Background: Postpartum hemorrhage and neonatal asphyxia are leading causes of maternal and neonatal mortality, respectively, that occur relatively rarely in low-volume health facilities in sub-Saharan Africa. Rare occurrence of cases may limit the readiness and skills that individual birth attendants have to address complications. Evidence suggests that simulator-based training and practice sessions can help birth attendants maintain these life-saving skills; one approach is called “low-dose, high-frequency” (LDHF). The objective of this evaluation is to determine the facilitating factors and barriers to participation in LDHF practice, using qualitative and quantitative information. Methods: A trial in 125 facilities in Uganda compared three strategies of support for LDHF practice to improve retention of skills in prevention and treatment of postpartum hemorrhage and neonatal asphyxia. Birth attendants kept written logs of their simulator-based practice sessions, which were entered into a database, then analyzed using Stata to compare frequency of practice by the study arm. The evaluation also included 29 in-depth interviews and 19 focus group discussions with birth attendants and district trainers. Transcripts were entered in Atlas.ti software for coding, then analyzed using content analysis to identify factors that motivated or discouraged simulator-based practice. Results: Practice log data indicated that simulator-based practice sessions occurred more frequently in facilities where one or two practice coordinators helped schedule and lead the practice sessions and in health centers compared to hospitals. The qualitative data suggest that birth attendants who practiced more were motivated by a desire to maintain skills and be prepared for emergencies, external recognition, and establishing a set schedule. Barriers to consistent practice included low staffing levels, heavy workloads, and a sense that competency can be maintained through routine clinical care alone. Some facilities described norms around continuing education and some did not. Conclusions: Designating practice coordinators to lead their peers in simulator-based practice led to more consistent skills practice within frontline health facilities. Ongoing support, scheduling of practice sessions, and assessment and communication of motivation factors may help sustain LDHF practice and similar forms of continuing professional development. Trial registration: Registered with clinicaltrials.gov #NCT03254628 on August 18, 2018 (registered retrospectively).
The study data presented here represent a sequential explanatory design [22], with the qualitative data collection following the quantitative data collection to help interpret the quantitative findings (Fig. 1). Records of providers’ participation in LDHF practice sessions were analyzed descriptively to understand fidelity to the intervention across study arms and facility types. Timeline of study intervention components and data collection presented in this manuscript First, with regard to the main study described in detail elsewhere [21], the study took place in 12 districts at 125 health facilities in the Western and Eastern regions of Uganda in 2014–2016. All public maternities in selected districts participated, including health centers III, which had lower monthly birth volume and were staffed by a few midwives, and health centers IV and hospitals, which had higher volume and staffing. All facilities in all study arms received brief onsite training sessions. Ugandan midwives certified as trainers led the 1-day HMS BAB training session at each facility, and all maternity unit staff were invited to participate, not just doctors, midwives, and nurses. The session included the expectation that maternity unit staff practice on a simulator for 10–15 min once per week for the following 8 weeks. After the 2-month practice period ended, the same trainers led the 1-day HBB training session and again suggested practice with specific scenarios for 10–15 min once per week for 8 weeks. For an additional 4 weeks, maternity unit staff were asked to practice scenarios including care for both mother and newborn. In study arms 2 and 3, practice sessions were facilitated by one of the birth attendants at the facility designated as the peer practice coordinator who received an extra half-day orientation, while in study arm 1, no one was assigned to coordinating or leading practice sessions (Fig. 1). Local stakeholders preferred to use the term “clinical mentor.” For clarity, we are referring to the clinical mentors as “practice coordinators” in this paper. Practice coordinators received no extra compensation but were reimbursed for travel costs to deliver program monitoring and evaluation records. In study arm 3 only, district trainers were asked to call peer practice coordinators to encourage practice sessions, and mobile phone airtime credits were provided. The study team provided detailed practice session plans to all facilities’ coordinators and health care workers.1 These sessions were recommended to occur during quiet periods on the labor ward or at the beginning or end of a provider’s shift. Practice could happen in small groups or individually with the practice coordinators. Many of the practice sessions provided opportunities for all maternity unit staff to practice together, not just birth attendants. This included recognizing warning signs, using clear communication, maintaining stock and correctly storing materials, and retrieving drugs and supply materials for birth attendants if needed. Three structured study tools were used to collect quantitative data. First, study staff listed all maternity unit staff who “perform or assist” deliveries at the facility, including names and cadres, using a paper form. This list was updated before each training session and at two subsequent data collection points: after the 6-month period of HMS BAB and HBB and after 20 weeks of practice (called midline). Second, each peer practice coordinator, (in study arms 2 or 3) or another designated person within the facility, (in study arm 1) was given a paper logbook on which to record practice sessions and asked to document practice session details, including date, provider’s name and cadre, topic practiced, and length of session. District trainers collected the HMS practice logs during the HBB training session. After finding low compliance with record-keeping around practice documentation after HMS, project staff reminded practice coordinators (study arms 2 and 3) and designated persons (study arm 1) of the importance of maintaining practice logs. Data collectors retrieved the HBB logs during the midline data collection. Project data entry staff entered the information from paper logs to a Microsoft Access database created with internal consistency checks. Other project staff