Journal Logo

Technical Reports

Three-Dimensional Printed Pediatric Airway Model Improves Novice Learners' Flexible Bronchoscopy Skills With Minimal Direct Teaching From Faculty

DeBoer, Emily M. MD; Wagner, Jennifer MS; Kroehl, Miranda E. PhD; Albietz, Joseph MD; Shandas, Robin PhD; Deterding, Robin R. MD; Rustici, Matthew J. MD

Author Information
Simulation in Healthcare: The Journal of the Society for Simulation in Healthcare: August 2018 - Volume 13 - Issue 4 - p 284-288
doi: 10.1097/SIH.0000000000000290
  • Free


Flexible bronchoscopy (FB) is an invasive procedure performed by pulmonologists, intensivists, and surgeons. Flexible bronchoscopy is a critical tool to assess for atypical pulmonary anatomy and to obtain samples from a bronchoalveolar lavage, which is needed to diagnose infections, neoplasms, and rare lung disease. Severe complications during pediatric FB are rare1; however, in adult interventional FB, trainees take longer to perform procedures than skilled attending bronchoscopists.2 A delay in the trainee's ability to identify lung lobes can increase procedure time and anesthesia time, which has been associated with postprocedural complications including hypoxemia and hypoventilation from poor lung expansion.3,4

A survey of pediatric pulmonary fellowship directors demonstrated that no consistent method is used to train fellows in the skills of FB5 and no accepted assessment tool exists to specifically evaluate competency in pediatric FB. Randomized controlled trials have shown that organized training using simulators and box models results in improved time to surgical competence.3,4,6,7 Three-dimensional (3D) printing has been successfully used in anatomy education and surgical education, with improvements in spatial understanding, task completion times, and provider satisfaction.8 Current FB simulators are rarely used by pediatric programs because they are cost prohibitive to many academic centers (upwards of US $60,000).6,9 Bronchoscopy of a 3D printed airway model provides a similar experience to bronchoscopy on live patients.10 Few of these bronchoscopy simulators have been validated for widespread utility in training.

The purposes of this article are to provide a description of the development of a 3D printed airway trainer and to provide preliminary data on its cost and its utility in teaching FB.


Airway Model

A pediatric airway model was locally created by the Department of Bioengineering at University of Colorado Anschutz Medical Campus. Volumetric chest computed tomography images obtained for clinical purposes of a 15-year-old adolescent boy without respiratory disease were used for airway and vessel segmentation using commercially available imaging software. Bronchi were modeled completely to the segmental branches and partially to the subsegmental branches. Imaging files were converted into an .stl file. This file was used to print a solid model of the airway tree and local vasculature using 3D printing technology. The airway tree was coated in flesh-colored silicone and the interior was dissolved leaving a hollow silicone shell. The hollow airway model and local vasculature were fixed in a flexible transparent silicone block to provide structure and allow external viewing of airway and vessel anatomy (Fig. 1). External holes were drilled into the model and six exchangeable colored markers were placed in segmental airways of six areas of the lungs (5 lobes and lingula). The cost of materials and labor to produce the described trainer was approximately US $2500: $500 estimated for creation of computer model, $500 for cost of materials, and approximately $1500 for manufacturing. This was produced on a fusion deposition modeler-type 3D printer that costs approximately US $32000 to purchase. Cost of the time to create the computer airway segmentation is included in this calculation. It can be assumed that a person familiar with the anatomy and with access and training on image segmentation could create a computer model in less than 12 hours. Using Internet search tools, a market survey was conducted to determine costs of other commercially available bronchoscopy trainers. Because Children's Hospital Colorado has six fellow trainees per year, the estimated annual cost to educate trainees was determined by dividing the cost of each model by 6 (Table 1).

