistics were presented in the form of tables and charts. Days were used as a time scale to calculate recovery time, and cumulative probabilities of survival after admission were estimated using a life table.

The Kaplan-Meier survival curve and the log-rank test were used to describe the survival function and to compare the survival curve of categorical predictor variables, respectively. A bivariate Cox proportional hazard regression model was created for each predictor. Subsequently, further variable selection for the multivariable Cox regression analysis was performed by stepwise backward variable selection with a p value≤0.25 as the cut-off value.2 22 The Cox proportional hazard assumption was tested for each covariate using Schoenfeld residual tests and graphically with a log–log plot of survival, and there was no variable that violated the proportional hazard assumption. Then, the adjusted HR with its 95% CI was used to measure the strength of association and p value<0.05 was used to identify the statistically significant association. Finally, the finding was reported based on the Strengthening the Reporting of Observational Studies in Epidemiology guideline (online supplemental file 4).

ResultsSociodemographic characteristics

Out of the 407 selected medical charts, only 388 were complete with a response rate of 95.5%. More than half (57.73%) of participants were boys and their median age was 11 months (table 1).

Table 1

Kaplan-Meir survival estimates for severe acute community-acquired pneumonia recovery time with different covariates in selected governmental hospitals of Addis Ababa, Ethiopia, from 1 January 2018 to 31 December 2020 (n=388)

Clinical characteristics and drug regimen

Of all participants, 9% showed danger signs on admission. The respiratory rate on admission varied between age groups; the mean respiratory rate on admission was 59.3±12.2 in infants, 54.2±12.6 in toddlers, 48±9 in preschoolers and 46.4±12 in school-age children. Among study participants, 84% of children received non-invasive respiratory support. The majority, 349 (89.9%) of the children, received ceftriaxone, and of these, approximately 83% recovered from their illness.

Comorbid medical diseases before or after admission

About 76% of the children had comorbidities, with hyperactive airway disease being the most common comorbidity, followed by severe acute malnutrition. Other medical comorbidities included moderate acute malnutrition, tuberculosis, meningitis, HIV, congestive heart failure and general developmental delay, which affected 96 (24.7%).

Treatment outcome and incidence rate of recovery from SCAP

About 91.5% (95% CI: 88.3% to 94.1%) of participants recovered from SCAP, 17 (4.4%) died and 10 (2.6%) referred to other facilities with an overall recovery rate of 11.5 per 100 person-day observations (95% CI: 10.37 to 12.76). The length of hospital stay ranged from 1 day to 57 days with the overall median length of stay of 6 days (IQR: 4–10).

Kaplan-Meir survival estimates for SCAP recovery time

In all observations, the median time to recovery from SCAP varied according to the different predictors, as shown in table 1. The Kaplan-Meier survival curve was used to estimate the survival status of children with SCAP. The curve tends to decline rapidly within the first 10 days, indicating that most children recover from their illness within this period (figure 1). Survival estimates of patients with SCAP differed significantly as a function of their weight-for-age, height-for-age and comorbidity (p value<0.000). A significant difference in recovery time was observed in underweight children at 8 days (95% CI: 7 to 9) compared with their peers at 5 days (95% CI: 5 to 6). The Kaplan-Meier survival curve was used to estimate the survival status of the groups of children with SCAP. A significant difference in recovery time was noted on the curve depending on the presence of other comorbid conditions and whether or not weight was normal for age (figures 2 and 3).

Figure 1

Overall Kaplan-Meier survival estimate of median recovery time among children admitted with severe community-acquired pneumonia.

Figure 2

Comparison of Kaplan-Meier survival estimate for median time to recovery among stunted and normal height-for-age children with severe community-acquired pneumonia.

Figure 3

Comparison of Kaplan-Meier survival estimate for median time to recovery among children with severe community-acquired pneumonia based on presence and absence of comorbidity.

Predictors of recovery time from SCAP

All independent variables were analysed in the Cox proportional hazard analysis with the dependent variable. Variables with a p value of <0.25 in the bivariate Cox regression analysis were selected for the multivariable Cox regression analysis, and a variable with a p value of <0.05 in the multivariable Cox regression analysis was declared a statistically significant association. Thus, age, stunting, duration before seeking care and comorbidity were significantly associated.

The rate of early recovery from SCAP decreased by 46% for each age unit (adjusted HR (AHR): 1.46, 95% CI: 1.16 to 1.84). In terms of nutritional status, recovery from SCAP decreased by 31% for children who were stunted compared with children who were not stunted (AHR: 0.69, 95% CI: 0.52 to 0.91). Similarly, children who sought care after 5 days of illness were 30% less likely to recover sooner (AHR: 0.70, 95% CI: 0.53 to 0.94) than their peers. Finally, the recovery rate from SCAP was 1.6 times higher in children admitted without comorbidity than in their peers (AHR: 1.59, 95% CI: 1.21 to 2.07) (table 2).

