Effect of an app-guided advanced cardiac life support algorithm on teamwork, non-technical skills, and workload in simulated in-hospital cardiac arrest: a randomized controlled feasibility trial Authors: Anne Kamphausen, Maximilian Pilz, Matthias Baumgärtel, Jens Christian Kubitz, Christian Engelen Published: 2026 ABSTRACT AbstractBackgroundIn-hospital cardiac arrest carries high mortality despite advances in monitoring, rapid response systems, and adherence to advanced life support (ALS) guidelines. Digital decision-support tools, including mobile applications, are increasingly explored to improve guideline adherence, yet their impact on cognitive workload, team performance, and ALS performance remains uncertain.MethodsWe conducted a prospective, randomized, simulation-based controlled feasibility trial with 40 interprofessional teams (80 participants) comparing ALS performance with and without the use of the ‘CPR Leader’ app. Two-person teams were randomized 1:1 to intervention or control and followed standardized in-hospital cardiac arrest scenarios (asystole and ventricular fibrillation). Primary outcomes included team performance (TEAM), non-technical skills (ANTS), and subjective workload (NASA-TLX). Secondary outcomes comprised objective resuscitation metrics, including time to rhythm analysis, defibrillation, medication administration, and chest compression quality.ResultsOverall NASA-TLX scores were high but comparable between intervention and control group (mean 49.2 ± 13.6 vs. 54.1 ± 14.6; p = 0.2), with higher effort reported in the app group and higher frustration in the control group. TEAM and ANTS scores showed no evidence of meaningful between-group differences. Across the two scenarios types, ALS performance metrics—including rhythm analysis, defibrillation, medication administration, and chest compression quality—were similar between groups, with the exception of earlier chest compression initiation in the control group during ventricular fibrillation.ConclusionsIn this exploratory randomized simulation study involving predominantly experienced in-hospital healthcare professionals, the use of a resuscitation support application did not demonstrate a clear improvement in adherence to guideline-recommended ALS algorithms or resuscitation performance compared with standard practice. These findings likely reflect the high level of participants’ clinical experience and should not be generalized to novice providers, mixed-experience teams, prehospital settings, or training environments. Further studies are warranted to evaluate the potential benefits of cognitive aids in less experienced populations and in real-world clinical practice.Trial numberNone.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13049-026-01673-7.Keywords: In-hospital cardiac arrest, Advanced life support, Resuscitation app, Teamwork, Cognitive workload Background In-hospital cardiac arrest remains a frequent and clinically devastating event, with persistently high mortality despite advances in monitoring, rapid response systems, and the critical importance of adherence to current advanced cardiac life support guidelines [1, 2]. Digital decision-support tools—particularly mobile applications—are increasingly explored as a means to enhance resuscitation guideline adherence during cardiac arrest. Existing studies indicate improvement in adherence to guidelines and faster execution of critical steps [3–5]. Resuscitation imposes high cognitive demands due to time pressure, complex algorithmic decision-making, and simultaneous multitasking requirements. High workloads have been noted to correlate with potentially unsafe levels of cognitive load and increased risk of human error [6]. Prior research indicates that the effect of tablet-based decision-support tools on workload remains unclear [7]. In contrast, Grundgeiger et al. demonstrated in a randomized controlled trial that team leaders supported by a digital aid reported significantly lower mental and physical demand and required less effort to achieve performance compared with unsupported teams [8]. These mixed findings suggest that digital guidance may either alleviate or increase cognitive load depending on context, user experience, and interface design. Evidence from lay-bystander CPR app research suggests that introducing decision-support tools during active resuscitation can initially impose additional cognitive burden, as rescuers must divide attention between patient assessment, task execution, and app interpretation, potentially delaying critical early steps such as checking for breathing or initiating chest compressions [9]. While this evidence originates from an untrained bystander population, it raises the question of whether comparable attention-related overhead could also affect experienced in-hospital teams, albeit potentially to a lesser degree given their higher baseline expertise. The inconsistency of prior evidence and the known cognitive challenges of resuscitation justify investigating how real-time app use affects workload, team interaction, and ALS performance in an in-hospital setting. Therefore, this study aims to determine whether real-time use of a resuscitation app