Training Programs for Thyroid Biopsy and Ablation: A Systematic Review

Abstract

Objective:

To evaluate the structure, components, and outcomes of educational programs designed to train clinicians in thyroid fine-needle aspiration (FNA) and ablative procedures.

Methods:

A systematic review of five databases was conducted for studies published from 2000 through April 2025 evaluating training interventions for thyroid FNA or ablative techniques. Two independent reviewers screened, extracted data, and assessed risk of bias for all eligible studies.

Results:

Ten studies met inclusion criteria: nine addressed FNA training, and one addressed radiofrequency ablation (RFA). All studies were rated at moderate to high risk of bias. Common instructional components included didactic sessions (n=5), hands-on simulation (n=9), supervised clinical experience (n=4), and multi-modal approaches (n=7). Simulation models varied in complexity, including food-based models, cadavers, and 3D-printed phantoms. Reported educational outcomes included improved learner confidence and perceived realism (reaction outcomes); enhanced accuracy, procedural speed, and knowledge (learning outcomes); and improvements in biopsy adequacy, malignant diagnostic yield, and procedural wait times (clinical outcomes). The single RFA study demonstrated a median nodule volume reduction of 82% at 12 months following simulation-based and supervised training.

Conclusion:

Structured, multi-modal training programs in thyroid FNA and RFA improve both learner performance and clinical outcomes. These findings support the integration of simulation-based learning and supervised procedural experience into thyroid intervention training.

Introduction

Thyroid nodules are common, particularly among women and with increasing age.¹ Diagnostic evaluation typically begins with ultrasound imaging of the thyroid, which provides sonographic details, such as nodule size and features linked to malignancy risk to help patients and clinicians decide whether a thyroid biopsy is necessary.^2–4 Thyroid fine needle aspiration biopsy (FNA) is an overall well tolerated and safe procedure.⁵ Yet, patient satisfaction and clinical efficacy (e.g., diagnostic rates) is influenced by the degree of clinician expertise performing the procedure, making adequate training crucial for optimal patient care.^{6, 7} Nonsurgical procedures for the care of patients with thyroid nodules, such as radiofrequency ablation (RFA), have proven to be safe and effective in reducing the burden associated with large thyroid nodules, with expanding indications for other clinical scenarios.⁸ Similar to thyroid FNA, operator expertise critically influences both patient experience and clinical outcomes of these procedures.⁸

Training programs for thyroid FNA and RFA can include different educational activities such as 1) didactic lectures/instruction, 2) hands-on practice on models and simulators, 3) supervised clinical experience, 4) ongoing assessment and feedback, and 5) a combination of these and others.^9–26 In recent years, there have been significant advancements in thyroid biopsy training models, including high-fidelity simulators, 3D-printed thyroid models, and ultrasound-guided biopsy trainers.^{9, 10, 14, 17} These newer models offer enhanced practicality and enable more realistic training experiences. Similarly, the expansion of ablative indications in thyroid pathology has prompted the creation of structured training programs.²⁶ Yet, the potential impact of these technological innovations and training program structures on learners’ proficiency and patient outcomes is not clear.

The goal of this systematic review was to evaluate the available evidence on training programs for thyroid biopsy and ablative techniques. By synthesizing findings across various studies, this review provides insights into training program components and educational outcomes, contributing to enhancing training strategies in thyroid biopsy and ablative techniques.

Materials and methods

Protocol registration and reporting

This systematic review was conducted by following a protocol and adhering to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.²⁷ This systematic review protocol was registered in PROSPERO (CRD42023490675).

Eligibility criteria

We included studies that evaluated training programs for thyroid FNA or ablation. Training programs were defined as structured activities that allowed learners to acquire proficiency in procedural skills.^{28, 29}. We included studies that involve any medical professional eligible to perform thyroid FNA or ablation in patient care as learners (e.g., medical students, residents, fellows, attendings). We included studies reporting educational outcomes, including both learner (e.g., training satisfaction) and/or clinical outcomes (e.g., nondiagnostic rate).^{30, 31} Observational, quasi-experimental, and randomized trials were eligible for inclusion.

Studies that included a description of an educational training program without clearly reporting educational or clinical outcomes (e.g., evaluation), reported on educational programs targeting mixed pathology (e.g., thyroid and parathyroid) or unrelated to thyroid nodules, cancer, thyroid FNA, or ablation, and where participants did not complete a formal educational activity were excluded. We also excluded studies that did not focus on professionals qualified to perform the procedure in the clinical setting or reporting on participants’ hypothetical reactions to the suitability of the educational activity for use in official training. Systematic reviews, commentaries, and narratives were excluded.