A, Airway model with exchangeable colored markers placed into subsegmental bronchi was created using 3D printing technology. Internal views: B, three segments of the right upper lobe but a marker is not visible from this scope position; C, orange marker located in a segment of the left lower lobe.
Summary of Training Costs

Education Pilot

Postgraduate year 2 pediatric residents were enrolled in this FB training pilot at the beginning of their 1-month inpatient pediatric pulmonology rotation. The Colorado Multiple Institutional Review Board approved this study and each study participant signed informed consent.

After enrollment, all participants received 12 minutes of introduction to the bronchoscope and airway anatomy and performed a prestudy assessment. This timed and observed FB assessment on the airway model was the primary method of evaluation. Participants were instructed to manipulate a retired 3.0-mm flexible pediatric bronchoscope to six areas of the lungs then locate and state the color of the marker in that lung area aloud to a study observer in the room who would evaluate responses after the completion of the FB. If multiple color-lung area combinations were reported, participants were scored based on the final time a color was mentioned. Three combinations of six colored markers were used in the assessments. The length of the procedure was timed; residents had a maximum of 600 seconds (10 minutes) to complete each assessment. During all assessments, the model was covered by a hospital blanket to prevent visualization of the bronchoscope's position.

After the prestudy assessment, residents were randomized to either the simulation trainee or resident control group in a blinded manner. Based on preliminary data, we estimated to have 80% power at a 0.05 significance level to detect a difference in time to complete an FB task between 18 trainees and 5 controls at posttraining assessment and therefore block randomized in a 2:1 manner. During the next month, simulation trainees performed up to three self-directed 15-minute practice sessions on the airway trainer, each followed immediately by an interval assessment, which were conducted identically to the prestudy assessment. The purpose of the interval assessments was to ensure that each participant attempted at least one complete bronchoscopy after each training session. During the practice sessions, residents were allowed to attempt FBs as many times as they wanted and were able to visualize the airways externally through the transparent silicone casing. Exact position in the model could not be viewed by looking at the model; however, the light from the bronchoscope revealed approximate position of the bronchoscope. They received one additional 15-minute practice session on the airway trainer with guidance from a pediatric pulmonologist. Practice sessions were scheduled according to availability of each residents schedule and had to accommodate duty hours, a 1 in 4-day call schedule (24-hour call shifts) as well as weekly half-day continuity clinic. The control group received no further practice or training over the month. All participants performed a poststudy assessment at the end of the month and a delayed assessment at least 2 months after the rotation to evaluate for retention of FB skills. Before the prestudy assessment and after the poststudy assessment, residents used a five-point Likert scale to rate their agreement with the following statement: “I can reliably reach each lobe of each lung using a bronchoscope in a pediatric patient.”

The primary outcome was the difference in the number of correctly identified lung areas at the poststudy assessment.

Secondary outcomes included time to complete the assessment (with maximum 600 seconds), number of colored markers visualized (out of 6 total), performance on the delayed assessment, and confidence measured on a five-point Likert scale.

As part of a limited expert assessment of the system, six local attending pediatric pulmonologists without experience with the airway model performed an FB on this model and were timed using the same assessment protocol used in this study.

Statistical Analysis

Median and interquartile ranges (IQRs) were reported, as distributions of the outcomes were nonnormal. Wilcoxon rank sum tests were used to compare performance outcomes between simulation trainees and control residents. Likelihood and confidence intervals are reported for simulation residents to achieve 6/6 correct marker identification. Effect size of the number of lung areas correctly identified and time to completion are reported at the poststudy assessment. Effect size was calculated as (mean of training group – mean of control group)/standard deviation. Participant confidence Likert data were dichotomized for ease of evaluation to agree (agree, strongly agree) or not agree (strongly disagree, disagree, neutral) and then was analyzed using Fisher exact test. Statistical significance was defined as P < 0.05.


Twenty-seven residents were approached and enrolled. No residents declined enrollment. Eighteen were randomized to the simulation trainee group and nine to the control resident group. All residents stated that they had performed zero FBs on patients or simulators in the past and had observed a median of one FB on a live patient (IQR = 0–1.5) before the study and one FB (IQR = 0.5–2) at the end of the study. No participant from either group performed a bronchoscopy on a live patient during the study period. Of the 18 simulation trainees, 13 completed all four practice sessions, four completed three, and one completed two sessions.