affects mental workload, communication patterns, and teamwork dynamics, while simultaneously improving adherence to advanced life support guidelines during simulated in-hospital cardiac arrest. Method Study design This prospective, randomized, simulation-based controlled study compared advanced cardiac life support performance with and without the use of a dedicated application. It adhered to the principles outlined in the Declaration of Helsinki, and its conduct was reviewed and approved by the institutional review board of PMU – Paracelsus Medical University, Nuremberg, Germany (IRB-2022-011) and hospital’s staff council. This study was not prospectively registered. Reporting of this trial was guided by the CONSORT extension for non-pharmacological interventions and the Reporting Guidelines for Health Care Simulation Research, with particular attention to intervention description, participant expertise, and scenario fidelity. Participants and setting Participants were recruited within the department of Anesthesia, Emergency Medicine, and Critical Care Medicine, General Hospital Nuremberg between October 2022 and February 2023. Participation was voluntary, and informed consent was obtained from all participants prior to the study. All participants were organized into two-person teams, each consisting of one physician and one nurse or anaesthesia technical assistant. Each team completed only one of the two simulation scenarios (ventricular fibrillation or asystole due to hypoxemia following aspiration). Scenario assignment was determined randomly and independently of study-arm allocation (intervention vs. control), resulting in 10 intervention and 10 control teams per scenario prior to the start of the trial. As teams were exposed to only a single scenario, no scenario order, washout period, or carry-over/learning effects between scenarios were applicable to this design. Details of the study design are given in Fig. 1. Study flow diagram Simulation scenarios followed the 2021 German Resuscitation Council (GRC) ALS algorithm [10] and were performed on a Resusci Anne® manikin (Laerdal Co., Stavanger, Norway). Simulations were conducted on an inpatient ward and recorded using three fixed cameras capturing the patient manikin and the entire team. Standard resuscitation equipment, including defibrillators, airway management tools, medications, and monitoring devices, was available. Each simulation involved the physician, a nurse, and a study team member to assist with resuscitation tasks (e.g., drug preparation, chest compressions). The study team member did not participate in decision-making. An attending anesthetist with over 15 years of simulation instructor experience conducted the sessions. Scenarios lasted approximately 10–12 min, allowing multiple rhythm analysis, medication administration, and team coordination tasks. Adequate treatment with defibrillation and medication in the simulation scenario „ventricular fibrillation“ is necessary to improve the patient’s condition. After return of spontaneous circulation, the ECG shows ST-segment elevation in the posterior leads. In the simulation scenario “asystole” due to hypoxemia following aspiration, the patient’s condition improves if an airway (e.g. endotracheal tube or Laryngeal mask) is correctly placed, CPR is performed correctly, and medication is administered appropriately. This scenario logic was built into the simulation to ensure a physiologically plausible and internally consistent course of the mannequin’s simulated vital signs in response to team actions; however, overall scenario success (i.e., simulated patient improvement) was not captured, quantified, or analyzed as a study outcome. The present study exclusively evaluated process-level performance metrics (see Outcome Measures). Following each scenario, participants received a brief, informal debriefing led by the simulation instructor, focused on immediate clarification of scenario events; a structured, formal debriefing protocol was not applied. Intervention teams used an Apple tablet (Apple Inc., Cupertino, CA, USA) with the ‘CPR Team Leader’ application (Version 1.0.9; Alphabet Medical GmbH; Bamberg, Germany), while control teams performed scenarios under standard conditions, with no cognitive aid, algorithm poster, or other decision-support tool available; control teams relied solely on their existing training and memory of the ALS algorithm. CPR Team Leader was a software application registered as a Class I medical device (MDD), designed to support resuscitation teams in adhering to current advanced life support (ALS) algorithms (Supplementary Fig. 1 ). The application guides users through the resuscitation algorithm depending on the initial cardiac rhythm, following the GRC resuscitation guidelines of 2021 [10]. Beyond displaying the relevant algorithm branch, the app provides integrated timers for rhythm-check and medication intervals, audible and visual reminders, and input fields for documenting administered medications and defibrillations, with automatic progression to the next algorithm step based on user input. A figure illustrating the user interface is provided in Supplementary Fig. 1. During the intervention