Information sources and search strategy

A comprehensive search of several databases from 2000 to April 2025 in any language was conducted. The databases included Ovid MEDLINE(R) and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, and Daily, Ovid Embase, Ovid Cochrane Central Register of Controlled Trials, Ovid Cochrane Database of Systematic Reviews, Scopus, and Web of Science. The search strategy was designed and conducted by an experienced librarian with input from the study’s principal investigator. Controlled vocabulary, supplemented with keywords, was used to search for training programs for thyroid biopsy and ablation for medical personnel. The actual strategy listing all search terms used is available in the Appendix.

Data management

References obtained using the search strategy were stored on Covidence, where they were screened for inclusion. Covidence is a web-based collaboration platform that streamlines the production of systematic reviews.³² Covidence was utilized for title and abstract screening and full text review.

Selection process

Two independent reviewers screened each reference in a duplicate manner. The screening process was divided into a title/abstract phase and a full-text phase. During the first phase, any decision conflict between reviewers was passed on to the full-text phase, where any conflict was resolved by either consensus or intervention of a third reviewer. Before the start of each phase, pilot testing and calibration were performed between reviewers to secure adequate inter-rater agreement (defined as a kappa index equal to or higher than 0.7).

Data collection process

Two independent members of the research team worked to collect information from each included study in a duplicate manner. Data collection conflicts were resolved by either consensus or intervention of a third reviewer.

Data items

We collected general study information such as first author, year of publication, country, setting, study design, sample size, participation rate if applicable, and funding. Participant demographics such as age, sex, training, specialty, and previous experience were also collected. We also collected details regarding the educational intervention (description, components, definitions, as applicable), educational outcomes (e.g., satisfaction, behavior), and clinical outcomes of intervention effectiveness (e.g., nondiagnostic biopsy rate, if applicable).

Outcomes

We followed the Kirkpatrick’s evaluation of educational programs model to identify and summarize the outcomes of thyroid FNA and ablation training programs.³¹ This framework was designed to help educators identify and conceptualize their training programs’ goals and evaluate the extent to which they are successfully meeting the educational program’s objectives. It includes four categories of outcomes: 1) reaction (e.g., students felt confident and satisfied with the thyroid biopsy training), 2) learning (e.g., knowledge of thyroid biopsy techniques increased after training), 3) behavior (e.g., participants applied biopsy techniques correctly in clinical settings), and 4) results (e.g., increase in successful, unsupervised thyroid biopsies by participants).

Risk of bias in individual studies

Risk of bias was assessed with the Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomized studies as needed.^{33, 34} This process was completed independently by two reviewers and followed a standardized piloted extraction form.

Data synthesis

We present a narrative summary of the identified educational programs and measured outcomes mapped to the Kirkpatrick framework for educational programs. This process required merging, when possible, outcomes that address the same construct and classifying these items according to Kirkpatrick’s framework for educational programs. Due to limited data and heterogeneity of study designs and outcomes measured, no meta-analysis was performed.^{31, 35}

Results

Study selection and characteristics

We identified 1,762 studies in our literature search and 10 met the inclusion criteria. (Figure 1).^{6, 9–12, 14, 15, 17, 18, 24, 36} Included studies were published between 2015 and 2022; most were conducted in the United States and Canada. (6/10).^{6, 12, 14, 15, 17, 36} Nine focused on training programs for thyroid FNA ^{6, 9, 11, 12, 14, 15, 17, 24, 36} and one for thyroid RFA.¹⁸ Ten studies evaluated participants before and after the training.^{6, 11, 12, 14, 15, 17, 18, 24, 36} One study implemented a comparative cohort between specialist and non-specialist participants with before and after training assessments.²³ The sample size ranged from 3 to 45 participants, with 135 participants included across all studies. Some studies included only attendings (3/10),^{6, 24, 36} others included only trainees (5/10),^{11, 12, 14, 17, 18} with two studies including both (2/10).^{9, 15} (Tables 1 and 2).

Figure 1.

Study selection.

Table 1.