During the prestudy assessment, there was no difference in the median number of lung area markers correctly identified (1 vs 1, P = 0.707) or time to completion of the assessment (600 vs 600 seconds, P = 0.432) between simulation trainee and control group (Tables 2, 3). Simulation trainees identified a significantly greater number of lung area markers correctly during the poststudy assessment (6 vs 1.5, P < 0.001, effect size = 1.8) and had a significantly shorter time to completion of the assessment (102 vs 600 seconds, P < 0.001, effect size = 1.6) (Tables 2, 3). By the third training session, 76.5% of simulation trainees had 6/6 accuracy (95% confidence interval = 54–99). After training, more simulation trainees than control residents agreed/strongly agreed that they could consistently identify lung anatomy during an FB (12/14 simulation trainees vs. 0/6 control residents, P = 0.0026).

Number of Markers Identified Correctly
Time to Completion of the Assessment

All residents completed a prestudy assessment, and all but one control completed the poststudy assessment. Of the enrollees, 6 (33%) of the 18 simulation trainees and 2 (22%) of the 9 control residents were lost to follow up and did not complete a delayed assessment.

During the expert assessment, the six pulmonary attendings correctly identified all six markers with a median (range) completion time of 29 (20–69) seconds.

Estimating a fellowship size of six people, the annual training costs per student are 43% to 250% less when using the training models described in this work compared with other trainers (Table 1). Annual maintenance and service costs are not included in these estimates for other trainers or for the 3D printer. The trainer presented here will not require annual maintenance or upkeep beyond routine cleaning with water.


A realistic anatomic model was created of a pediatric airway for learners to practice manipulating the bronchoscope and to learn spatial lung anatomy. With four short and mostly self-directed training sessions, an improvement was observed in the identification of lung lobes on the model and a decrease in procedural time from prestudy to poststudy assessments. Although the pilot was not powered to measure retention, there was a trend toward improved performance of simulation trainees months after completing the training.

In many pediatric pulmonology fellowships, FB competency is measured by the number of FBs completed, rather than by skills assessment.5 The number of opportunities for pediatric FB procedures may be limited at some training sites, so it is important that learners maximize the impact of each procedure they are able to perform. Practicing FB skills on a 3D printed airway model may improve basic bronchoscope skills and knowledge of airway anatomy, which could allow trainees to more safely participate in FB on live patients.

A pediatric airway model may also facilitate standardization of FB and FB training across centers. Although procedural training varies, key aspects of procedural skills curricula focused on novice learners should include spaced repetition, feedback, and independent practice. Spaced repetition models where deliberate gaps are introduced between training encounters have been shown to improve retention of knowledge and skills for a variety of learners.11 A spaced repetition model of training was attempted but was condensed and somewhat irregular because of variability in resident's schedules and a short overall study period of 1 month. Feedback has been shown to be one of the most effective ways to improve learning and is particularly effective when it is integrated into successful completion of a motor task.12,13 The model's transparent silicone casing allowing visualization of anatomical relationships between the thoracic vasculature and the airways is unique compared with other models.10,14 The model's use of exchangeable colored markers is also unique and provides direct feedback to the trainee about the location of the bronchoscope and the correct pathway to each lung area.

The concept of deliberate practice, in which trainees practice micro skills in an iterative process that provides feedback about their performance, has been repeatedly associated with faster progression toward expertise.15,16 This model could be employed using a deliberate practice model where students focus on improving their skills in bronchoscope position (colored markers) or the efficiency of scope navigation (time to completion) at any given training session based on their areas of weakness. In training of lumbar punctures (LP), it has been shown that learners retain more skills if they are given less direct instruction and instead are able to struggle independently on an LP model.17 This FB model provided direct feedback to learners about the position of the bronchoscope and also encouraged independent exploration of skills by limiting the exposure to an instructor who may provide too much help and hence decrease long-term skill retention. By changing the pattern of the exchangeable colored markers, instructors could conduct a standardized assessment of trainee's basic FB skills that can be repeated over time with new color combinations. The attributes of the trainer allow for an objective repeatable assessment of anatomic knowledge and basic FB motor skills, two components required in a skill-based assessment of FB competency and entrustability.