scenarios, the tablet was operated by the team leader, who opened the application at the start of the scenario. The extent and pattern of app use during the scenario varied between participants and was not standardized. Interactions with the app were captured on video but were not formally rated or quantified as part of the performance assessment. The CPR Leader application was introduced at the hospital in summer 2022 to support adherence to resuscitation guidelines and has since been implemented in routine clinical practice. Medical colleagues with extensive professional experience in emergency medicine reviewed the app in advance to ensure completeness and correctness of the algorithm content. Prior to study initiation, all departments of the hospital received structured instruction and online training on the use of the application. Participants had therefore already been familiarized with the app in the context of routine departmental implementation prior to the simulation scenarios, and did not receive additional dedicated training immediately before the study scenarios. Outcome measures Two trained raters (CE, AK) reviewed video recordings independently to assess adherence to the resuscitation protocol, team performance, and non-technical skills. Primary outcome Team performance and non-technical skills were evaluated using the Team Emergency Assessment Measure (TEAM) and the Anaesthetists’ Non-Technical Skills (ANTS) framework. All simulations were independently reviewed by two trained raters (CE, AK), who assessed observable behaviours according to the structured rating scales. The mean score of both raters was used for analysis. Inter-rater reliability was assessed using the intraclass correlation coefficient (ICC). Team performance Team performance was assessed using the German version of Team Emergency Assessment Measure (TEAM). The TEAM instrument evaluates teamwork performance, focusing on leadership, task management, and inter-team communication in high-pressure environments during emergencies [11, 12]. Non-technical skills Non-technical skills were assessed using the Anaesthetists’ Non-Technical Skills (ANTS) system [13, 14]. ANTS is a validated behavioural marker system comprising four categories: task management, team working, situation awareness, and decision-making. At the end of each simulation, participants completed the NASA Task Load Index (NASA-TLX) to assess perceived workload. Subjective workload Participant´s perceived workload during the simulation was measured using the NASA Task Load Index (NASA-TLX). The NASA Task Load Index (NASA-TLX) measures perceived workload across six domains: mental, physical, and temporal demand, performance, effort, and frustration [15]. High NASA-TLX scores are observed in physically demanding tasks (overall score 62) or in activities requiring active environmental scanning, such as visual search tasks (overall score 58) [16]. In comparison, routine clinical tasks for physicians typically yield an overall workload score of approximately 40 in real-world practice [17]. The global NASA-TLX RAW score was calculated as the unweighted mean of the six subscales (Mental Demand, Physical Demand, Temporal Demand, Performance, Effort, Frustration), each scored from 0–100, yielding a theoretical range of 0–100 [18, 19]. The Performance subscale was scored in its standard NASA-TLX direction (anchored from ‘Perfect’ to ‘Failure’), such that higher values indicate poorer self-rated performance, consistent with the direction of all other subscales, where higher values indicate greater perceived workload. Secondary outcome Objective resuscitation performance was evaluated using data recorded during simulation based on the German resuscitation guidelines published in 2021 [10]. Key performance metrics included the time to first chest compression, time to first rhythm analysis, time to first defibrillation, time to first administration of adrenaline, adherence to recommended rhythm and medication intervals, no-flow time, and timing of chest compression provider changes. Data analysis This study was designed as a feasibility study to inform the planning of future, adequately powered trials. Accordingly, the sample size was determined by the number of participants available during the study period rather than by a formal sample size calculation. Based on the study by Siebert et al., the group size was set at 10 teams per study arm. As no formal sample size calculation, and thus no power calculation for a given effect size, was performed, all results are of exploratory nature. Consequently, no correction for multiple testing has been performed, as all p-values are of descriptive nature and interpreted exploratory only. Continuous parameters are summarized by reporting mean, standard deviation, median, interquartile-range, minimum, and maximum. Categorical values are reported as absolute and relative frequencies. All measures are given stratified by study arm and for the full population. Mean differences between the study arms are reported together with their 95% confidence intervals based on a t-distribution. Descriptive p-values were computed from Welch t-tests for continuous data and from