Context and intervention

Bangladesh’s MOHFW maintains a three-tier system for delivering healthcare services and is managed by two directorates: Directorate General of Health Services (DGHS) in the Health Service Division and Directorate General of Family Planning (DGFP) in the Medical Education and Family Welfare Division.14 19 20 Public sector healthcare service delivery follows the IMCI protocol for management of sick children at the community-level and facility-level. Implementation of the PSBI management guidelines is done by trained paramedics (Sub-Assistant Community Medical Officers (SACMO) and supported by Family Welfare Visitors (FWV)) at the UH&FWC. The SACMO has 3 years training on general healthcare, including child health, and the FWV has 18 months training on midwifery and contraceptive management.19 20 The Upazila Health Complex ((UHC); catchment ~250 000 persons), a subdistrict level hospital providing inpatient care, serves as the primary referral facility for PSBI cases. If a case cannot be managed at the UHC, then the infant is referred to secondary-level or tertiary-level care.16 19

Per the national guidelines, the SACMO is the designated provider to assess the infant following a decision algorithm and provide management accordingly. Infants with PSBI are classified into three subcategories based on severity of illness (figure 2).16 Critical illness (CI) cases require inpatient management; thus, hospital referral with prereferral antibiotics is the only option for these infants. Young infants classified as clinical severe infection (CSI) or isolated fast breathing (IFB) in infants<7 days are eligible for simplified antibiotic treatment (SAT)—when hospital referral is not feasible for families—with injectable gentamicin once daily for 2 days and oral amoxicillin two times a day for 7 days. If referral is not feasible, caregivers of infants with CSI are instructed to return to the UH&FWC the next day for the second gentamicin injection. The FWV may provide the second injection if the SACMO is not present at the UH&FWC. Infants with IFB (<7D) only receive the oral amoxicillin regimen. The SACMO follows up with caregivers over telephone on the fourth day, and if the infant’s condition has not improved, advises the caregiver to seek care at the UHC. On the eighth day of treatment, the family receives a home visit from the Family Planning Inspector (FPI), who are trained as supervisors of frontline workers, to assess treatment compliance and the condition of the infant.16

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Management and treatment pathway for young infants with PSBI assessed and treated at UH&FWCs according to Bangladesh’s infection management guidelines. *Critical Illness signs: unconscious, convulsions or history of convulsions, unable to feed at all, no movement on stimulation, unable to cry, persistent vomiting, bulging fontanelle, cyanosis. †Clinical severe infection signs: severe chest in-drawing, hypothermia (<35.5°C), fever (>37.5°C), movement only on stimulation, not feeding well (based on history and observation). ‡Isolated fast breathing (<7 days): respiratory rate equal to or greater than 60 breaths per minute as the single sign of infection in infants aged <7 days. §UHC is the primary referral facility; however, the infant may be referred to the district hospital if it is closer to the UH&FWC. ¶The SACMO counsels the caregiver on providing medicine to the infant at home and prescribes oral amoxicillin two times a day for 7 days. FPI, Family Planning Inspector; PSBI, possible serious bacterial infection; SACMO, Sub-Assistant Community Medical Officers; SAT, simplified antibiotic treatment; UHC, Upazila Health Complex; UH&FWC, Union Health & Family Welfare Centres.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

MaMoni HSS support to PSBI case management and programme monitoring data on referral decision, treatment and follow-up of PSBI cases (0–59 days) during the period of October 2015–December 2016. CI, critical illness; CSI, clinical severe infection; IFB, isolated fast breathing; HSS, FPI, Family Planning Inspector; MaMoni HSS, MaMoni Health System Strengthening project; PSBI, possible serious bacterial infection; MOHFW, Ministry of Health and Family Welfare; SACMO, Sub-Assistant Community Medical Officers; UHC, Upazila Health Complex; UH&FWC, Union Health & Family Welfare Centres.