General Description of Studies

First Author, Year, Country	Participants	Setting	Sample	Time of assessment and type of study	*Outcomes	Thyroid Training Model	Objectives
FNA Training
Gilbert, 2016, US	Endocrinology and cytopathology fellows	Academic	4	One time after training. Single cohort one-time assessment	4	No physical model (design a novel Multidisciplinary Thyroid Biopsy Clinic)	Improve the educational experience of endocrinology and cytopathology fellows. Provides fellows the opportunity to perform multiple biopsies—a maximum of 3 initial passes for each patient—that are evaluated on-site immediately and assessed for quality of technique and adequacy of the specimens.
Atri, 2021, Canada	Trained sonographers	Academic	4	Before and after training. Single cohort two-time assessment	4	Thyroid phantoms (blue phantom, Sarasota, Florida)	To assess consistency and long-term progress in thyroid biopsy performed by trained sonographers under supervision of a radiologist.
Luz, 2022, Brazil	Radiology residents	Academic	20	Before and after training. Single cohort two-time assessment	1, 2	Food-based model (Chicken hearts, three-layer pork meats, gel bottles, RF Medical Radiofrequency Generator and RF Electrodes)	To evaluate the efficacy of a training program in ultrasound-guided fine needle puncture using a cost-effective model.
Baba, 2023, Japan	Physicians who specialized in thyroid gland treatment, senior and junior residents, and students	Academic	45	Before and after training (two training groups, experts and trainees). Comparative cohort two-time assessment	2	Thyroid phantoms (two phantoms)	To evaluate the effectiveness of puncture training using our phantom for junior and senior residents and medical students.
Davis, 2018, US	All senior general surgery residents	Academic	12	Before and after training. Single cohort two-time assessment	1,2	Food-based model (olive target) and thyroid phantoms (Blue Phantom Ultrasound Central Line Training model©)	(1) Allow general surgery residents to learn an important skill while providing a convenient new service for our patients; (2) Measure resident safety and proficiency in performing US-guided FNA; and (3) Evaluate resident feedback regarding the experience, as well as the possibility of adding this to the general surgery curriculum.
Ghai, 2015, Canada	Sonographers	Academic	4	Before and after training. Comparative cohort two-time assessment	4	Thyroid phantoms (blue phantom, Sarasota, Florida)	To develop and implement a program where selected sonographers would be trained to perform thyroid biopsies independently under the supervision of a radiologist, with the goal of improving efficiency and quality.
McCrary, 2017, US	Medical students with minimal US training	Academic	17	Before and after training. Single cohort two-time assessment	2	Cadaver	To determine if a fresh cadaver model for the instruction of ultrasound –guided (FNA) of thyroid nodules is a practical method for instruction.
Samargandy, 2022, Canada	Otolaryngology—Head & Neck Surgery residents	Academic	14	Before and after training. Single cohort two-time assessment	1,2	Thyroid phantoms (Blue Phantom Ultrasound Central Line Training model©)	To compare the efficiency between long-axis and short-axis thyroid UGFNA techniques in trainees. Our secondary outcomes were to define the comfort level and learning curves of trainees
Zhang, 2019, China	Radiology residents	Academic	12	Before and after training. Single cohort two-time assessment	2	Food-based model (tofu model)	To construct a cost-effective and easily attainable training phantom for ultrasound-guided fine needle thyroid nodule targeting punctures and to evaluate its usefulness for puncture training.
RFA Training
Li, 2022, Taiwan	Endocrinologists who never performed echo-guided aspiration cytology and RFA	Academic	3	Before and after training. Single cohort two-time assessment	4	Food-based model (matrix consisted of a reproducible combination of gelatin, Metamucil, maisena (cornstarch), and water)	To investigate the feasibility of the food-assisted and -simulated training model of thyroid RFA in three young endocrinologists who had never performed echo-guided aspiration cytology and thyroid RFA in real patients.

Abbreviations:FNA:Fine Needle Aspiration; RFA: Radiofrequency Ablation; ROSE:rapid on-site evaluation;;

^*Outcomes: (1) reaction (2) learning (3) behavior (4) clinical outcomes

Response rate: Not reported, except for Samargandy, 2022 where fourteen out of seventeen eligible residents joined the study.

Funding: Not reported, except for Samargandy, 2022 reported funding from a foundation. And Luz, 2022, Baba, 2023 and McCrary, 2017 reported no funding.

Table 2.