Pearson’s chi-square tests for categorical data. For the NASA-TLX analysis, differences and p-values have been adjusted for the random effect “team”. Inter-rater reliability was assessed using the intraclass correlation coefficient (ICC) together with its 95% confidence interval (CI). All statistical analyses were conducted using the statistical software R, version 4.6.1. Results Participant characteristics In total, 40 interprofessional teams (80 participants) were included in the analysis, with an equal allocation to the intervention and control group. Baseline demographic variables—including sex distribution, age, clinical experience, specialty background, and advanced training—did not differ between groups. The majority of participating physicians were specialists in anesthesiology with extensive clinical experience (> 7 years in 70% of cases). No imbalance between groups was observed in any demographic variable (Tables 1 and 2). Table 1Demographic characteristics of physiciansIntervention (n = 20)Control(n = 20)Total (n = 40)Gender n (%) Male13 (65%)10 (50%)23 (58%) Female7 (35%)10 (50%)17 (43%)Age n (%) > 40 years12 (60%)10 (50%)22 (55%) 36–40 years3 (15%)6 (30%)9 (23%) <= 35 years5 (25%)4 (20%)9 (23%)Work Experience n (%) > 7 years14 (70%)16 (80%)30 (75%) 4–6 years1 (5%)0 (0%)1 (3%) 0–3 years5 (25%)4 (20%)9 (23%)Specialist physician n (%)15 (75%)15 (75%)30 (75%) Board qualification in prehospital emergency medicine (EMS) n (%)16 (80%)13 (65%)29 (73%) Board qualification in critical care n (%)9 (45%)7 (35%)16 (40%)Specialty n (%) Anesthesia18 (90%)19 (95%)37 (93%) Internal Medicine2 (10%)1 (5%)3 (8%) Demographic characteristics of physicians Board qualification in prehospital emergency medicine (EMS) n (%) Board qualification in critical care n (%) Table 2Demographic characteristics of nursesIntervention (n = 20)Control(n = 20)Total (n = 40)Gender n (%) Male10 (50%)9 (45%)19 (48%) Female10 (50%)11 (55%)21 (53%)Age n (%) > 40 years8 (40%)12 (60%)20 (50%) 36–40 years4 (20%)3 (15%)7 (18%) <= 35 years8 (40%)5 (25%)13 (33%)Work Experience n (%) > 7 years17 (85%)17 (85%)34 (85%) 4–6 years2 (10%)1 (5%)3 (8%) 0–3 years1 (5%)2 (10%)3 (8%)Advanced Training in Anaesthesia and Intensive Care Nursing n (%)15 (75%)15 (75%)30 (75%)Anaesthesia Technician n (%)3 (15%)4 (20%)7 (17.5%)Advanced Training in Emergency Nursing n (%)1 (5%)0 (0%)1 (2.5%)No specialty certification n (%)1 (5%)1 (5%)2 (5%) Demographic characteristics of nurses Primary outcome Assessment of subjective workload (NASA-TLX) The overall NASA-TLX RAW score did not differ meaningfully between groups (mean 49.2 ± 13.6 in intervention vs. 54.1 ± 14.6 in control group; median 50 in intervention vs. 54 in control group; mean difference − 4.9 with 95% CI [-12, 2.6]; p = 0.2 from mixed linear regression with random intercept team), with broad but overlapping ranges (24–85 vs. 16–78). Subscales Two domains showed exploratory signals with directionally distinct effects. Effort was higher with the app (62.4 ± 21.1 vs. 51.0 ± 24.3; median 70 vs. 50; mean difference 11 with 95% CI [1.9, 21]; p = 0.02), suggesting added exertion to manage the digital guidance alongside resuscitation tasks. Effort was higher with the app (62.4 ± 21.1 vs. 51.0 ± 24.3; median 70 vs. 50; mean difference 11 with 95% CI [1.9, 21]; p = 0.02), suggesting added exertion to manage the digital guidance alongside resuscitation tasks. Frustration was higher without the app (32.6 ± 25.9 vs. 46.8 ± 24.0; median 50 vs. 23; mean difference − 14.0 with 95% CI [-25, 3.0]; p = 0.013), potentially reflecting uncertainty or lack of structured prompts when relying on memory/experience alone. Frustration was higher without the app (32.6 ± 25.9 vs. 46.8 ± 24.0; median 50 vs. 23; mean difference − 14.0 with 95% CI [-25, 3.0]; p = 0.013), potentially reflecting uncertainty or lack of structured prompts when relying on memory/experience alone. Other domains - Mental Demand (54.6 ± 25.9 vs. 61.0 ± 23.2; 95% CI for mean difference [-18, 5.5]; p = 0.3), Physical Demand (47.3 ± 28.0 vs. 54.9 ± 25.7; 95% CI for mean difference [-20.0, 4.6]; p = 0.2), Temporal Demand (42.4 ± 22.8 vs. 49.9 ± 24.3; 95% CI for mean difference [-19, 3.9]; p = 0.2), and self-rated Performance (55.9 ± 22.6 vs. 60.9 ± 19.0; 95% CI for mean difference [-16, 6.0]; p = 0.4) showed only small differences and comparable medians. Details are given in Fig. 2. Additional results are displayed in the Supplementary Table S1 in the Appendix. Results of NASA Task Load Index Assessment of team performance (TEAM) Team performance did not differ between groups across the TEAM subscales Leadership, Teamwork, and Task Management, nor on the Global Rating (Table 3). Mean (SD) values were closely aligned across all TEAM domains (e.g., Leadership 6.8 (1.0) vs. 6.7 (1.7); Teamwork 22.4 (4.4) vs. 21.7 (5.6); Task Management 6.0 (1.6) vs. 5.5 (2.0); Global Rating 7.4 (1.9) vs. 7.2 (2.3); all p ≥ 0.4)). These distributions indicate comparable central tendencies and spread, with no signal of superiority for either condition. Inter-rater reliability was excellent (ICC = 0.959; 95% CI 0.95–0.97), supporting robustness of the ratings. Table 3Results of team emergency assessment measure (TEAM)InterventionN = 20Control N = 20Total N = 40Mean Diff.