Study protocol training for life: A stepped wedge cluster randomized trial about emergency obstetric simulation-based training in a low-income country

Background: Globally perinatal and maternal mortality rates remain unacceptably high. There is increasing evidence that simulation-based training in obstetric emergencies is associated with improvement in clinical outcomes. However, the results are not entirely consistent. The need for continued research in a wide variety of clinical settings to establish what works, where and why was recommended. The aim of this study is to investigate the effectiveness of an emergency obstetric simulation-based training program with medical technical and teamwork skills on maternal and perinatal mortality in a low-income country. Methods: A stepped wedge cluster randomized trial will be conducted at the medium to high-risk labour ward at Mulago Hospital, Kampala, Uganda, with an annual delivery rate of over 23,000. The training will be performed using a train-the-trainers model in which training is cascaded down from master trainers to local facilitators (gynaecologists) to learners (senior house officers). Local facilitators will be trained during a four-day train-the-trainers course with an annual repetition. The senior house officers will be naturally divided in seven clusters and randomized for the moment of training. The training consists of a one-day, monodisciplinary, simulation-based training followed by repetition training sessions. Scenarios are based on the main local causes of maternal and neonatal mortality and focus on both medical technical and crew resource management skills. Kirkpatrick’s classification will be used to evaluate the training program. Primary outcome will be the composite of maternal and neonatal mortality ratios. Secondary outcome will comprise course perception, evaluation of the instructional design of the training, knowledge, technical skills, team performance, percentage of ventouse deliveries, percentage of caesarean sections, and a Weighted Adverse Outcome Score. Discussion: This stepped wedge cluster randomized trial will investigate the effect of a monodisciplinary simulation-based obstetric training in a low-income country, focusing on both medical technical skills and crew resource management skills, on patient outcome at one of the largest labour wards worldwide. We will use a robust study design which will allow us to better understand the training effects, and difficulties in evaluating training programs in low-income countries. Trial registration: ISRCTN98617255, retrospectively registered July 23, 2018. The aim of this study is to investigate the effectiveness of a simulation-based emergency obstetric training program on different levels of Kirkpatrick in a low-income country with a stepped wedge cluster randomized trial. A stepped wedge cluster randomized trial will be conducted at the medium to high-risk labour ward at Mulago Hospital, Kampala, Uganda, with an annual delivery rate of over 23,000. A stepped wedge trial is a cluster-randomized trial in which all study groups (clusters) receive the intervention by a computer-generated random sequential roll-out of the training sessions over time (Fig. 2). Stepped wedge design with seven clusters and subsequently eight study periods of each 7 weeks This design has several advantages; all clusters are trained, the effect can be estimated from both within- and between-cluster comparisons, it is possible to control for time effects, and the design is useful when it is preferable to implement the intervention in stages because of logistical, practical or financial constraints [34–37]. However, it may be tough to create fixed clusters of students without contamination during working hours, difficult to plan training sessions during holidays and exam periods, and a challenge to train a new cluster strictly every 7 weeks. Therefore, a prospective pre-post intervention analyses will also be done with 16 months prior to the start of the program and 16 months from the program onwards. The training will be performed using a train-the-trainers model in which training is cascaded down from master trainer to local facilitators to learners. Local facilitators will be a selection of gynaecologists from Mulago Hospital based on their clinical and teaching experience. They will be selected by the head of department and will train and implement new medical technical skills and crew resource management skills in a hierarchical way (top-down). The gynaecologists will be trained during a 4 day train-the-trainers course including a test day with junior house officers (intern doctors) and annual repetition day. Afterwards, the local facilitators will train the senior house officers (SHOs) in their first-, second- or third-year of their gynaecology training program. The SHOs will be naturally divided in seven clusters of approximately six to nine persons and randomized for the moment of training. The local facilitators will be paid for the training sessions. The training consists of a one-day, monodisciplinary, simulation-based