checked the original paper records against the database. This database was merged with a list of all birth attendants assigned to each facility’s maternity unit, using the provider’s name and ID number, through a combination of the merge functions in Stata version 15 [23] and manual checks by study staff. Third, in study arm 3 only, district trainers were asked to telephone the practice coordinators to encourage them to hold practice sessions and to keep written records of their calls. Call records were entered into a Microsoft Access database and analyzed in Microsoft Excel. Several indicators were compared across study arms: (1) the proportion of facilities submitting a provider practice log following HMS and HBB and the combined practice period, (2) the mean number of practice sessions and 95% confidence interval per month per provider over 2-month periods (July–August, September–October, and November–December) for each type of training content2, (3) the total number of practice sessions per provider (for HMS and HBB combined, over the project duration) was calculated at the facility level, and the mean for all facilities, (4) the mean and 95% confidence interval of total practice sessions per provider stratifying by the following factors: (a) higher (above the total mean) or lower documented practice, (b) higher level of facility (hospital or health center IV) or lower level, (c) study arms with a practice coordinator (study arms 2 and 3), or study arm 1. All quantitative data analysis was completed using Stata version 15. In May and June 2015, focus group discussions and semi-structured interviews were conducted with nurses, midwives, and health facility in-charges in 24 selected facilities, in order to gather contextual information about the project intervention and rich descriptions of the training and practice sessions. The sample was stratified based on the study arm and level of simulator practice sessions achieved, and a preliminary analysis of whether the facility had achieved higher- or lower-than-average practice levels. Districts and then facilities were purposively sampled, using data from the practice logs and study team members’ knowledge of contextual factors. Each group of 12 high- and 12 low-practice facilities included different levels of health facilities; the hospital from each district was included. The sample size was selected with the aim of achieving saturation in the higher and lower practice subgroup in each study arm. The study team contacted facilities prior to the day of data collection to schedule the focus group discussions; however, sometimes communication gaps and scheduling challenges—such as providers going back to school, on maternity leave, and being out for illness or on leave—limited data collection. The study team did not track refusals, but refusals were unusual because the study team adjusted their activities to accommodate participants’ schedules. Planned and achieved data collection is presented in Table 1. Within each facility, data collectors invited all staff who assisted with or conducted deliveries to participate in the focus group discussion. They recruited the practice coordinator (or facility in-charge, in control districts) in each selected facility and the district trainer in each district to participate in in-depth interviews. Two Ugandan researchers with graduate-level education in qualitative research conducted the interviews and focus group discussions in English using an interview guide that had been pre-tested in facilities outside the study area. Interview and focus group discussion topics included experiences with the two training sessions, perceived challenges in completing practice sessions and strategies to overcome those challenges, recommendations for improving the LDHF approach, and any perceived changes in clinical practice following the training and practice sessions. Interviews and focus group discussions lasted about 1 h. For the participants’ convenience, data collection took place at or near the health facility where they worked, in a setting that enabled auditory privacy, without any non-participants present. No follow-up interviews or focus group discussions were conducted. The interviewers were not involved in other aspects of study implementation and received a 2-day orientation related to the qualitative component’s objectives, tools, consent process, research ethics, recruitment and interview, and focus group moderation techniques. The interviewers made field notes on the day of data collection and, within a few weeks of the interviews, transcribed their own audio recordings verbatim. Transcripts were spot-checked for quality against the audio recordings by a study co-investigator. Sample size for qualitative data collection, by adherence level to practice guidelines 1Sample was split evenly between Eastern and Western regions. All facilities per study arm were in the same district 2All districts sampled included both high and low practice facilities 3Practice coordinators in control arm had a proxy for the qualitative study Transcripts of focus group discussions and interviews were coded in Atlas.ti version 7 [24] using a codebook based on predetermined and emergent codes. Four study co-investigators and one interviewer coded the interviews following an exercise of coding an initial sample of interviews to achieve consistency. The analysts communicated frequently about how the codes were being applied and the process of coding transcripts. In the interest of reflexivity, investigators made an effort to be aware of own biases, such as being predisposed to think that LDHF is an effective training approach, and our own limitations as outsiders to the participating health facilities. A qualitative content analysis approach was used to identify themes and answer research questions [25], influenced by Friese’s Noticing-Collecting-Thinking approach [26]. After coding each transcript, Atlas.ti enabled generation of query “reports” of all data from similar codes; these reports were produced separately for facilities in the higher and lower practice categories, and by type of facility (hospitals vs. from health centers) and study group (study arm 2 or 3 vs. 1). Analysts conducted multiple reviews of reports and identified themes present across transcripts and determined which themes were expressed more strongly within participant subgroups. Institutional Reviews Boards at the Johns Hopkins Bloomberg School of Public Health in Baltimore, Maryland, United States, and Makerere University in Kampala, Uganda and the Ugandan government approved this study. All participants provided verbal informed consent prior to participation, as per the approved protocol. The study is registered on clinicaltrials.gov #{“type”:”clinical-trial”,”attrs”:{“text”:”NCT03254628″,”term_id”:”NCT03254628″}}NCT03254628.