[95% CI]p-value Leadership 0.17 [-0.76, 1.1]0.7 Mean (SD)6.8 (1.0)6.7 (1.7)6.8 (1.4) Median (Q1 - Q3)7 (6–8)8 (6–8)7 (6–8) Min – Max5–82–82–8 Unknown101 Teamwork 0.67 [-2.6, 3.9]> 0.9 Mean (SD)22.4 (4.4)21.7 (5.6)22.0 (5.0) Median (Q1 - Q3)24 (19–26)23 (18–26)24 (19–26) Min – Max11–2812–2811–28 Unknown101 Task Management 0.45[-0.74, 1.6]0.4 Mean (SD)6.0 (1.6)5.5 (2.0)5.7 (1.8) Median (Q1 - Q3)6 (5–7)6 (4–7)6 (5–7) Min – Max2–82–82–8 Global Rating Scale 0.23 [-1.1, 1.6]0.7 Mean (SD)7.4 (1.9)7.2 (2.3)7.3 (2.1) Median (Q1 - Q3)8 (6–9)8 (6–9)8 (6–9) Min – Max3–103–103–10 Results of team emergency assessment measure (TEAM) 0.17 [-0.76, 1.1] 0.67 [-2.6, 3.9] 0.45 [-0.74, 1.6] 0.23 [-1.1, 1.6] Assessment of non-technical skills (ANTS) Across all four ANTS categories—Task Management, Team Working, Situation Awareness, Decision Making—there were no clear between group differences (Table 4). Mean category scores ranged from 2.6 to 3.3, reflecting moderate performance across domains without directional trends favoring the intervention. Inter-rater reliability for ANTS varied from moderate to excellent (ICC 0.642–0.945), indicating acceptable consistency across raters and domains. Table 4Results of ANTSIntervention N = 20Control N = 20Total N = 40Mean Diff.[95% CI]p-value Task Management 0.58[-1.5, 2.6]0.6 Mean (SD)11.9 (2.6)11.3 (3.7)11.6 (3.2) Median (Q1 - Q3)12 (10–14)12 (8–15)12 (9–14) Min – Max6–165–165–16 Team Working -0.15[-2.2, 1.9]0.9 Mean (SD)11.6 (2.9)11.8 (3.4)11.7 (3.1) Median (Q1 - Q3)12 (10–14)12 (9–15)12 (9–15) Min – Max6–166–166–16 Situation Awareness 0.25[-1.6, 2.1]0.8 Mean (SD)11.8 (2.5)11.6 (3.3)11.7 (2.9) Median (Q1 - Q3)12 (11–13)13 (9–14)12 (10–14) Min – Max6–166–166–16 Decision Making -0.20[-1.8, 1.4]0.8 Mean (SD)8.4 (1.8)8.6 (2.9)8.5 (2.4) Median (Q1 - Q3)9 (7–10)9 (6–11)9 (7–10) Min – Max5–113–123–12 Results of ANTS 0.58 [-1.5, 2.6] -0.15 [-2.2, 1.9] 0.25 [-1.6, 2.1] -0.20 [-1.8, 1.4] Secondary outcomes Simulation - scenario “asystole” Across all key ALS performance metrics, no clear differences were detected between the intervention and control group. Time to first rhythm analysis was similar between groups (119.6 ± 63.7 s vs. 115.1 ± 28.6 s), as was the time to first administration of adrenaline (185.2 ± 65.7 s vs. 199.2 ± 28.2 s). Deviations from recommended 120-second rhythm analysis intervals and 3–5-minute adrenaline intervals showed large inter-team variability, but no systematic intergroup differences. Chest compression performance—including start time, total compression duration, no-flow time, and proportion of no-flow time—did not differ remarkably. Timing of ventilation onset and securing of the airway likewise showed no group differences, and choice of airway device (ITN vs. LMA) was similarly distributed between groups. All results are displayed in Table 5. Table 5Time course of advanced cardiac life support „asystole“Intervention N = 10Control N = 10TotalN = 20Mean Diff.[95% CI]p-value Start of Chest Compression [sec] 0.00[-5.8, 5.8]> 0.9 Mean (SD)17.7 (7.4)17.7 (4.3)17.7 (5.9) Median (Q1 - Q3)18 (11–21)19 (15–21)18 (14–21) Min - Max9–3410–239–34 Chest Compression time in total [sec] -14 [-68, 41]0.6 Mean (SD)728.7 (68.3)742.4 (44.4)735.6 (56.5) Median (Q1 - Q3)744 (733–776)748 (714–775)748 (724–776) Min - Max563–789670–808563–808 No Flow Time [sec] 1.4 [-27, 30]> 0.9 Mean (SD)82.4 (33.4)81.0 (26.8)81.7 (29.5) Median (Q1 - Q3)84 (52–109)73 (59–99)77 (58–101) Min - Max28–13755–13928–139 % No Flow Time 0.40 [-3.6, 4.4]0.8 Mean (SD)11.4 (4.5)11.0 (4.0)11.2 (4.2) Median (Q1 - Q3)12 (7–14)10 (8–13)11 (8–14) Min - Max4–178–214–21 Time to first analysis [sec] 4.5[-43, 52]0.8 Mean (SD)119.6 (63.7)115.1 (28.6)117.4 (48.1) Median (Q1 - Q3)107 (70–187)117 (98–136)116 (78–139) Min - Max44–21170–16644–211 First Administration of Epinephrine [sec] Mean (SD)185.2 (65.7)199.2 (28.2)192.2 (49.7)-14 [-63, 35]0.5 Median (Q1 - Q3)171 (153–259)206 (176–216)193 (163–226) Min - Max81–272150–23781–272 Start Ventilation [sec] 47 [-38, 132]0.3 Mean (SD)122.1 (114.9)75.5 (43.3)98.8 (87.8) Median (Q1 - Q3)72 (49–227)61 (39–117)69 (44–129) Min - Max0–32026–1440–320 Secure Airway [sec] 0.70[-123, 125]> 0.9 Mean (SD)251.2 (159.9)250.5 (88.9)250.9 (126.0) Median (Q1 - Q3)279 (72–371)252 (211–299)259 (157–335) Min - Max47–467101–38347–467 Airway Device ITN6 (60%)2 (20%)8 (40%) LMA4 (40%)8 (80%)12 (60%) Time course of advanced cardiac life support „asystole“ 0.00 [-5.8, 5.8] -14 [-68, 41] 1.4 [-27, 30] 0.40 [-3.6, 4.4] 4.5 [-43, 52] -14 [-63, 35] 47 [-38, 132] 0.70 [-123, 125] Simulation scenario “ventricular fibrillation” Time to first rhythm analysis (92.0 ± 43.4 s vs. 94.5 ± 29.7 s) was comparable between intervention and control teams. Time to first defibrillation did not differ clearly between intervention and control teams (112.8 ± 40.9 s vs. 133.3 ± 31.1 s, p = 0.224), although a numerical trend toward earlier defibrillation was observed in the intervention group. No differences were detected in adherence to recommended rhythm analysis or defibrillation intervals. Medication administration patterns for adrenaline and amiodarone showed substantial variability across teams, but did not differ clearly between groups. The only notable performance difference in the ventricular fibrillation scenario concerned earlier initiation of chest compressions in the control group (18.1 ± 4.7 s vs. 12.4 ± 5.3 s; p = 0.021). All other resuscitation parameters—including no-flow time, airway management timing, and airway device selection—showed no group differences. All results are displayed in Table 6. Table 6Time course of advanced cardiac life support „Ventricular Fibrillation“Intervention N = 10Control N = 10TotalN = 20Mean Diff.