Table 2

Predictors associated with recovery time from SCAP among paediatrics patients in selected governmental hospitals in Addis Ababa, Ethiopia (n=388)

Discussion

This study investigated the median recovery time after SCAP and determined predictors of recovery time in children. The median recovery time after SCAP was 6 days overall (IQR: 4–10). According to the British Thoracic Society guideline for the treatment of SCAP, this finding was long, as the average recovery time in the guideline is 3 days.11 Compared with studies conducted in Debre Markos and Gonder, the recovery rate and recovery time were lower and longer, respectively.2 21 This difference could be due to the fact that in the current study area, debilitated children from all over the country with complicated, chronic and fatal cases were referred to these hospitals due to the advancement in healthcare delivery and practice, resulting in longer recovery time and higher mortality rate. In a study conducted in Jimma, Ethiopia, the median recovery time for 76% of children was 3 days, which was twice as long as the results of this study.27 The differences in the sample size and sociodemographics could be the possible reasons for the discrepancy.

In a study at a rural health centre in Gambia, the mean recovery time was 4.5 days, which is shorter than in this study.16 Differences in socioeconomic status, quality of care and health choice behaviour of the community could contribute to this discrepancy. Similarly, the recovery time in this study is longer than in Vanderbilt (2.3 days) and Nepal (2 days).24 28 Differences in severity, case mix and comorbidity may explain this discrepancy. However, the recovery time in this study was shorter than in the study in Poland (10.1 days).29 This could be due to the fact that the study was conducted between the years 2007 and 2011 (variation in study period).

In this study, younger children were reported to recover faster than older children. This finding is consistent with the results in Debre Markos, Vanderbilt and Italy.2 28 30

Compared with children with normal nutritional status, stunted children take longer to recover. This is because malnutrition exacerbates the effects of disease, leading to more severe flare-ups, complications and a longer recovery time.31 This finding is consistent with the study conducted in Debre Markos and Gonder.2 11

Another important predictor of the median time to recovery from SCAP was the time to seek medical care. Children who presented to the healthcare facility early (≤5 days) recovered faster than those who presented late. This is because early diagnosis and treatment of pneumonia reduces the likelihood of developing complications leading to prolonged hospitalisation and poor treatment outcomes, and this finding is consistent with study results in Debre Markos and Gambia.2 16

Finally, children with comorbidity take longer to recover compared with children without comorbidity. This is because the immune system of children suffering from several diseases at the same time is significantly weakened, which prolongs the duration of hospitalisation and delays recovery. This result is relatively consistent with the results of other previous studies in Debre Markos, Gambia and Nepal.2 11 16

In general, the strongest side of this study was the recruitment of study participants from different facilities in the study area to increase the representativeness of the sample. In this study, younger age and early presentation to a healthcare facility were shown to shorten recovery time. On the other hand, the presence of comorbidities and stunting prolonged recovery time. Therefore, measures that increase the recovery rate and shorten the recovery time, such as primary prevention to eradicate malnutrition and increase healthcare utilisation in the community, should be strengthened. However, the study was not without limitations as it was conducted retrospectively based on secondary data and failed to include some variables, particularly sociodemographic and socioeconomic characteristics of parents and caregivers’ perceptions of SCAP which could be potential predictors of the outcome variable. Therefore, future researchers should consider a prospective cohort study design to include other factors such as sociodemographic and socioeconomic characteristics of parents and caregivers’ perceptions of SCAP.

Data availability statement

Data are available upon reasonable request.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

An ethical clearance was obtained for this study from the institutional review board of the School of Nursing and Midwifery, College of Health Sciences, Addis Ababa University. This board gave an ethical clearance letter with protocol number '046/SNM/21'. After that, a written permission letter to access patient’s charts was obtained from responsible body of each institution. The data generated from the medical records of children were deidentified in order to maintain privacy and anonymity of the study participants. This study adhered to the ethical principles for research in human subjects stipulated in the Declaration of Helsinki.

Acknowledgments

The authors would like to thank College of Health Sciences, Addis Ababa University, for their financial support during data collection. The authors would also like to thank the staff of Tikur Anbessa Hospital, Menelik II Hospital, Zewditu Memorial Hospital and Yekatit 12 Hospital for their co-operation and the supervisors and data collectors. Finally, the authors would like to thank the institutional repository of Addis Ababa University for preserving the thesis and allowing retrieval of this article from the thesis (http://etd.aau.edu.et/handle/123456789/28584).