Summary of training programs

First author	Specialty	Previous Experience	*Teaching strategies used			**Outcomes
			Didactic instruction	Training	Interactive
Atri	Radiology	More than 10 years of experience performing US examinations	X	X	X	Clinical outcomes
Baba	Surgical Oncology	The physicians were specialized in breast and thyroid gland treatment. The residents and medical students do not have prior experience.	X	X	-	Learning
Davis	Surgery	None	X	X	X	Reaction; Learning
Ghai	Sonography	Had more than 10 years of experience in performing US examinations	X	X	X	Clinical outcomes
Gilbert	Endocrinology and cytopathology	NR	-	-	-	Clinical outcomes
Li	Endocrinology	None	X	X	X	Reaction; Learning
Luz	Radiology	None	-	X	-	Clinical outcomes
McCrary	NR	Minimal US training/no students had previous experience related to thyroid or neck US.	-	X	-	Learning
Samargandy#	Otolaryngology	Mixed experience (with and without experience)	-	X	-	Reaction; Learning
Zhang	Radiology	None	-	X	-	Learning

NR = Not reported

^#Samargandy: age range 25–34 and women 14%.

^*Description of teaching strategies used: 1. Didactic (instructions, lecture, videos), 2. Training (on phantoms, hands-on practice on models and simulation), 3. Interactive session (direct clinical experience/skills under supervision, debriefing),

Risk of bias

Overall, the included studies were considered to have a moderate to high risk of bias due to study design limitations and presence of potential confounders. (Table 3).

Table 3.

Risk of Bias.

Study ID	Selection							Comparability	Outcome
	Representativeness of the exposed cohort	Selection of the non exposed cohort/comparison group	Ascertainment of exposure				Demonstration that outcome of interest was not present at start of study	Comparability of cohorts on the basis of the design or analysis	Assessment of outcome (					Was follow-up long enough for outcomes to occur	Adequacy of follow up of cohorts
			secure record	structured interview or survey	written self report	ND			secure record	structured interview where blind to case/control status	interview not blinded to case/control status	written self report or medical record only	ND
Gilbert 2016	somewhat representative of the average in the community	drawn from a different source	X				yes	N/A				X		yes	no statem
Baba 2023	somewhat representative of the average in the community	drawn from the same community as the exposed cohort	X				no	N/A					X	yes	complete follow-up - all subjects accounted for
Samargandy 2022	no description of the derivation of the cohort	no comparison group				X	yes	N/A						yes	complete follow-up - all subjects accounted for
Li 2022	truly representative of the average in the community	no comparison group		X			yes	N/A			X			yes	complete follow-up - all subjects accounted for
Luz 2022	somewhat representative of the average in the community	no comparison group	X				yes	N/A	X					yes	complete follow-up - all subjects accounted for
Atri 2021	truly representative of the average in the community	no comparison group	X	X			yes	N/A	X		X			yes	complete follow-up - all subjects accounted for
Zhang 2019	somewhat representative of the average in the community	no comparison group	X				yes	N/A	X			X		yes	complete follow-up - all subjects accounted for
Davis 2018	somewhat representative of the average in the community	no comparison group	X				no	N/A	X			X		yes	subjects lost to follow up (small number lost - > 70 % or description provided of those lost)
McCrary 2017	selected group of users eg nurses volunteers	no comparison group		X			yes	N/A			X			yes	complete follow-up - all subjects accounted for
Ghai 2015	truly representative of the average in the community	no comparison group		X			yes	N/A					X	yes	complete follow-up - all subjects accounted for

Abbreviations: ND: No Description

Overview of Training Programs

Most training programs included hands-on practice for thyroid biopsy, didactic instructions, supervised clinical experience with patients, or a combination of these methods. Nine studies included hands-on practice using various models for thyroid FNA,^{6, 9, 11, 12, 14, 15, 17, 24, 36} with some studies using more than one model: five used phantom models (e.g., Blue Phantom, Sarasota, Florida),^{6, 9, 12, 17, 36} five used food-based models (e.g., grapes as targets, chicken breast with olives),^{11, 12, 18, 24} and one used a cadaveric model¹⁴. Five studies incorporated didactic instruction, covering neck anatomy, thyroid malignancy, and FNA techniques.^{6, 9, 12, 24, 36} Four studies provided direct clinical experience, where participants performed biopsies under supervision until they were deemed competent. ^{6, 12, 24, 36} Unique approaches included online theoretical courses (one study)²⁴, simulation with a standardized patient (one study)¹⁰, and using a multidisciplinary thyroid biopsy clinic model (one study).¹⁵ A hands-on practice model was utilized for thyroid ablation, employing chicken hearts, three-layer pork meats, gel bottles, and ablation equipment (Tables 1 and 2).¹⁸