This model is cost-effective compared with other bronchoscopy trainers. Virtual reality bronchoscopy simulators are very expensive ranging from US $60,000 to US $76,000. Physical model bronchoscopy trainers are currently sold for approximately US $2500 to US $4300, with fewer features than are available on the 3D printed model described in this study. Three-dimensional printers are also very expensive but can be used for presurgical planning,18 and the costs of the printer can be offset by its clinical uses and can also be used to facilitate student coursework in manufacturing using 3D printing. For institutions without access to a 3D printer, service bureaus can print models for an upcharge of approximately 10%.


Curricula using task trainers are effective at teaching procedural skills and advancing learners to competence.3,4,6,7 It has not yet been established whether the basic FB skills gained during training on this model transfer to live patients in which other conditions not replicated by this model (ie, lung movement, mucous production, complications from the underlying pathologic reason an FB is indicated) may affect FB performance. Our expert assessment revealed that six pulmonologists without experience with the airway model correctly identified all six markers with a faster median completion time of 29 seconds. This supports the validity of the model as a measure of at least some aspect of skill gained during training; however, our use of the model as both trainer and evaluation method is a confounder that limits what we can say about real-world performance.

Other limitations in this study included using second-year residents as a convenience sample. Residents were chosen as the study sample because of insufficient numbers of pediatric pulmonology fellows at this institution (2 per year). At this training program, FB is only performed by pediatric pulmonologists and rarely by pediatric surgeons; therefore, it is highly unlikely any resident who does not pursue an elective month in pediatric pulmonology would be exposed to FB during other points of their training. Given that pediatric FBs are unlikely to be performed by graduating residents not pursuing a career in pediatric pulmonology, the residents enrolled in this study may have less motivation to improve FB skills as compared with fellows pursuing a fellowship in pediatric pulmonology and hence the training may have a larger effect if used with incoming fellows.

Future Directions

Potential airway trauma was not assessed, and it is possible that speed the learners demonstrated on the model does not accurately reflect true bronchoscopy skills on live patients. Future studies should consider evaluating transfer of skill from an anatomically accurate 3D printed model to FB on live patients in a nationwide or worldwide program. These studies will have to overcome challenges around the lack of a validated pediatric FB assessment tool and the low number of total pediatric pulmonology fellows at any single program. New models will be needed to evaluate advanced bronchoscopy skills such as scope manipulation in patients with airway abnormalities, copious secretions, or substantial airway inflammation. In addition, other methods may be needed to test knowledge of indications and contraindications, reasoning skills for atypical patients, management of complications encountered during the procedure, or diagnostic bronchoscopy maneuvers (bronchoalveolar lavage, suctioning, and biopsy). There are now plans to create models of different ages and with different airway pathologic findings to help train some of these advanced skills. If the model were to be used to train advanced learners or as a summative assessment, markers of more colors could be placed at different depths to increase the difficulty of the test and to make it more difficult for the trainee to determine a pattern of colors.

In conclusion, we present evidence that an anatomically accurate airway trainer can be created using 3D printing technology and when used with exchangeable colored markers can facilitate training of basic FB skills using a deliberate practice model. Creation of similar anatomically accurate models could be completed for other basic endoscopic procedures including laryngoscopy, esophagogastroduodenoscopy, and colonoscopy. Given the somewhat rigid nature of the materials involved, procedures involving fixed, hollow anatomies such as the nasal passages, ear canals, or oral pharynx seem most amenable to 3D printing using this technique. Intestinal endoscopic trainers could also be created with this technology; however, this would require computed tomography images with noncollapsed segments of bowel and also would not replicate gut motility. Educators and clinicians could consider this concept for procedures that have limited trainers beyond learning on patients and are performed on fixed hollow anatomic sites.