training followed by repetition training sessions. Scenarios are based on the main local causes of maternal and neonatal mortality and include crew resource management. Primary outcome will be the composite of maternal and neonatal mortality ratios. Secondary outcome will comprise course perception, evaluation of the instructional design of the training, knowledge, technical skills, team performance, percentage of ventouse deliveries, percentage of caesarean sections, and four items of the Weighted Adverse Outcome Score [21]. The training will be implemented at Mulago hospital, the national referral hospital in Kampala, Uganda and the teaching Hospital for Makerere University. This hospital serves the population living in Kampala and the surrounding districts. The maternity wards of Mulago Hospital include a low-risk ward and a medium to high-risk ward. Each year, over 31,000 women give birth in Mulago Hospital. Over 23,000 women deliver at the medium to high-risk ward. Only staff from the medium to high-risk ward will be included in this study. The staff consists of 45 gynaecologists, 60 SHOs, and 45 midwives. The SHOs differ in level of education towards their gynaecologist training (first-, second-, and third-year SHOs). During a 24-h shift one gynaecologist is on-call duty. During daytime, six SHOs, and eight midwives provide obstetric care while four SHOs and six midwives are available during night. The study concerns a simulation-based training program focusing on medical technical skills and crew resource management in emergency obstetric scenarios in a low-income setting. Except for vacuum extraction and neonatal resuscitation training, no other simulation-based training was done before. The local facilitators will train all first-, second- and third-year SHOs since they have a central and coordinating role in providing emergency obstetric care on the labour ward in this hospital. With over 23,000 women giving birth at the ward in 1 year, and only gynaecologist on-call duty during a 24-h shift, midwives ask the SHOs to handle in case of an emergency obstetric situation. Before the SHOs start with their three-year training to become a gynaecologist, they have worked already in the hospital as an intern during their training to become a doctor. During the period as an intern, they are already responsible for triage of patients and learn how to perform a caesarean section. After their internship, students got selected to become a resident in obstetrics and gynaecology. During a shift, the SHOs are working together in teams of first-, second-, and third-year students. They need to call and help each other during emergency obstetric situations. The training program consists of a full-day simulation-based training and half-day repetition training sessions. To be included, SHOs must work at the medium to high-risk maternity ward of the Mulago hospital. Written informed consent to participants in this study have to be obtained at the beginning of the first training day. Full body simulators (Noelle® and Pedi® Blue neonate, Gaumard) will be used for all training sessions. Communication with Noelle® will be done by an actor, usually a non-local doctor who is not involved in the training day. Other training materials (e.g. disposables, balloon ventilators) will be obtained from Mulago Hospital labour ward and reused as much as possible. Course content (e.g. syllabus, instructor manual, slides, observation forms) will be developed in cooperation with staff members of the obstetrics and gynaecology department in Mulago Hospital and Medsim, a medical simulation centre in Eindhoven, The Netherlands. All materials will be written in English. The instructional design features described by Issenberg et al. and Cook et al. will be used to design the training program [38, 39]. All training sessions will take place at the skills lab of the Makerere University College of Health Sciences, situated at Mulago hospital. The first step will be the train-the-trainers course for the local facilitators provided by a Dutch team, consisting of two obstetricians, a communication expert and a simulation technician. They are all certified simulation educators. The train-the-trainers course will take 4 days. During the first day, principles of simulation-based education (e.g. learning theories, crew resource management, debriefing techniques) and the training program for the SHOs will be discussed and registered. Additionally, local protocols for obstetric emergencies will be restructured during the first day and the local trainers will be introduced to the full body simulators. From the second day on, the local facilitators will start practicing during obstetric simulated scenarios on each other. This includes preparing, leading and debriefing a simulated scenario, including technical set-up (simulators, audio-video equipment, presentation equipment). On the last day the trainers will educate junior house officers (intern doctors) in simulation-based obstetric training as a final test. The Dutch team of master trainers will