Evaluation of training program outcomes

1. Reaction to Training

Reaction to training outcomes were reported in three studies.^{12, 17, 18} Participants generally found the training models realistic and applicable to real-world scenarios, with significant increases in comfort levels with procedures post-training (Tables 4 and 5). Luz et al. evaluated a training program for thyroid biopsy involving 20 participants including a brief online theoretical module and a single in-person session using a low-cost gelatin-based model.¹⁸ They assessed confidence in performing thyroid FNA before and after the training using a 10-point Likert scale. Confidence scores improved from a mean of 2.5 before training to 8.5 after training (p<0.05), reflecting increased confidence and procedural performance. However, the magnitude of improvement did not significantly differ between first-year and second-year residents. Davis et al. evaluated a training program for thyroid biopsy, including didactic instruction, simulation on a high-fidelity manikin, and supervised clinical experience, involving 12 participants. Before training, 62% of participants were not comfortable performing the procedure; after training all participants felt comfortable performing the procedure. Moderate comfort increased from 15% to 50%, and extreme comfort from 0% to 8% post-training; however, no statistical analysis was reported for these changes.¹² Samargandy et al. assessed comfort levels with long-axis and short-axis techniques before and after training, involving 14 participants. The training included pre-session instructional videos and repeated simulation-based practice using the Blue Phantom thyroid model. Before training, 62% of participants were not comfortable with the procedure, which improved to 0% discomfort after the training. Additionally, 71% of participants found the long-axis technique easier to perform, and 64% found the long-axis technique either easier or no different to learn compared to short-axis post-training.¹⁷

Table 4.

Results Skills and Clinical Outcomes

Author	Method for Assessment (survey, chart review, interview)	Outcome Assessed	Result
Atri 2021	Database from the institution which included data on number of procedures performed and final cyopathology report, during program implementation period		Initial Postimplementation (Baseline):
		Malignant;	Malignant: 144 (6%),
		Unsatisfactory;	Unsatisfactory: 534 (21%),
		Patient wait times	Wait time: 49 days;
			Late Postimplementation:
			Malignant: 219 (7%),
			Unsatisfactory: 284 (10%),
			Wait time: 39 days
			P < 0.001
Luz 2022	Direct observational assessment (logged before and after the course by a third-year resident physician)	Time to perform the procedure	Time to perform:
		First year residents
		Before the course	173.7 s±91.28 s,
		After the course	19.2 s±112.8 s
		Second year residents
		Before the course	112.8 s±17.66 s
		After the course	14.3 s±9.36 s)
		Performance time before course for both groups	P = 0.1617
		Performance time after course for both groups
		Interaction between time and groups	P = 0.2974
		Average number of punctures on matrix surface before and after the course
		First year residents
		Before the course	2.2
		After the course	1.1
		Second year residents
		Before the course	1.5
		After the course	1.0
		Punctures before course for groups	P = 0.0747
		Punctures after course for both groups	P = 0.0527
		Interaction between punctures and groups	P = 0.1368
Baba 2023	Unclear	Major axis (first time, second time);	**Senior residents (80 to 53.2, p=0.06)
			Junior residents (199.6 to 119.4, p<0.05),
			Students (214.6 to 157.8, p=0.096);
		Short axis (first time, second time);	**Senior residents (39.9 to 34.4, p=0.516)
			Junior residents (96.4 to 55.2, p<0.05)
			Students (125.9 to 87.1, p=0.09);
			Senior residents (2 to 0)
		Erroneous punctures (first time, second time)	**Junior residents (20 to 0)
			Students (31 to 3);
		Vessel injury (major axis/short axis ratio)	25 to 1
		***Improvement rates	**Junior residents (major axis: 0.58, short axis: 0.62)
Davis, 2018	Knowledge terst	Test scores (% overall knowledge)	p value = 0.05
		% pre-test	40.91
		% pot-test	61.82
Ghai 2015	Direct observational assessment; Chart review (adequate sample data)	Complications	0% major procedural complications;
		Patient wait times
		Before	80.7 days
		After	28.3 days (p<0.001);
		Unsatisfactory rate
		Specialist sonographer	288 (21.4%) (p=0.015)
		Physician	336 (25.4%)(p=0.0497)
		Malignant samples
		specialist sonographer	73 (5.4%)(p=0.823)
		physician	68 (5.1%), (p=0.888)
Li 2022	Follow-up ultrasonography was performed at 1, 3, 6, and 12 months after thyroid RFA	Changes in the volume of nodules (VRR at 12 months)	82%
		Major complications	Hoarseness (2/23)
		Minor complications	NSAIDs for wound pain control (3/23)
McCrary 2017	Survey/evaluation (10 items)	Pre- and Postinstruction Skills Assessment (10 items);	Students showed significant improvement in 8 skills (P < .05)
			1. Correctly identifies nodule with ultrasound probe (9 to 15)
			2. Correctly characterizes nodule as cystic or solid (9 to 15)
			3. Adjusts ultrasound image depth appropriately to optimize view of the nodule, noting the position of adjacent vasculature (6 to 15)
			4. Inserts needle at the appropriate distance and angle from the transducer (1 to 14)
			5. Needle tip is visualized throughout the procedure (2 to 9)
			6. Aspirates while advancing needle tip (2 to 12)
			7. Needle tip is advanced into the nodule (1 to 10)
			8. Successfully aspirates colored water into syringe (1 to 10)
		Pre- and Postinstruction Cognitive Assessment (9 items);	Statistically significant change in cognitive knowledge regarding contraindications (p = 0.001)
		Pre-test	0.529
		Post-test	1.35
Samargandy 2022	direct observational assessment	Average time taken to complete FNA	41 seconds to 52 seconds (p=0.74);
		Success in initial UGFNA attempt	86% successful, 14% unsuccessful
		Successful UGFNA technique	99/118 (83.9%)
Zhang 2019	Doctor assessment: Two doctors with more than 3 years of experience in FNAB judged whether the procedure was successful	Puncture success rates
		Before training	34.4 ± 14.2%,
		After training	66.7 ± 19.5% (p = 0.003);
		Handedness (right, left)	100% right handed;
			No significant difference between right- and left-handed needling (p = 0.101)
		Depth of target attempts (superficial, deep)	No significant difference between superficial and deep (p = 0.458)
Gilbert 2016	Retrospective chart reviews 2 years before and 2 years after instituting the MTBC	Nondiagnostic thyroid FNA biopsies
		Before MTBC,	24%
		After MTBC	5% (P < .001)
		Diagnosis of benign nodules
		Before MTBC,	48%
		After MTBC	78% (P < .001)
		Size of nodules biopsied (Mean size)
		Before MTBC,	23.63 mm
		After MTBC	19.94 mm (P < .001)
		Surgical referrals
		Before MTBC,	21%
		After MTBC	0%