[95% CI]p-value Start of Chest Compression [sec] 5.7 [0.97, 10]0.021 Mean (SD)18.1 (4.7)12.4 (5.3)15.3 (5.7) Median (Q1 - Q3)20 (15–21)11 (8–17)16 (11–21) Min - Max11–258–228–25 Chest Compression time in total [sec] 1.2 [-52, 55]> 0.9 Mean (SD)765.0 (58.5)763.8 (55.1)764.4 (55.3) Median (Q1 - Q3)770 (697–807)751 (736–809)751 (723–808) Min - Max694–856675–855675–856 No Flow Time [sec] -20 [-49, 10]0.2 Mean (SD)92.2 (13.6)111.8 (40.4)102.0 (31.0) Median (Q1 - Q3)95 (82–99)106 (83–121)95 (83–116) Min - Max67–11757–19257–192 % No Flow Time -2.6[-6.8, 1.6]0.2 Mean (SD)12.1 (2.0)14.7 (5.7)13.4 (4.4) Median (Q1 - Q3)13 (11–14)13 (11–15)13 (11–14) Min - Max8–158–258–25 Time to first analysis [sec] -2.5 [-38, 33]0.9 Mean (SD)92.0 (43.4)94.5 (29.7)93.3 (36.2) Median (Q1 - Q3)74 (60–142)93 (73–112)85 (69–112) Min - Max48–16444–15544–164 Time to first Defibrillation [sec] -21 [-55, 14]0.224 Mean (SD)112.8 (40.9)133.3 (31.1)123.1 (36.9) Median (Q1 - Q3)101 (85–163)137 (117–157)126 (99–158) Min - Max65–17861–17361–178 Administration of Epinephrine 0.2 between 1./2. defibrillation0 (0%)2 (20%)2 (10%) between 2./3. defibrillation2 (20%)1 (10%)3 (15%) between 3./4. defibrillation5 (50%)7 (70%)12 (60%) between 4./5. defibrillation2 (20%)0 (0%)2 (10%) between 5./6. defibrillation1 (10%)0 (0%)1 (5%) Administration of 300 mg Amiodarone 0.4 between 1./2. defibrillation0 (0%)1 (10%)1 (5%) between 2./3. defibrillation0 (0%)1 (10%)1 (5%) between 3./4. defibrillation5 (56%)5 (50%)10 (53%) between 4./5. defibrillation3 (33%)1 (10%)4 (21%) between 5./6. defibrillation1 (11%)0 (0%)1 (5%) Start Ventilation [sec] -27 [-81, 27]0.3 Mean (SD)75.5 (30.6)102.3 (72.0)88.9 (55.6) Median (Q1 - Q3)75 (50–92)73 (63–100)73 (54–96) Min - Max38–12241–25438–254 Secure Airway [sec] 5.7 [-145, 157]> 0.9 Mean (SD)321.2 (181.4)315.5 (134.5)318.4 (155.4) Median (Q1 - Q3)307 (203–356)308 (220–388)307 (207–372) Min - Max103–634134–539103–634 Airway Device > 0.9 ITN7 (70%)8 (80%)15 (75%) LMA3 (30%)2 (20%)5 (25%) Time course of advanced cardiac life support „Ventricular Fibrillation“ 5.7 [0.97, 10] 1.2 [-52, 55] -20 [-49, 10] -2.6 [-6.8, 1.6] -27 [-81, 27] 5.7 [-145, 157] Discussion In this simulation-based randomized controlled study, real-time use of a resuscitation application neither improved nor impaired observer-rated teamwork (TEAM), non-technical skills (ANTS), or overall perceived workload compared with standard practice. Adherence to resuscitation guidelines was largely comparable between groups, with the exception of earlier initiation of chest compressions in the control group during ventricular fibrillation. Consistent with the feasibility nature of this study, these findings should be regarded as exploratory and hypothesis-generating rather than confirmatory. Both groups experienced a high level of cognitive workload, as reflected by NASA-TLX scores (mean 49.2 ± 13.6 vs. 54.1 ± 14.6), with only small differences between conditions. This finding contrasts with prior work by Grundgeiger et al., who reported reduced workload in app-supported teams [8]. First, the complexity and realism of the in-hospital simulation scenarios used in this study may have imposed a uniformly high cognitive demand that could not be substantially mitigated by app support. Second, our study involved experienced clinicians, the majority of whom had more than seven years of clinical experience. Such individuals are likely to rely on well-established mental models and automated decision-making processes during resuscitation, potentially limiting the additional benefit of external cognitive aids. Interestingly, the opposing effect observed here—higher effort while using the app but greater frustration without a cognitive aid, with no differences in overall workload— may reflect a trade-off between structured guidance and operational overhead: the app provides cognitive support and reduces uncertainty, but simultaneously requires attention, interaction, and interpretation during time-critical tasks. Notably, the extent and pattern of app use during the intervention scenarios varied between participants and was not standardized; although interactions with the app were captured on video, they were not formally rated or quantified. It therefore remains unclear to what extent the higher effort scores reflect genuine cognitive overhead from the resuscitation task itself, as opposed to unfamiliarity with the interface, divided attention between the app and the clinical scenario, or interface-related friction. The absence of differences in TEAM and ANTS scores further supports the notion that digital cognitive aids may have limited impact in highly experienced