World Health Organization (WHO) clinical guidelines for children are intended to provide evidence-based guidance to health workers in diverse clinical settings, with a focus on low-resource and smaller health facilities [1–3]. The guidelines for pneumonia were updated in 2013 and changes in severity classification and treatment recommendations for children 2-59 months of age were made . The previous WHO pneumonia classification separated pneumonia into non-severe, severe, and very severe categories . Pneumonia with chest indrawing in children aged 2-59 months was considered at least “severe” with parenteral antibiotics being recommended. The 2013 revision described “pneumonia” and “severe pneumonia” classifications and included children with chest indrawing and no other danger signs (chest-indrawing pneumonia) in the pneumonia group that could be safely managed outside of a hospital . While this change was carried through in revised primary care and hospital guidelines [2,3], the community case management guidelines (iCCM) for community health workers (CHWs) still recommend that CHWs refer children with chest-indrawing pneumonia to a hospital .
The 2013 revisions were informed by evidence that oral antibiotics are equivalent to intravenous antibiotics for most children with pneumonia [7–12] and studies demonstrating that many children with “severe” or chest-indrawing pneumonia could be safely treated at home [13–18].
However, the studies which influenced this guideline change were conducted in limited contexts and under controlled trial conditions. The observed pneumonia mortality rates were very low compared to most low- and middle-income country (LMIC) settings, the rates of wheezing and viral aetiology were high, and large numbers of patients were excluded. This raised concern that they may not be representative of the populations for whom WHO guidelines are intended – particularly in higher mortality contexts with higher rates of bacterial pneumonia and in the absence of the level of monitoring usually associated with clinical trials . While these were set up as non-inferiority trials, the dilution of patient populations with children with self-limiting viral lower respiratory tract infections, such as mild-moderate bronchiolitis, risks lowering the power to detect a difference in treatment outcomes for children with bacterial pneumonia, even in very large studies .
A subsequent study from Kenya sought to evaluate the appropriateness of this change in guidelines, adding to the evidence showing that chest indrawing in sub-Saharan Africa hospital contexts, along with other factors not currently incorporated in the WHO classification of severity of pneumonia (such as moderate pallor and moderate underweight), were significant risk factors for mortality [20,21].
This review aimed to examine all published studies from LMICs which analysed home treatment of chest indrawing pneumonia, in comparison to referral or admission for inpatient hospital treatment, to establish the degree to which we can be confident in recommending such management, and under what conditions.
We conducted a systematic search of Medline, Embase, and PubMed (for articles not yet indexed in Medline) for all relevant articles published since January 1, 2000 (search conducted on September 1, 2020). We mapped search terms to medical subject headings where possible, using Boolean operators to combine searches into our final systematic search query. We used synonyms of “pneumonia”, “chest-indrawing”, “home treatment” and “child” to target the search strategy, with oversight from an experienced medical librarian to ensure all relevant papers were identified. We also searched reference lists of all included references for eligible studies. The specific search terms used for our Medline search and further details of the search strategy, information sources, and data collection processes are included in Appendix S1 in the Online Supplementary Document.
Assessment of study eligibility
We included studies published since 2000 evaluating outcomes for children (aged 28 days to 4 years) with WHO-defined pneumonia with chest-indrawing (“chest-indrawing pneumonia”) treated at home (Table 1). Two reviewers (CW and PW) independently screened the titles and abstracts of all returned studies. We obtained full texts for studies screened by either reviewer, with the two reviewers independently assessing them for inclusion. We resolved disagreements by discussion and, where appropriate, consulted a third reviewer (HG). No reviewer was blinded to the journal titles, study authors, or affiliated institutions.
Table 1. Inclusion and exclusion criteria for studies in this review
ALRI – acute lower respiratory tract infection; LMICs – low- and middle-income countries
We used a standardised data extraction form to extract data relevant to our review. Two reviewers (CW and PW) independently extracted data from each eligible study and entered it into an Excel spreadsheet (Microsoft, Redmond, US). We resolved disagreements by discussion and contacted study authors where appropriate to resolve any uncertainties. We did not attempt a meta-analysis of extracted data as our primary goal was to understand treatment outcomes with respect to population and context. We categorised context, population and outcome data, then qualitatively synthesised results to determine whether children with WHO-defined chest-indrawing pneumonia could be safely managed at home, and in what contexts.
Assessment of study quality and risk of bias
We assessed the quality and risk of bias of all included studies by using the Effective Public Health Practice Project (EPHPP) Quality Assessment Tool [22,23]. Using this tool, two reviewers, PW and CW, independently rated studies as strong, moderate, or weak with respect to selection bias, study design, confounders, blinding, data collection method, withdrawals and dropouts, and a global rating. Where disagreements occurred, a third reviewer, HG, carried out a final assessment (Table S1 in the Online Supplementary Document).