provide the local trainers feedback. Afterwards, the local trainers will facilitate simulation-based training sessions for all SHOs without supervision of the master trainers in the above mentioned training scheme. Each training session will be organised by two local trainers (main and repetition training). Finally, the master trainers will provide an annual one-day train-the-trainers repetition course. The main training of the SHOs will comprise a one-day (8-h), simulation-based, obstetric training, focusing on 50% medical technical skills and 50% crew resource management skills (i.e. teamwork skills) (Fig. 3). Set up of the main training There are seven groups with SHOs, each group consists of six to nine SHOs. The training scenarios will be postpartum haemorrhage, eclampsia, fetal distress with a vacuum-assisted vaginal delivery and resuscitation of the newborn, breech delivery, and a repetition scenario of postpartum haemorrhage. These scenarios relate to local leading causes of perinatal and maternal mortality and obstetric healthcare problems. Crew resource management skills such as speak up, leadership, situational awareness, and decision-making will be integrated in every scenario with increasing difficulty levels. Learning goals and learning objectives will be defined for the scenarios in collaboration with the local trainers. The main training will start with a general introduction and a knowledge test. Afterwards, the concept of simulation-based training will be explained and the SHOs will set their individual learning goals. Subsequently, the five scenarios will be covered. Every scenario will start with a short introduction of the scenario, followed with performing the scenarios by two to three trainees. The trainers and remaining SHOs will observe the trainees via synchronised video broadcast. Every trainee will participate in at least two scenarios. After each scenario a debriefing by means of the video recordings will be provided by the trainers. The video will be used to show what was done/not done. After this, the students will reflect upon their learning experiences. The debriefing will contains three different phases; reaction, analysis and take home phase. The instructors will provide feedback on both medical technical skills and crew resource management skills. When all five scenarios are completed, the predefined learning goals will be evaluated. Finally, all SHOs will undergo the same knowledge test and they will be asked to fill in an evaluation questionnaire on course perception. After the main training, SHOs will be invited to take part in repetition training sessions. Each repetition training session will comprise half a day. During a repetition training, one clinical scenario will be executed (and repeated). New scenarios will be designed for these sessions based on the same emergency obstetric situations in the main training. However, expectations for the level of performance will be raised and some extra elements such as hand hygiene will be added to keep it challenging. Each repetition training will start with an introduction, in which learning goals will be defined. At the end of every training, learning goals will be evaluated and summarized. Simulation-based obstetric training in a low-income country will decrease the composite of maternal and perinatal mortality ratios. Questions to be answered: The primary outcome of this study will be the combined mortality proportion (CMP). This will concern a composite mortality rate, including maternal and perinatal mortality. Maternal mortality ratio (MMR) is defined as the number of maternal deaths per 100,000 live births [40]. Perinatal mortality ratio (PMR) is defined as the number of stillbirths and deaths in the first week of life per 1000 live births. In Mulago hospital only deliveries with a gestational age of 28 weeks or more or a birth weight of more than one kilogram are registered. These deliveries will be included in the analysis. Expressed as proportions, the combined mortality proportion (CMP) holds: CMP = MMR/100,000 + PMR/1000. Maternal mortality and perinatal mortality will be prospectively registered using the patient registration books in Mulago Hospital. Data extraction from these registration books will be without identification of the subjects. To evaluate the perception of the trainees of the training program, we will use a 42-item questionnaire about ten instructional design features of the training program including feedback, repetition, curriculum integration, difficulty range, learning strategies, clinical variation, controlled environment, individualization, defined outcomes, and simulator validity (the ID-SIM) [41]. The data will be treated as ordinal data at the item level. Suggestions for improvement can be made in an open remark. A 30-item multiple choice knowledge test about technical and non-technical skills was development and evaluated by gynaecologists from Mulago Hospital and the Netherlands. SHOs will be asked to fill in the questionnaire at the onset and end of the full-day main training. Mean values