Implementation strategies and data collection

In August 2015, implementation of the new guidelines started in 53 UH&FWCs of 10 subdistricts within 3 districts of Bangladesh—Lakshmipur, Noakhali and Habiganj. We provided support to the MOHFW to ensure facility readiness, build capacity of providers and promoted awareness and community engagement with the public sector healthcare system (figure 1). Additional details on our implementation strategies are provided in online supplementary table 1. Project staff extracted data from UH&FWC registers maintained by the SACMO during monthly supervisory meetings at the subdistrict hospital. The information collected included age and sex of the infant, illness classification, referral decision, gentamicin treatment and follow-up. Data on oral amoxicillin treatment were not available.

Our analysis followed an explanatory sequential mixed methods design where qualitative data were analysed to help explain the quantitative results.29 We analysed data for 4597 sick young infants collected over a 15-month period during early implementation (October 2015–December 2016). A total of 4590 records were included after excluding records missing age (n=5) and illness classification (n=2). Qualitative data included three focus group discussions with SACMOs (n=2) and FPIs (n=1) to explore providers’ opinions about guideline implementation; key-informant interviews with MOHFW managers at the national and subdistrict levels and documentation of programme learnings by project staff. We used the WHO PSBI Monitoring and Evaluation Framework12 to identify key indicators and guide our analysis of implementation research outcomes (figure 1). We performed χ² tests of independence to examine the relation between PSBI classification, referral feasibility and prereferral treatment.

Reaching families to promote awareness and care-seeking for young infant illness

We oriented government providers in the community and an existing pool of unpaid Community Volunteers (1 per 250 persons) on the guidelines to promote awareness of services and increase referrals to primary health facilities. The Community Volunteer programme model required minimal inputs from MaMoni HSS and was largely sustained by community ownership and coordination between the volunteers and government frontline workers (online supplementary table 1).

We estimated capture of PSBI cases, trends in utilisation and source of community referrals to assess reach and acceptability of the programme. The number of PSBI cases by month remained low, but there was an increasing trend in the total number of young infants identified at the UH&FWC over the programme period (figure 4). The most frequently cited source of referral was a Community Volunteer (47%) followed by self-referrals (36%) (table 1). Based on the PSBI incidence rate for young infants in South Asia (95.4/1000 live births), we estimated that 11.2% (95% CI 10.3 to 12.1) of the expected PSBI cases in the programme area sought care from UH&FWCs during this 15-month period.3 30 Qualitative findings with providers, MOHFW managers and project staff attributed low utilisation rates to lack of awareness of services available at the UH&FWC and caregivers’ preference for private sector providers.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Utilisation of the UH&FWC for young infants and PSBI cases identified from October 2015 to December 2016. PSBI, possible serious bacterial infection; UH&FWC, Union Health & Family Welfare Centres.

Reflections on programme learnings and recommendations for scale-up

Ensuring facility readiness to provide PSBI services was feasible due to multistakeholder engagement and collaboration, which has been shown to be critical for implementation research and scale-up of interventions.17 18 31 Barriers varied by facility depending on managing MOHFW Directorate, either DGFP or DGHS, because each has separate supply chains from the national to subdistrict levels. Most of the programme area UH&FWCs are under administrative management of DGFP, which did not have supply of injectable gentamicin required for PSBI management in the local supply chain. While DGHS had injectable gentamicin available, there was not an adequate supply available for distribution to the UH&FWCs. To support plans for scale-up, we collaborated with stakeholders, including MOHFW managers in both Directorates, to identify bottlenecks and plan for the necessary financing in absence of project support. As a result, in the current health sector plan, both Directorates have secured the necessary budget for the procurement of essential commodities for PSBI management in UH&FWCs, incorporated plans for capacity building and allocated resources for training of providers in PSBI management guidelines.14 Following global guidance, the government has also undertaken initiatives to align the PSBI guidelines with the existing IMCI programme, including reporting through District Health Information System 2 for those UH&FWCs that are managed by DGHS.12 13 A similar electronic information system is under development, with support from MaMoni HSS, for UH&FWCs managed by DGFP. Both electronic systems present an opportunity for reducing the recordkeeping burden for SACMOs, improving monitoring practices and quality of reporting.