Abbreviations: MTBC: Multidisciplinary Thyroid Biopsy Clinic; UGFNA: Ultrasound-guided Fine Needle Aspiration; VRR: Volume Reduction Rate

^**Baba 2023: Participants were divided into physicians (group A, n = 10), senior residents (group B, n = 8), junior residents (group C, n = 12), and students (group D, n = 15).

^***The improvement rate was defined to identify the difference between the numerators for the first and second time practicing the procedure, and for the first time practicing the procedure

Table 5:

Results Satisfaction and Knowledge

Study	Outcome evaluation	Result
Satisfaction
Davis, 2018	Proportion of participants that felt “not comfortable” with US-guided FNA before and after course	Before	After
		62%	0%
Luz, 2022	Mean Likert scale evaluation* from participants on comfort performing US-guided FNA before and after course	First year residents:
		Before	After
		0.8	6.2
		Second year residents:
		Before	After
		2.2	7.8
Samargandy, 2022	Proportion of participants that felt “not comfortable” with US-guided FNA before and after training	Before	After
		62%	0%
Knowledge
Samargandy, 2022	Mean proficiency** in overall US-guided FNA performance before and after training	Before	After
		2.43	3.93
Davis, 2018	Average score on knowledge test	Pre-test	Post-test
		40.91	61.82
		p-value: 0.05

Abbreviations: US-FNA: Ultrasound-guided Fine Needle Aspiration

^*The Likert scale evaluation is a series of questions with responses ranging from 0 (the worst) to 10 (the best). The participants were asked how safe they felt performing US-guided FNA before and after the course.

^**The Global Proficiency Scale ranges from a score of 1 (not proficient) to a score of 5 (very proficient)