of the knowledge tests will be compared. Clinical performance in the simulated postpartum haemorrhage scenarios will be assessed by three independent Dutch simulation instructors through reviewing the videotaped training sessions. The assessors will be blinded for the day of training and whether the scenario was the first or the last of the day. A skills checklist based on literature and clinical experience will be used [42]. The mean score of the clinical performance of the first and last scenario of the SHOs’ main training will be compared. Team performance will also be assessed by three independent Dutch simulation instructors through reviewing the videotaped training sessions. The Clinical Teamwork Scale (CTS) will be used [43]. The CTS contains questions about communication, situational awareness, decision-making, role responsibility, and patient friendliness. The mean score of the clinical performance of the first and last scenario of the main training will be compared. The assessors will be blinded for the day of training and whether the scenario was the first or the last of the day. The percentage of deliveries by vacuum extraction and caesarean sections will be prospectively collected from Mulago Hospital’s patient registration book. The Weighted Adverse Outcome Score (WAOS) is defined as the total weighted score of each adverse outcome divided by the total number of deliveries (Fig. 4) [44]. Because of registration difficulties, only four out of 10 index measures (maternal death, intrapartum or perinatal death, uterine rupture, Apgar score less than 7 after 5 min) will be assessed. Weighted Adverse Outcome Score (WAOS). ICU, intensive care unit; NICU, neonatal intensive care unit [21]. Maternal and perinatal mortality ratios will also separately be evaluated, because these ratios are not independent in the combined mortality proportion. Finally, the ratio of maternal and perinatal mortality per total number of deliveries will be analysed. Power calculation was carried out as described in both Hussey et al. and Woertman et al. [35–37] First the sample size calculation for a standard randomized clinical trial (RCT) was calculated. To show a reduction in CMP of 20% with an alpha of 0.05 and a power of 80%, a total of 6398 deliveries will be needed for a simple RCT design. The design effect was calculated assuming an intracluster correlation (ICC) of 0.05, a cluster size of 3343 deliveries per year, and seven clusters. Considering the design effect, we will need 2367 deliveries per measurement period. To achieve this number at least 5 weeks for each period will be needed. However, to obtain logistical possibilities, the duration of each step will be 7 weeks with a total study duration of 56 weeks. Statistical significance will be accepted at a two-sided p-value < 0.05. All data will be collected, secured and stored in Mulago hospital. Access to the final trial dataset will be limited to persons who have to perform statistical analyses and to interpret results. The statistical analysis of the stepped wedge design will be analysed in different steps. First of all, differences in patient characteristics (age, parity, gestational age, single/multiple pregnancy, neonatal gender, birth weight) across clusters will be investigated with Kruskal-Wallis for numerical data and with exact chi-square statistics for categorical data. The characteristics that seem to be different across clusters (p < 0.05) will enter into the generalized linear mixed effects model (GLMM) for estimation of treatment effect. Here the outcome is the event (i.e. composite mortality rate, including maternal and perinatal mortality) on the individual and we will use a logit link function to model the probability of the event. In the logit scale, the cluster indicator will serve as random effect on the intercept, the selected patient characteristics will enter this model as linear predictors, the period of the stepped wedge is treated as categorical effect, as well as the treatment effect. In mathematical terms the model can be formulated as follows: with the logit function given by logit(x) = log x − log(1 − x) and Yij: the binary outcome on patient j in cluster i, Zi: the random effect of cluster i, assumed to be normally distributed Zi~N0σC2, and with σC2 the between-cluster variance, Pij: a vector of indicators Pij = (Pij1, Pij2, …, PijR)T for patient j in cluster i that indicate in which period the observation is taken, with ∑r=1RPijr=1 and R the number of periods, Xij: a vector of confounders Xij = (Xij1, Xij2, …, XijS)T for patient j in cluster i, and S the number of covariates. Tij: the treatment indicator for patient j in cluster i. μ: the overall intercept, αr: the effect of period r βs: the effect of confounder Xijs γ: the effect of treatment [45]. Ethical approval was obtained from both the Mulago Research and Ethics Committee (Protocol MREC: 674), and the Uganda National Council for Science and Technology (UNCST, SS 3927). Written informed consent to participants in this study was obtained at the beginning of the first training day.