Prevention and treatment of suspected pneumonia in Ethiopian children less than five years from household to primary care

Abstract Aim Ethiopia has implemented the integrated community case management to reduce mortality in childhood diseases. We analysed prevention, care seeking and treatment of suspected pneumonia from household to health facility in Ethiopia. Methods Analyses were based on a survey in four regions that included modules covering 5714 households, 169 health posts with 276 health extension workers and 155 health centres with 175 staff. Caregivers of children aged 2‐59 months responded to questions on awareness of services and care seeking for suspected pneumonia. Pneumonia‐related knowledge of health workers was assessed. Results When a child had suspected pneumonia, 46% (95% CI: 25,68) sought care at health facilities, and 27% (95% CI: 12,51) received antibiotics. Forty‐one per cent had received full immunisation. One‐fifth (21%, 95%: 19,22) of the caregivers were aware of pneumonia treatment. Sixty‐four per cent of the health extension workers correctly mentioned fast or difficult breathing as signs of suspected pneumonia, and 88% suggested antibiotics treatment. Conclusion The caregivers' awareness of suspected pneumonia treatment and the utilisation of these services were low. Some of the health extension workers were not knowledgeable about suspected pneumonia. Strengthening primary health care, including immunisation, and enhancing the utilisation of services are critical for further reduction of pneumonia mortality.

Ethiopia achieved a two-thirds reduction of under-five mortality during the Millennium Development Goals era. 3 Still, pneumonia remains the number one killer of children below the age of 5 years in this country. 4 Following the WHO recommendation, the integrated community case management of childhood illnesses was adopted in 2010 as a national programme for the implementation at the community level through health extension workers. The goal was to enhance the access to treatment for most of the socio-economically impoverished and hard to reach communities. However, the 2016 Ethiopia Demographic and Health Survey reported that less than a third of caregivers sought health care for their child with suspected pneumonia, and a very few (3%) received antibiotic treatment. 3 Studies that investigated the background to the low utilisation of integrated community case management of childhood illnesses services have identified demand as well as supply-side factors. The demand-side factors included caregivers perceived poor service quality provided at health post, 4 lack of knowledge on childhood danger signs and available services, and care-seeking preferences. 5 The supply-side factors included the health extension workers' lack of expertise in treating suspected pneumonia cases. 6 Further, the unavailability of essential drugs and the health extension workers' inadequate supervision and training also contributed to the low utilisation of care. 7  This study aimed to describe the prevention, care seeking and treatment of suspected pneumonia from households to health facilities at the start of the intervention.

| Study setting, design and population
The basis of Ethiopia's three-tiered health system is the primary healthcare unit, composed of five health posts and their affiliated health centre. Around 37 000 health extension workers stationed at 18 000 health posts to provide preventive, promotive and essential curative services. This cross-sectional study was part of the larger The sample size was estimated using a double population proportion formula for the evaluation of the effectiveness of the Optimising the Health Extension Programme intervention. The following assumptions were employed: the proportion of children less than five years in the surveyed households was 0.65, and 7% of the children were expected to have suffered from suspected pneumonia during the last 2 weeks. 9 The design effect was set at 1.3, 5% level of significance, and 80% power was used to detect any differences of 10-20 percentage points of appropriate care seeking for different common childhood illnesses between baseline and end-line surveys in intervention and comparison areas. Hence, a total of 6000 households were expected to include 3494 children aged 2-59 months and 245 cases of suspected pneumonia in children less than five years.
All caregivers of children aged 2-59 months, who resided in the study districts, were included in the study. For every cluster with 30 households, the corresponding health post and health centre, together with their respective health extension workers and health centre staff serving these clusters, were included in the study.

| Data collection
The structured questionnaires used for data collection were developed based on previous literature and major survey instruments.
This study was based on one survey that included different questionnaire modules for selected households, health posts that were served by health extension workers, and health centres with their healthcare staff, that is health officers and nurses. The survey questionnaires and data collection tools and procedures were piloted and accustomed to the local context and also translated into the local languages, that is Amharic, Tigrigna and Oromiffa. Data were