Ethical approval for this study was not required.
We retrieved 1521 references from database searches, screened 1131 unique articles and identified 26 articles for full-text screening. Of these, 12 were excluded because it was impossible to extract data specific to patients with chest indrawing pneumonia (n = 8) [24–31] or specific to those managed at home (n = 4) [32–35]. We included 14 studies in the qualitative synthesis (Figure 1).
Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram.
Characteristics of included papers can be found in Table 2 and Table S2 in the Online Supplementary Document. Of the 14 papers, six were randomised controlled trials (RCTs) [13–15,17,36,43], one was a planned subgroup analysis of a larger RCT , six were observational studies [16,18,37,40–42], and one was a case report . Aside from the case report, the number of participants included in each paper ranged from 117  to 13 266 . Half (7/14) of the papers were published after 2015.
Table 2. Settings and PICOS criteria of included studies
CHW – community health worker, iCCM – Integrated Community Case Management, IMCI – Integrated Management of Childhood Illness, IM – intramuscular, IV – intravenous, LHW – lady health worker, RCT – randomised controlled trial, mo – months, d – days
*Different studies used varying definitions of treatment failure, but all definitions included clinical deterioration as well as persistence of chest indrawing and/or fever. See main text and Table S3 in the Online Supplementary Document for details.
The studies took place in 9 lower-middle income countries and 2 low-income countries. Seven studies were predominately in urban settings [14,15,18,36,37,39,41], and seven were in predominantly rural settings [13,16,17,38,40,42,43]. Three studies involved community-based health workers [13,17,42], eight involved primary care clinics [16,18,39,40,43] or hospitals [14,15,18,39,41], and 3 were of “day clinics” [36–38].
All studies included children 59 months of age or younger, except one  which included patients up to 12 years of age. Most inclusion criteria for studies were consistent with pneumonia as per the 2005 or 2014 WHO guidelines [4,5]. Exclusion criteria varied, but children with severe pneumonia (ie, pneumonia with danger signs) were excluded in most studies, as were patients with comorbidities or other conditions for which antibiotics would have been indicated.
Four studies reported vaccination coverage, with generally high coverage ranging from 74% to 97% receiving all age-appropriate vaccines according to local guidelines [14,18,39,43] (Table S2 in the Online Supplementary Document).
Many patients were excluded following screening in most of these studies, with numbers of enrolled patients ranging from 5%  to 45%  of the total number of patients screened. While non-severe or non-chest-indrawing pneumonia was the most common exclusion reason (99.7% of exclusions in one study  were due to non-severe pneumonia), other studies had numerous other exclusion reasons, including one in which consent was refused for 61% of screened children , one in which 25% of screened patients were excluded for a history of three or more episodes of wheeze , and another in which 18% were excluded due to a history of asthma  (Table S3 in the Online Supplementary Document).
Objectives of included studies
All studies either compared management of pneumonia with chest indrawing at home vs hospital management, compared management algorithms that mandated hospital management in one arm and allowed home treatment in another, or were observational studies of home care.
Characteristics of health care resources
Oxygen saturation was documented as having been measured in patients in eight of the 14 studies [14,36–41,43], although in two of these it was only available to patients in one arm of the study [39,43]. All studies had clear protocols for diagnosis, categorisation, management, and referral of pneumonia, though there was variation in how strictly these were adhered to. Most (11/14) studies described follow-up frequency, all requiring review at least twice in the first week, including at least one review by day three (Table S2 in the Online Supplementary Document).
Training and education of caregivers
Ten of the 14 studies described the training procedures for study staff, which ranged from one day  to one week  and included a variety of techniques, such as lectures, videos, practical sessions, and role play, as well as assessment and ongoing supervision. Six studies described caregiver education, which generally included teaching on how to administer medication and review for danger signs. One study reported the use of a video and formal assessment of understanding (Table S2 in the Online Supplementary Document).
The primary outcome of eight studies [13–15,17,18,39,41,42] was treatment failure, however varying definitions of treatment failure were used. By all definitions, clinical deterioration, and/or persistence of fever or chest indrawing were considered indicative of treatment failure (Table S4 in Online Supplementary Document).
Other studies’ primary outcomes were appropriateness of treatment compared to guidelines , successful treatment via a day clinic [36–38], and effect of oximetry on severe pneumonia diagnosis and referrals [40,43].
Trial evidence on clinical outcomes
Of the seven included randomised or cluster-randomised trials, four [13–15,17] directly compared treatment failure in a community setting with referral or inpatient management of chest-indrawing pneumonia (Table 3 and Table 4). Of these, two studies [13,17] were on “lady health worker” (village-based community health worker) programs in rural Pakistan, and two involved urban tertiary care facilities in Pakistan  or India . These four trials found similar or better treatment outcomes for patients in the community management arm relative to the inpatient management or referral arm. The lower risk of treatment failure on intention-to-treat analysis in one trial  became insignificant after per-protocol analysis and may have been influenced by numerous patients in the inpatient management arm voluntarily withdrawing from the study and self-discharging from the hospital.
