We performed this educational initiative in the 12-bed Pediatric Intensive Care Unit (PICU) and the 15-bed Neonatal Intensive Care Unit (NICU) of the Level-IV University Children’s Hospital, University Medical Center Hamburg-Eppendorf, Germany, between August 2019 and July 2021 after approval of the local ethical review committee (Ethikkommission der Ärztekammer Hamburg, Germany).
A multi-professional team of physicians and nurses supervised the project. During the planning phase, we collected ideas and suggestions on potential improvements and current problems related to pediatric MV from the medical and nursing teams of both ICUs. In addition, we regularly exchanged ideas via a “Kanban Board” through surveys and regular team feedback rounds (Fig. 1). Checklists for set-up and settings according to treatment goals were fixed permanently and visibly to each ventilator. Additionally, pocket cards were distributed to all staff summarizing essential educational topics and mnemonic aids. A team consisting of intensive care nurses, physicians, medical education specialists, and communication experts developed the instructive content for a 30-minute educational film. The film covered the above-mentioned treatment goals, ventilation, oxygenation, inhaled nitric oxide (iNO), and complications of MV. We trained staff members by showing the film, answering questions, and discussing critical topics during training sessions in both ICUs. In addition, each staff member had access to the film from home and any hospital computer for self-study (Fig. 1).
In joint discussions and regular feedback rounds, a highly standardized approach was developed and adapted to set up ventilators and to establish specific start settings according to patient weight categories. After several team feedback rounds during staff meetings, we expanded the contents of the checklists, pocket cards, and educational film (Fig. 1).
We developed a theory test (TT) following the specific topics of a validated testing tool for pediatric mechanical ventilation(O'Boyle et al., 2014) to evaluate physicians’ and nurses’ theoretical knowledge at baseline. A questionnaire recorded the participant’s profession and experience. We randomly assigned the participants to one of two TTs of equal difficulty using a randomization list. The two TTs consisted of 15 multiple-choice questions on the following topics: ventilation, oxygenation, iNO, high-frequency oscillatory ventilation (HFOV), and complications of MV.
A practical skill test (PST), consisting of four sequential Objective Structured Clinical Examinations (OSCE)(Brannick et al., 2011; Burgess et al., 2020; Nicholls et al., 2016; Nyangeni et al., 2022) was taken directly at a modified ICU workplace. We randomly assigned the participants to one of the two equally difficult PSTs using a randomization list. Assessors observed the participants during the PST and assigned a maximum of 30 points based on performance-structured checklists. The PST simulated four clinical challenges that may occur during mechanical ventilation: A) set up and connect a ventilator including a humidification system (while time was measured), B) start a ventilator and set ventilation parameters and alarm limits for a postoperative patient (newborn, 3 kg body weight with healthy lungs), C) modify MV according to a blood gas analysis showing respiratory acidosis or alkalosis, D) adjust the ventilator settings following a sudden improvement or deterioration of lung compliance indicated by corresponding device alarms. After the intervention, the staff was re-tested using the other previously unassigned TT and PST.
We agreed on MV treatment goals consisting of a lung-protective ventilation strategy for all patients using synchronized mandatory intermittent ventilation (SIMV) and HFOV as a rescue MV mode. We specified recommendations regarding the initiation of MV, ventilator setup, and settings for different weight categories from 0.5 to 80 kg. The specifications referred to PIP, PEEP, inspiration time, respiratory rate, pressure support, trigger sensitivity, and TV and minute volume for SIMV; and mean airway pressure, amplitude, frequency for HFOV. In addition, we defined oxygenation targets to avoid hyperoxia, MV monitoring frequency, alarm limit ranges for all respirator settings, and the use of humidified gas for patients ventilated longer than 24 h.(M. C. J. Kneyber et al., 2017; Pediatric Acute Lung Injury Consensus Conference, 2015; A. H. van Kaam et al., 2021) We emphasized a standardized display setting showing pressure, flow, volume curves, and alarm limits.
Before and after the intervention, we performed random and unannounced checks of the ventilator and monitor settings in all invasively and non-invasively ventilated patients in both ICUs. The checks were performed approximately twice per week, but not on consecutive or fixed days, and at varying day times to ensure unpredictability (Fig. 1). In addition, ventilator displays and alarm limits were checked. Treatment goal violations were categorized as follows: a) ventilator setup: humidification not turned on, or no water in the humidification system; b) ventilator display: absent or incomplete displays of pressure, volume, flow curves or alarm limits; c) volume target monitoring: missing or non-optimal (inappropriately high or low) limits for minute volume or TV; d) pressure target monitoring: delta pressure > 15 mmHg, PIP > 30 mmHg; missing, or inappropriately high or low limits for PEEP or CPAP pressure; e) saturation limits (SpO2) monitoring: inappropriately high or low limits.
Statistical Analysis
A required minimum sample size of N = 44 participants was calculated to detect a 10% improvement in participants’ overall performance, assuming an overall performance mean of 75% and a standard deviation of 15% with a specified power of 90% and a significance level of 0.05. Continuous variables were expressed as mean (95% confidence interval [CI]). A random allocation sequence was generated using the sample function in R with a 50% probability. Discrete data were compared between groups with the Chi-square test, and effects were reported as Cramer’s V effect sizes. A paired two-tailed t-test was used to compare continuous variables before and after the intervention, and effects were reported as Cohen’s D effect sizes. For the performance comparison, only those participants who had taken part at both test times were included. Linear regression models were calculated for predictor variables (study phase, profession, professional experience) to analyze their impact on the participants’ theoretical knowledge and practical skills performance. Treatment goal adherence over the study period was visualized using a statistical process control chart and adherence rate analyzed using a mixed-effects model to account for multiple checks per patient. P values less than 0.05 were considered significant. Statistical analyses were performed using R 4.1.2 (2021-11-01) (R Core Team, Vienna, Austria).