teams. Both instruments are well validated for assessing non-technical skills [20, 21] but may be less sensitive to incremental improvements in expert populations, where ceiling effects are expected [22–24]. Accordingly, the absence of differences is consistent with previous literature reporting limited responsiveness of behavioural marker systems among experienced clinicians [22–24]. Consistent with these findings, app use also did not result in clear improvements in objective ALS performance metrics, including time to rhythm analysis, medication administration, airway management, or chest compression quality. The only notable difference was the earlier initiation of chest compressions in the control group during ventricular fibrillation. This isolated finding may reflect a small procedural overhead associated with initial interaction with the application, similar to effects previously described for smartphone-guided CPR in untrained rescuers [25]. Since no differences were observed in subsequent intervals—including rhythm analysis, defibrillation, medication timing, or no-flow time—this initial overhead does not appear to propagate throughout the remainder of the resuscitation algorithm in trained teams. Consistent with these findings, app use also did not result in clear improvements in objective ALS performance metrics. The use of the resuscitation app did not lead to clear improvements in adherence to ALS performance metrics, including time to rhythm analysis, medication administration, airway management, and chest compression quality. The most clearly detected performance difference observed was an earlier initiation of chest compressions in the control group during ventricular fibrillation. It is conceivable that interaction with the application introduces a brief cognitive and procedural overhead during the early phase of resuscitation, when rapid action is critical; however, the present study was not sufficiently powered to characterize this potential effect conclusively. Notably, time to first defibrillation was numerically shorter in the intervention group than in the control group (112.8 ± 40.9 s vs. 133.3 ± 31.1 s), although this difference might be explainable by chance (p = 0.224), likely reflecting the limited statistical power associated with the small sample size. Given that time to first shock is widely regarded as one of the most critical determinants of outcome in VF/pulseless VT arrests, this numerical trend warrants attention despite its exploratory nature. It is plausible that the explicit visual and acoustic prompts provided by the app—guiding teams through rhythm analysis and, where indicated, prompting defibrillation—actively contributed to earlier shock delivery in the intervention group, rather than the delay in the control group reflecting a lack of familiarity with ALS guidelines, particularly given that the majority of participants were experienced clinicians with regular exposure to resuscitation training and practice. This finding should therefore be interpreted as a hypothesis-generating signal rather than a definitive effect, and warrants confirmation in adequately powered future studies. These findings highlight the importance of baseline competence: when guideline adherence is already high, additional decision-support may offer limited benefit, with the possible exception of time-critical, high-stakes decisions such as defibrillation timing. In contrast, studies involving less experienced providers or training settings have demonstrated clearer improvements in performance and adherence when cognitive aids are used: Grundgeiger et al. (2021) reported improved CPR performance and reduced workload with a cognitive aid [8], while Siebert et al. observed benefits in neonatal resuscitation for trainees [3]. Taken together, these findings suggest that the effectiveness of digital decision-support tools is strongly context-dependent and closely linked to user expertise. The nominal differences in effort, frustration, and time to start of chest compressions should be considered hypothesis-generating, as multiple outcomes and subscales were tested without adjustment for multiplicity. Nonetheless, these findings contribute to the ongoing discussion regarding the role of digital tools in acute care settings. While such tools offer structured guidance and may enhance confidence and reduce uncertainty, their integration into high-performance clinical workflows must be carefully considered. In expert teams, where decision-making is rapid and largely automatized, additional interfaces may compete for attention rather than provide substantial benefit. In line with this, although it represents a different type of intervention, the use of an interactive CPR assist device synchronized with ventilation and defibrillation yielded comparable effects by improving adherence to guideline-based processes and reducing hands-off times in a simulated setting. However, these benefits were primarily observed under controlled conditions and do not necessarily translate into a meaningful advantage for experienced teams operating in real-world, high-performance environments [26]. Conversely, in less experienced teams or high-stress