Table 3. Outcomes of studies comparing home vs hospital management in which primary outcome was treatment failure in patients with chest-indrawing pneumonia
HR – hazards ratio, ITT – intention to treat, PP – per protocol, CI – confidence interval
*In all cases “intervention” refers to home-based management and ‘control’ refers to management in hospital.
†Day 6 treatment failure except where noted.
Table 4. Outcomes of studies with comparison groups other than home vs hospital management, or with primary outcomes other than treatment failure in patients with chest-indrawing pneumonia
RR – relative risk, OR – odds ratio, aOR – adjusted odds ratio, mo – months
One other randomised trial from urban Bangladesh  found no difference in treatment outcomes of patients with chest-indrawing pneumonia and severe malnutrition treated in a day clinic compared to treatment in a local hospital, as long as patients were referred appropriately when needed. Subgroup analysis of a cluster-randomised trial in urban Tanzania  found that management of pneumonia with an algorithm which included oximetry measurements and point of care testing of CRP enabled more patients to be managed without antibiotics and in the community, with an improvement in outcomes. However, the number of patients with chest indrawing in each arm was low.
These trials reported low treatment failure rates for those treated in the community (median = 7.9%, range = 3.2-11.7), and mortality was very low (median = 0.9%, range = 0-0.2).
These trials included several important safety measures in selecting patients for community-based treatment and monitoring them and responding to deterioration (Table S2 in the Online Supplementary Document). Except for one focussing on children with malnutrition , trials generally excluded children with malnutrition or other chronic illnesses or comorbidities (including anaemia, wheeze, asthma), those who had already received treatment or were re-presenting, those who may be unable to comply with oral treatment (due to vomiting, for example), and anyone who would otherwise meet WHO criteria for severe pneumonia. Except for the two lady health worker programs in Pakistan, all trials included routine assessment with pulse oximetry to identify those with hypoxaemia (low blood oxygen level). After the initial visit, all studies required regular patient review for at least one week, ranging from daily to every 3-4 days. Studies included training for health care workers, including on how to counsel caregivers, although few studies reported this in detail.
Other evidence on feasibility and appropriateness of community-based treatment
Observational studies from a range of African and Asia-Pacific contexts showed similarly low treatment failure rates (median = 7.0%, range = 2.2, 10) and very low mortality (range = 0%-0.6%) among children with chest-indrawing pneumonia treated in the community (Table 3 and Table 4).
A qualitative synthesis of additional data on context, intervention, and processes identified additional lessons on the feasibility and appropriateness of community-based treatment of chest-indrawing pneumonia (Table S5 in the Online Supplementary Document).
The inclusion of pulse oximetry was found to improve pneumonia diagnosis and identification of hypoxaemia in a cluster randomised trial among rural primary care facilities in Ethiopia , an observational study among rural primary care facilities in Malawi , a secondary analysis of an RCT in urban Tanzania , and an observational study in urban Papua New Guinea . The Malawi study found that, if health care workers had followed the revised WHO guidelines (in the absence of pulse oximetry), 42% of children with hypoxaemic pneumonia (SpO2<90%) would not have been referred compared to 8% using the existing guideline (which recommended referral for all those with chest indrawing) . The Tanzania study found that pulse oximetry, alongside point-of-care CRP testing, better identified low-risk patients for outpatient treatment and was associated with better outcomes compared to usual IMCI care . The Ethiopian study found increased adherence to treatment recommendations, including referral follow-through, but no significant effect on outcomes . The Papua New Guinea study also found that pulse oximetry helped caregivers understand and trust care plans provided by health care workers .
While community health workers’ assessments and that of more highly trained health professionals were generally concordant [17,42], children with comorbidities, moderate malnutrition, or delays in care-seeking may be less likely to receive appropriate diagnosis and categorisation . All studies included specific training on risk assessment for participating health care workers.
Studies found mixed effects of community-based treatment guidelines on the number and proportion of patients presenting, being referred, and adhering to treatment recommendations. An observational study in Bangladesh that introduced community-based management of chest-indrawing pneumonia after observing low referral completion rates found increased numbers of children presenting with chest-indrawing pneumonia to primary care facilities, lower referral initiation rates, and minimal change in the number or proportion following through with referral recommendations . Other studies found low rates of referral completion (30%-47%), particularly among those in the hospital-care arm [13,43], and moderate rates (5%) of participants discharged from hospitals against medical advice .