Provider adherence to the guidelines varied by infection classification and referral feasibility. Substantial time in training and supervision focused on management of CSI cases when referral was not feasible because these infants require two gentamicin injections and follow-up on the fourth and eighth day. Therefore, providers may have prioritised the outpatient management of CSI cases, indicated by the high level of adherence to treatment and follow-up of these cases. Prereferral antibiotic treatment was significantly lower for cases that accepted referral. From a clinical perspective, referral without prereferral antibiotics is a life-threatening lapse in care. Newborns with severe infections can deteriorate rapidly if left untreated; thus, timely treatment with antibiotics is critical.32 33 Provider reservations about providing injections to critically ill infants may contribute to this lapse in care. Strategies to improve providers’ confidence to provide prereferral injections, and emphasis on follow-up of all PSBI cases, should be included in future training and supervision sessions.

Uncertainty around treatment outcomes, especially for referred cases, is a major barrier to monitoring the long-term effectiveness of the programme. The guidelines recommend that SACMOs follow-up with the caregiver, or hospital providers, to ensure families reach the referral facility; however, this follow-up is not recorded in the registers, so the health outcomes of these infants are not routinely tracked. Families that accept referral are more likely to have infants with serious illness and in the current system are more likely to be lost to follow-up. Interventions to mitigate the financial burden of hospital referral, including transfer arrangements, could improve feasibility of referral for families. Systems for tracking referral cases need to be established and tested during rollout of the guidelines, and indicators for monitoring the care and outcomes of these cases should be integrated into routine systems. For PSBI cases receiving outpatient treatment, the day 8 follow-up visit is critical for assessing treatment outcomes. However, we found this visit was not regularly happening. We responded to this barrier by facilitating follow-up during the programme period, which allowed for these visits to occur, but limited our ability to assess the functionality of these visits in the routine system. This illustrates a common trade-off with implementation research, which others have suggested could be balanced by studying interventions within the context of multiple settings.17 34 Further investigation into improving functionality of these visits is needed as this strategy is scaled-up in Bangladesh and within multicountry contexts to improve external validity.

Since our data are from early implementation, we anticipated low rates of capture of PSBI cases at UH&FWCs. Qualitative findings attributed low utilisation to caregivers’ preference for private sector providers and lack of awareness of services available at the UH&FWC. Bangladesh’s private sector offers high-end services for wealthy families, while a large economy of informal providers retails services in the community.19 Previous studies on care-seeking for neonatal illness in Bangladesh have identified a preference for informal providers as families may perceive their practices as more convenient, affordable or culturally appropriate.19 35–37 In the early stages of implementation, it is important to focus on number of cases identified and trends in utilisation.12 Community Volunteers were frequently cited as a source of referral, which indicates penetration of our community engagement programme. Increasing trends in utilisation suggest improved awareness and acceptability of UH&FWC services among families of young infants as the programme matured, which over time, may contribute to a higher capture of PSBI cases in the community.

Through our reflection on programme leanings, we identified some important limitations in our study data. Programme monitoring was based on facility records and interviews with providers, which limited our ability to investigate barriers to programme feasibility and acceptability from the caregivers’ perspective. Future studies should include interviews with caregivers to collect data on predictors of newborn infections, explore the complexities around referral feasibility and adapt strategies to improve access to and acceptability of services. Missing data in facility records limited our ability to draw conclusions about the effectiveness of simplified treatment and compare treatment outcomes across subcategories. Indicators on treatment and follow-up of cases were especially vulnerable to this limitation because it was not possible to determine in the analysis phase if blank fields indicated that treatment/follow-up was not completed, or if the fields were blank due to recording errors. There is variation among providers in how they record and report data, which was not controlled for in our data collection. Therefore, our data are subject to reporting bias and performance on care indicators should be interpreted judiciously. Future programme efforts should review register completeness during supervision, or periodic programme review, to promote accountability and address challenges associated with incomplete registers in a timely manner.