Key notes
• The Ethiopian Ministry of Health invests in improving care seeking for pneumonia and other common childhood illnesses.
• Based on a survey, we show low utilisation of services for children with suspected pneumonia, inadequate coverage of pneumonia-relevant vaccinations and insufficient knowledge about suspected pneumonia among primary healthcare workers.
• The Ethiopian primary healthcare system needs to improve the prevention and management of suspected pneumonia. collected using electronic tablets with logical controls to minimise inaccuracies. There were fifteen data collection teams, and each team comprised of a supervisor and seven data collectors. Data collectors and supervisors held a minimum of first academic degree and were trained for 2 weeks on data collection techniques, procedures, quality assurance and ethical considerations of the study. A supervisor revisited a small sample of households from each cluster to ensure data quality, and the consistency of the original and a repeat interview was compared. Details of the study protocol and tool development are described elsewhere 8 and have been registered as ISRCTN12040912.

| Measurements
In this study, suspected pneumonia was defined as a child aged 2-59 months who had cough combined with either fast or difficult breathing due to chest problems within the 2 weeks before the survey. 3 Thus, care seeking was defined as a child with suspected pneumonia for whom advice or treatment was sought from a relevant care provider that included government health facilities or private providers. Moreover, the treatment of suspected pneumonia with antibiotics was estimated as the percentage of children with suspected pneumonia who sought care from an appropriate health provider and received antibiotics.
Child immunisation and vitamin A supplementation are key strategies to prevent pneumonia or mortality from pneumonia. 2,4 The vaccination status of children aged 12-23 months was primarily assessed by reviewing immunisation cards. If such cards were absent, the caregivers were asked to report the type of vaccines their children had received. Different probing techniques were used by the data collectors to minimise bias, such as considering the route and timing of the vaccine administration and the potential benefit of the vaccine.
Full immunisation was estimated as the proportion of children aged 12-23 months who had received BCG vaccination, three doses of pentavalent vaccine, three or more doses of the oral polio vaccine and measles immunisation. Immunisation with the pneumococcal conjugate vaccine was estimated as the proportion of children aged

12-23 months who had received three doses of this vaccine. Vitamin
A supplementation coverage was determined as the proportion of children aged 6-23 months who had received vitamin A supplement within 6 months before the survey.
The caregivers' awareness of the availability of pneumonia treatment was estimated as the proportion of caregivers who had heard messages regarding pneumonia treatment. The socio-economic status was represented by a wealth index that was generated by principal component analysis for each household based on ownership of assets. The households were after that categorised into quintiles from the poorest to the least poor.
The pneumonia-related knowledge of the health extension workers and health centre staff was assessed by asking these health workers to mention signs and management of suspected pneumonia, general danger signs of illness, feeding problems and acute malnutrition. The health workers' knowledge of suspected pneumonia was estimated as the proportion of health extension workers or health centre staff that correctly mentioned cough combined with fast or difficult breathing as signs of suspected pneumonia. The knowledge of signs of severe pneumonia also included stridor or chest in-drawing.
The health facility-level treatment of childhood suspected pneumonia was assessed at the health post and health centres by reviewing the registers of sick children 2-59 months during the last 3 months before the survey. Antibiotic treatment was described as the proportion of children registered to have suspected pneumonia who had received antibiotics.

| Data analysis
Descriptive statistics, that is proportions or means along with their corresponding 95% Confidence Interval (CI), were used to summarise the characteristics of the study participants and health facilities.
The analyses were performed using STATA version 14 (STATA, Corp). Written informed consent was sought from each participant in the household interview after providing information about the study to each household representative and caregiver. Similarly, written informed consent was also secured from the health extension workers and health centre staff during the health facility data collection.

| Characteristics of caregivers and their children
In this baseline survey, 194 clusters were included. The remaining six clusters were excluded because of social unrest. In these clusters, 2532 caregivers with 3110 children were interviewed. A bit more than half of the caregivers had no formal education. Almost a quarter of the children were aged 2-11 months (Table 1).

| Caregivers' awareness of healthcare services, care-seeking preferences and utilisation
Less than a third (28%, 95% CI: 26,30) of the caregivers had heard about pneumonia treatment, mainly from health centres or health extension workers at health posts. A majority (89%) of the caregivers were aware of the availability of a health post in their community, and 65% stated that they had <30 minutes' walk to reach the nearest health facility. One-third (33%) of the caregivers had visited the health post one or more times during the last 6 months. A fifth (21%, 95% CI 19,22) of the mothers preferred to consult the health extension worker if their children had symptoms of suspected pneumonia, while a majority (86%) preferred to seek care at the health centre (Table 1).

| Healthcare providers' training, supervision and knowledge
A majority (83%) of the health extension workers were trained in the integrated community case management programme, and 78% had received supportive supervision during the 6 months before the survey. In these supervision sessions, 57% had discussed the diagnosis or treatment of pneumonia and 67% about acute malnutrition.    (Table 2). Most of the children with suspected pneumonia received antibiotics at the health post and health centres, Table 3. Three-fourth of children with suspected pneumonia received amoxicillin at the health centres, and fewer were treated at the health posts. The use of co-trimoxazole for treating pneumonia was more frequent at health posts than at health centres.