2. Learning Outcomes

Studies evaluated the effect of the training programs on technical skills (e.g., time to perform the biopsy, success rate) and knowledge about the procedure (Tables 4 and 5). Six studies evaluated the acquisition of technical skills post-training with simulator models.^{9, 11, 12, 14, 17, 18} Baba et al. conducted a before-and-after study evaluating a simulation-based training program involving 45 participants who performed ultrasound-guided thyroid FNA on a custom 3D-printed thyroid phantom. The intervention included a brief didactic session on thyroid anatomy and biopsy technique, followed by hands-on puncture practice using the phantom. Major-axis puncture times decreased from 199.6 seconds to 119.4 seconds (p<0.05), while short-axis times decreased from 96.4 seconds to 55.2 seconds (p<0.05).⁹ Luz et al. conducted a before and after study involving 20 radiology residents, including 10 first year (R1) and 10 second year (R2) participants, to evaluate a short training program in ultrasound-guided thyroid FNA using a low-cost gelatin-based model with a grape target.¹⁸ The intervention included a 30-minute online theoretical module and a one-hour in-person practical session. Both groups showed significant reductions in procedure time and the number of superficial punctures post-training. In the R1 group, the number of punctures decreased from 2.2 ± 0.92 to 1.1 ± 0.71, and in the R2 group, from 1.5 ± 0.32 to 1.0 ± 0.0 (p < 0.0001). Although R2 residents performed faster overall, there were no statistically significant differences in pre- or post-training performance between groups, suggesting that the training was equally effective for both levels.¹⁸ Zhang et al. evaluated a training program involving 12 participants, reporting an increase in puncture success rates from 34.4%±14.2% to 66.7%±19.5% post-training (p = 0.003).¹¹ Samargandy et al. assessed proficiency improvements in technical skills among 14 participants, with mean scores increasing from 2.21 to 3.5 for correct ultrasound probe use and from 2.5 to 3.71 for appropriate needle handling (p<0.05).¹⁷ Davis et al. evaluated knowledge improvement in 12 participants, reporting an increase in test scores from 40.91 to 61.82 (p<0.05), with 82% of participants showing measurable improvement. McCrary et al. assessed cognitive knowledge and technical skills in 17 participants, reporting reductions in task completion times and error rates, including correct ultrasound probe identification improving from 53% to 88% (p<0.05).^{12, 14}

3. Transfer of Learning and clinical outcomes of FNA and RFA training

Three studies assessed the translation of acquired skills into clinical practice, demonstrating successful skill transfer from simulation-based environments to real patient care settings (Tables 4 and 5). ^{6, 24, 36} Atri et al. evaluated an FNA training program implemented in a clinical setting involving 4 participants. Before training, the malignancy detection rate was 6%, the unsatisfactory sample rate was 21%, and patient wait times averaged 49 days. After training, malignancy rates increased to 7% (p<.001), unsatisfactory rates declined to 10% (p<.001), and wait times were reduced to 39 days (p<.001), suggesting improved clinical performance post-training. ⁶ Ghai et al. assessed an FNA training intervention conducted in a specialist sonographer setting with 4 participants. Patient wait times declined from 80.7 to 28.3 days (p<.001), and unsatisfactory/nondiagnostic findings were reduced from 25.4% to 21.4% (p=0.015). Additionally, no major procedural complications were reported.³⁶

Li et al. conducted a retrospective study evaluating a stepwise RFA training program for three endocrinologists with no prior procedural experience. The program included over 100 simulated ablations using a food-based model (chicken hearts, pork layers, gel bottles) followed by supervised clinical cases. After training, participants independently treated 23 patients, demonstrating the ability to apply newly acquired skills to real-world patient care.²⁴

Two studies explicitly reported patient-level clinical outcomes following provider training, supporting the long-term impact of training programs on care quality and procedural effectiveness (Tables 4 and 5). ^{15, 24} Gilbert et al. evaluated an FNA training program implemented in a clinical setting with 4 participants. Before training, nondiagnostic biopsy rates were 24%, benign nodules were diagnosed in 48%, and surgical referrals occurred in 21% of cases. After training, nondiagnostic rates dropped to 5%, benign diagnoses rose to 78% (p<0.001), and surgical referrals were eliminated (0%, p<0.001). These outcomes underscore the potential of training to directly improve diagnostic yield and clinical decision-making. ¹⁵ Li et al. reported that after completing the RFA training program, the three trained endocrinologists independently treated 23 patients with benign thyroid nodules. At 12-month follow-up, the median volume reduction rate of treated nodules was 82.0% (IQR: 62.17–87.63%; p<0.05), with minimal complications. These data demonstrate the effectiveness of the training program in enabling clinicians to deliver successful, independent RFA treatment with favorable patient outcomes. ²⁴

Discussion

In this systematic review, we identified 10 studies, including 135 participants, in which nine studies assessed thyroid FNA training programs and 1 study focused on RFA. Educational programs included didactic instruction (N=7), hands-on practice (N=8), clinical assessment under supervision (N=4), and a multi-modal approach (N=7). These training programs were associated with positive learner reactions and improvements in procedural skills and knowledge. Furthermore, a limited number of programs (N=4) reported clinical outcomes and showed an improvement in diagnostic/satisfactory rates of biopsies.