environments with lower baseline adherence, cognitive aids may play a more prominent role in improving performance and patient safety. Limitations This study has several limitations. First, the sample consisted primarily of experienced anesthesiology and critical care professionals, which may limit generalizability to less experienced teams or other clinical settings. Second, the study was conducted in a simulated environment, which may not fully replicate the complexity, stress, and unpredictability of real in-hospital cardiac arrests. Third, the study was conducted at a single centre, which may limit the generalizability of the findings to other institutions or healthcare systems. The relatively small sample size may reduce statistical power to detect subtle differences in performance or workload. Given the feasibility nature of this study, no formal power calculation was performed prior to the study, and the study may therefore have been underpowered to detect clinically relevant effects. Furthermore, video ratings for TEAM and ANTS were performed by two investigator-raters (CE, AK) who were also co-authors involved in study design, data collection, and interpretation, and who were aware of the study hypothesis. Videos were rated in a fixed, chronological rather than randomized order, and raters had access to scenario timing data during rating. Disagreements were addressed by averaging both raters’ scores. While inter-rater reliability (ICC) was high, this reflects consistency between raters and does not exclude the possibility of shared expectancy bias, particularly given the visible presence or absence of the tablet, which likely made allocation difficult or impossible to blind. An independent, blinded third rater or a sensitivity analysis on a subset of videos was not feasible within the scope of this study but is recommended for future research. Cardiopulmonary resuscitation events are relatively infrequent in routine clinical practice, and consequently, the use of the tablet-based assist device is also likely to be sporadic. More frequent exposure and routine use might improve familiarity, reduce cognitive load, and potentially alter the observed effects on performance. Furthermore, the extent and pattern of app use during the intervention scenarios varied between participants and was not standardized; although interactions with the app were captured on video, they were not formally rated or quantified. Consequently, it cannot be fully determined to what extent the higher Effort scores observed in the intervention group reflect genuine cognitive overhead associated with the resuscitation task itself, or rather unfamiliarity with the interface, divided attention between the app and the clinical scenario, or interface-related friction. Future studies should systematically capture and analyze app interaction patterns to better disentangle these potential contributing factors. Finally, the study evaluated only short-term effects of real-time app use and did not assess long-term retention, clinical outcomes, or patient-centered endpoints. Conclusion In summary, in this exploratory randomized simulation-based feasibility trial involving predominantly experienced in-hospital ALS providers, app-guided resuscitation support was not associated with a clear improvement in observer-rated teamwork, non-technical skills, perceived workload, or selected ALS process metrics. The findings should be interpreted in light of the small sample, high baseline expertise, and limited statistical power and should not be extrapolated to novice providers, mixed-experience teams, prehospital settings, or educational contexts. Future research should focus on identifying the contexts in which digital decision-support tools provide the greatest benefit. This includes evaluating their impact in novice providers, interdisciplinary teams with variable experience levels, and real-world clinical settings. Additionally, optimization of user interface design and integration into existing workflows may help minimize cognitive overhead and enhance usability during time-critical situations. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Author contributions JK, AK and CE were the principal investigators and designed the overall study. AK and CE supervised data collection and analyses, and both AK and CE supported interpretation of data and drafting the manuscript. AK and CE collected data and prepared data for analyses, and MP, AK and CE analyzed data, interpreted data. JK, AK and CE searched literature and drafted the manuscript. All authors have provided comments and critical revisions to the manuscript. Funding No funding. Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Paracelsus Medical University, Nuremberg, Germany (IRB-2022-011), as well as the hospital staff council. Written informed consent was obtained from all participants before participation. This simulation-based feasibility study was not prospectively registered. Consent for publication Consent for publication was obtained from all participants prior to publication. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement No datasets were generated or analysed during the current study.