| D ISCUSS I ON
We have shown that Ethiopian caregivers' awareness of the avail-

| Household-level knowledge, care-seeking preference and utilisation
We found that only a quarter of caregivers were aware of pneumonia treatment, which was considerably lower than similar studies in other low-income countries. 11 This low awareness could be explained by limited interaction between the health workers and caregivers.
The caregiver's awareness of the existence of pneumonia treatment was more limited among those who had never received a home visit by the health extension workers or had never visited the health post.
Home visiting by health professionals is considered a cost-effective strategy to enhance community awareness of childhood illnesses 12 and improving utilisation of services. 13 Visiting health facilities also pneumonia, was also shown by a previous national study. 15 The low reported preference for the health extension worker might be related to the caregivers' perception of poor quality of care provided by these health workers coupled to a poor awareness of the availability of pneumonia treatment. 13 In a previous study, mothers questioned the competence of the health extension workers. 16 Those who cannot afford treatment at alternative health facilities used the services of the health extension workers at the health post. 5 We also found a low coverage of child immunisations, partic- This study showed that less than half of the caregivers sought care when their children suffered from suspected pneumonia. This level of care utilisation is low as compared to reports from other low-and middle-income countries. 19 The reported low level of care utilisation could be attributed to the caregivers' lack of awareness of available pneumonia treatment 20 and perceived poor quality of service. 21

| Knowledge of healthcare providers and treatment of pneumonia
Our knowledge assessment of healthcare providers illustrated that two-thirds of the health extension workers correctly mentioned the signs of suspected pneumonia: nevertheless, less than half recalled signs of severe pneumonia or general danger signs. These findings imply that they lacked the knowledge to classify childhood suspected pneumonia correctly. This problem was further illustrated by another sub-study of this project, which showed that the health extension workers had great difficulties incorrectly classifying common diseases such as suspected pneumonia. 22 The reported finding is in line with previously reported inadequate knowledge and management of pneumonia by the health extension workers. 6 Also, other groups of community health workers in African countries reportedly share similar difficulties. 23 Such problems could be related to an insufficient capacity building, including training and supportive supervision. The provision of in-service training and supportive supervision is part of the government's programme. 6,7 Our analysis showed that nearly a fifth of the health extension workers never received appropriate training.
Three-quarters of these health workers received supportive supervision, which was lower than indicated as an acceptable level, that is >90%. 24 The health centre staff who supervises the health extension workers should also have the required knowledge and skills. 25 We found that not all health centre staff were knowledgeable about suspected childhood pneumonia. A previous study also reported a lack of knowledge of the health extension workers' activities among supervisors, 26 indicating a need of refresher training also to supervisors.
This study reported that only a quarter of children in the household survey with suspected pneumonia received antibiotics, which is lower than similar studies in other low-income countries. 19 The facility register review, however, reported that a majority of the children classified to have suspected pneumonia were treated with antibiotics. This difference in reported antibiotic utilisation suggests that caregivers have difficulties in reporting signs and symptoms of suspected pneumonia. 20 At health facilities, amoxicillin was the most frequently prescribed antibiotic. Still, more than a quarter received co-trimoxazole. Cotrimoxazole is less effective in the treatment of pneumonia. 27   This study is part of an evaluation of a complex intervention that aims at increasing utilisation of primary child health services in Ethiopia. 8 In another study of the same project, it was revealed that referral practices of sick children were weak from the health post to health centre, and from health centre to hospital. 28 Children with severe pneumonia were, however, all referred for management at higher level. We showed that the child immunisation coverage was low. Other sub-studies have revealed that vaccination coverage was equitably distributed from a socio-economic 29 and geographic perspective. 30

| CON CLUS ION
The caregivers of children below the age of 5 years had low awareness and utilisation of treatment of suspected pneumonia and pneumonia-relevant vaccinations. Therefore, demand-creation strategies to improve community awareness of suspected pneumonia combined with efforts to enhance the quality of care may increase the utilisation of services. Continuous capacity building, supportive supervision, and mentoring of health extension workers and health centre staff are needed, as well as ascertaining the availability of relevant pharmaceutical drugs.

ACK N OWLED G EM ENTS
We would like to forward our deepest gratitude to the study participants.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data for this manuscript were primarily collected by the