In Vivo Central Venous Line Placement after Simulation Training: Evaluating the Transferability of Technical Performance Using Hand Motion Analysis

This study aims to assess technical performance correlation between central line placement during simulation training on a manikin and in-vivo procedures on real patients using validated hand motion metrics.

Materials and Methods:

Thirty-nine PGY2 anesthesia and interventional radiology (IR) trainees were recruited to perform ultrasound-guided central venous line placement on a standardized manikin during simulation training. Subsequently, they performed the procedure on real patients following the same methodology. All trials were video recorded, and performance was assessed using validated hand motion metrics—Path Length (PL), Rotational Moves (RM), Translational Moves (TM), and Time^1-3. Participants were mentored by IR and anesthesia experts during training and in-vivo performance. Procedure steps were standardized across all trials.
Motion data from the first and final simulation trials were compared using Wilcoxon signed-rank test. Spearman’s correlation was used to assess the agreement between final simulation performance and real-patient performance for both the overall procedure and specific procedure segments⁴. Data are reported as median ± interquartile range and p values were adjusted using Benjamini-Hochberg correction.

Results:

A total of 184 simulation trials and 29 in-vivo trials (3 excluded due to methodological differences) were analyzed. Median number of simulated trials was 5 (IQR: 5–11.5).

Significant performance improvement was observed after training, with lower metrics in the final trials for the needle hand (e.g., RM: 118 ± 63 vs. 164.5 ± 77.5, adj. p< 0.001) and the probe hand (e.g., RM: 66 ± 75.2 vs. 123 ± 110.5, adj. p< 0.001). A significant reduction in procedure time was seen (57 ± 33 vs. 97.4 ± 45.2, adj. p< 0.001).
Using motion data of the entire procedure, no significant correlation was found between the final simulation and the in-vivo performance for all metrics. However, when breaking the procedure down into segments, correlation analysis showed that participants who had fewer translational movements during needle access in simulation performed  the needle access phase in the operating room with less translational movements (Rho=0.65, adj. p=0.002).

Conclusion: Hand motion data show there is transferability of skill gained during simulation to in-vivo performance for at least one portion of a routine interventional procedure. This suggests that specific portions of procedures are more reproducible and may be more appropriate for assessing performance between simulation and in vivo settings.