Evidence from four trials suggests similar or lower treatment failure rates among children with chest-indrawing pneumonia treated at home compared to referral or inpatient management. However, these trials were all conducted in Pakistan or India and excluded patients with significant comorbidities, making generalisability limited – particularly in populations where malnutrition or other comorbidities are common and settings where the aetiology and associated features of pneumonia differ.
Evidence from other interventional and observational studies supports the feasibility of home management of chest-indrawing pneumonia, if safeguards are in place, with relatively low rates of treatment failure and mortality. These studies included settings with higher HIV prevalence and likely higher rates of bacterial pneumonia than peri-urban Pakistan and India, where the RCTs were conducted. They also represented care models based in the community, primary care facilities, secondary and tertiary hospitals, and “day clinics”. In all these studies, the conditions for safe outpatient management of chest-indrawing pneumonia were carefully defined. Children with hypoxaemia or danger signs and those with HIV, malnutrition, anaemia, or other comorbidities, were all managed as inpatients. A strong emphasis was placed on appropriate training of the health care staff, education of caregivers, and appropriate and timely follow-up of all patients.
Pulse oximetry was a core part of risk assessment in most included studies (although two of the earlier RCTs did not include oximetry), identifying severely ill children with hypoxaemia who otherwise may not have been appropriately referred or treated with oxygen. This finding concurs with other studies on pulse oximetry that have found pulse oximetry is an objective measure that improves hypoxaemia detection and risk stratification and provides health care workers and patients/families with greater confidence in treatment plans [40,44–49].
These findings are supported by a recent study on the management of chest-indrawing pneumonia by community health workers in Bangladesh, India, Ethiopia, and Malawi, which found equivalent treatment outcomes compared to facility-based care . This study involved modification of community health worker guidelines (iCCM) to include pulse oximetry and allow community-based treatment for children with chest-indrawing pneumonia, using strict risk assessment and follow-up procedures and providing close supportive supervision (including 3-monthly refresher training).
Implications and interpretation
Based on existing evidence, we suggest that home treatment of chest-indrawing pneumonia may be appropriate for low-risk patients with adequate safety and care structures (Table 5).
Table 5. Recommendations for home treatment of chest-indrawing pneumonia for children aged 2-59 mo
Where it is safe and feasible, the management of chest-indrawing pneumonia at home has benefits. Home management is less costly to both the health services and to families, and usually more convenient and acceptable to families . Home management options are particularly valuable in remote geographies and in populations who face substantial barriers in accessing hospital care, with opportunities to improve treatment adherence, including compliance with referrals when they are needed.
However, current community and primary care structures and processes in many LMIC contexts are unlikely to provide the required risk assessment and safety net without substantial investment. Clinical risk assessment is poorly taught and practised, with respiratory signs being one of the most consistently missed elements of existing IMCI practice [53–55]. Pulse oximeters are rarely available or used in primary care settings [40,43,46,56–58] despite inclusion in WHO primary care guidelines and “priority medical device” lists [59,60].
Evidence from the included studies suggests that training, adequate equipment, supportive supervision, and structures for referral and review are essential to implementing home care. The inclusion of a risk assessment checklist may provide additional practical support to health care workers, particularly lower-level health care workers who would not usually be managing more severely ill patients without a referral (Appendix S2 in the Online Supplementary Document).
Implementation of home care for children with chest-indrawing pneumonia will vary between contexts, with this review identifying examples from urban hospitals, rural health facilities, and community settings. The day clinic model presented in three of the papers from Bangladesh is interesting, enabling a hybrid inpatient/outpatient service where patients attend a facility (which has oxygen, suction, and other services) from 8 am to 5 pm each day for parenteral antibiotics, and return to their home at night. However, more research is needed on the use of this model in other contexts.
This review was limited by the number and quality of studies included. We identified few randomised trials, from narrow geographical and epidemiological settings, and their trial conditions may not reflect what is possible in actual practice. To address these issues, we included interventional and observational studies that could provide more ‘real-life’ data on the conditions required to provide safe care in the community. Future studies exploring the implementation of home management of chest-indrawing pneumonia in more routine practice settings and in diverse contexts will give greater clarity and confidence about how to safely manage these children without admission.
Home treatment of chest-indrawing pneumonia can produce treatment outcomes comparable to hospital-based care for carefully selected populations in certain contexts. However, home treatment should only be recommended in children who are low risk and in contexts where adequate care and safety provisions are available. Prospective operational research into home treatment of chest indrawing pneumonia in settings outside of South Asia, particularly Sub-Saharan Africa, would help with determining the generalisability of the findings of existing trials.
Disclaimer: The authors alone are responsible for the views expressed in this publication and they do not necessarily represent the views, decisions, or policies of the World Health Organization.
Acknowledgements: Thanks to Poh Chua, research librarian, for substantial technical support in setting up and running the database searches, and Helen Thomson and Haset Samuel for administrative support.