1. DeBoer EM, Prager JD, Kerby GS, Stillwell PC. Measuring pediatric bronchoscopy outcomes using an electronic medical record. Ann Am Thorac Soc 2016;13:678–683.
2. Stather DR, MacEachern P, Chee A, Dumoulin E, Tremblay A. Trainee impact on procedural complications: an analysis of 967 consecutive flexible bronchoscopy procedures in an interventional pulmonology practice. Respiration 2013;85:422–428.
3. Nagendran M, Gurusamy KS, Aggarwal R, Loizidou M, Davidson BR. Virtual reality training for surgical trainees in laparoscopic surgery. Cochrane Database Syst Rev 2013;8:CD006575.
4. Nagendran M, Toon CD, Davidson BR, Gurusamy KS. Laparoscopic surgical box model training for surgical trainees with no prior laparoscopic experience. Cochrane Database Syst Rev 2014;1:CD010479.
5. Leong AB, Green CG, Kurland G, Wood RE. A survey of training in pediatric flexible bronchoscopy. Pediatr Pulmonol 2014;49:605–610.
6. Colt HG, Crawford SW, Galbraith O 3rd. Virtual reality bronchoscopy simulation: a revolution in procedural training. Chest 2001;120:1333–1339.
7. Goova MT, Hollett LA, Tesfay ST, et al. Implementation, construct validity, and benefit of a proficiency-based knot-tying and suturing curriculum. J Surg Educ 2008;65:309–315.
8. Langridge B, Momin S, Coumbe B, Woin E, Griffin M, Butler P. Systematic review of the use of 3-dimensional printing in surgical teaching and assessment. J Surg Educ 2017;S1931-7204(17):30183–30186.
9. Wahidi MM, Silvestri GA, Coakley RD, et al. A prospective multicenter study of competency metrics and educational interventions in the learning of bronchoscopy among new pulmonary fellows. Chest 2010;137:1040–1049.
10. Byrne T, Yong SA, Steinfort DP. Development and assessment of a low-cost 3D-printed airway model for bronchoscopy simulation training. J Bronchol Interv Pulmonol 2016;23:251–254.
11. Pashler H, Rohrer D, Cepeda NJ, Carpenter SK. Enhancing learning and retarding forgetting: choices and consequences. Psychon Bull Rev 2007;14:187–193.
12. Ericsson KA. Deliberate practice and the acquisition and maintenance of expert performance in medicine and related domains. Acad Med 2004;79:S70–S81.
13. Kruglikova I, Grantcharov TP, Drewes AM, Funch-Jensen P. The impact of constructive feedback on training in gastrointestinal endoscopy using high-fidelity Virtual-Reality simulation: a randomised controlled trial. Gut 2010;59:181–185.
14. Al-Ramahi J, Luo H, Fang R, Chou A, Jiang J, Kille T. Development of an innovative 3D printed rigid bronchoscopy training model. Ann Otol Rhinol Laryngol 2016;125:965–969.
15. Moulaert V, Verwijnen MG, Rikers R, Scherpbier AJ. The effects of deliberate practice in undergraduate medical education. Med Educ 2004;38:1044–1052.
16. Ericsson KA. Deliberate practice and acquisition of expert performance: a general overview. Acad Emerg Med 2008;15:988–994.
17. Brydges R, Nair P, Ma I, Shanks D, Hatala R. Directed self-regulated learning versus instructor-regulated learning in simulation training. Med Educ 2012;46:648–656.
18. George E, Barile M, Tang A, et al. Utility and reproducibility of 3-dimensional printed models in pre-operative planning of complex thoracic tumors. J Surg Oncol 2017;116:407–415.

Bronchoscopy training; simulation; pediatric pulmonology; task trainer; spaced learning

Copyright © 2018 Society for Simulation in Healthcare