The results from this review support the effectiveness of multi-modal training for thyroid procedures, particularly FNA. The identified programs commonly included combinations of didactic instruction, hands-on learning with simulation models, and supervised clinical practice. This multimodal structure reflects a structured educational framework that allows learners to acquire foundational knowledge, apply it in a simulated environment, and then translate skills into clinical practice. In line with competency-based approaches, several programs emphasized repeated practice, formative assessment, and progression based on skill acquisition rather than time alone. This aligns with modern procedural education frameworks that advocate for staged learning, including observation, simulation, and performance verification, to ensure that trainees achieve proficiency before independent practice. ^{29, 37} Together, these findings highlight the importance of integrating multiple modalities and competency-based milestones into training curricula to enhance both learner confidence and clinical performance in thyroid procedures.^{29, 37}

The included studies evaluated clinical outcomes showcasing potential improvement in nondiagnostic rates. Several reported reduced unsatisfactory samples, improved malignant yield, and shorter patient wait times after training. These findings reflect not only enhanced technical skills but also improved clinical efficiency. Evidence from other procedural disciplines shows that simulation-based, competency-driven training supports long-term skill retention, particularly when paired with deliberate practice and feedback.³⁸ This underscores the value of integrating didactics, simulation, and supervised clinical practice within a structured, competency-based framework to achieve lasting improvements in performance and patient outcomes.

The learning curve for FNA and RFA has been extensively studied, highlighting the importance of structured training and experience to achieve proficiency. For example, Penín et al. evaluated 1,618 participants performing 1,872 FNA biopsies and identified 60 procedures as the threshold for competency, with significant gains in diagnostic accuracy as experience increased.³⁹ Similarly, Russ et al. found that proficiency in RFA for benign thyroid nodules is achieved after 60 cases, with early improvements in technique efficacy and ablation outcomes.⁴⁰ Kuo et al. described distinct learning phases for RFA, estimating procedural mastery at 65 cases and highlighting factors, such as nodule composition and power output, that influence treatment efficacy and the pace of skill acquisition.³⁹ Sinclair et al. emphasized the need for prior expertise in ultrasonography and needle-based procedures to enhance training in thyroid ablation.⁸ These underscore the value of integrating didactics, simulation, and supervised clinical practice within a structured, competency-based framework to achieve lasting improvements in performance and patient outcomes.

From a research perspective, integrating emerging technologies such as artificial intelligence (AI) and extended reality (XR, including virtual, augmented, and mixed reality) holds significant potential to enhance training programs. The adoption of AI in training programs has the potential to standardize education, enhance technical expertise, and ultimately improve patient outcomes. AI-driven medical education and healthcare simulation offer opportunities to enhance trainee competency, particularly in areas that rely on ultrasound, radiology, pathology, and other specialized fields.³⁸ In addition to AI, simulation-based training plays a key role in procedural education by offering a safe, structured environment for repeated practice and feedback. Studies have shown that task-specific simulators improve accuracy, reduce errors, and support long-term skill retention.^41–43 For example, Mascheroni et al. demonstrated that metrics-based simulation training to proficiency significantly outperformed traditional training for cardiac resynchronization therapy (CRT) implantation, with 93% of trainees reaching expert-defined performance benchmarks.⁴⁴ These findings underscore the importance of integrating simulation into competency-based curricula for thyroid procedures, thereby ensuring the acquisition of skills and procedural safety prior to clinical practice. Yet, we did not identify the use of these emerging education tools in the described training programs.

This systematic review was conducted using robust methods, including evaluating multiple data sources, an a priori protocol with predefined inclusion criteria, and independent, duplicate completion of all steps. It provides an overview of contemporary training programs for thyroid biopsy and RFA. In this study, we employed the Kirkpatrick model to assess the impact of training programs for thyroid FNA and RFA. This structured framework provides a comprehensive lens for evaluating the effectiveness of the identified programs. However, the evidence predominantly consisted of small, non-randomized studies, which is common in educational research. Due to the heterogeneity of study outcomes, we were unable to perform a meta-analysis. Additionally, the inclusion of diverse healthcare professional populations, while broadening the scope of the review, may have introduced variability in training program effectiveness that could not be fully accounted for. Moreover, the reliance on self-reported outcomes in some of the included studies may introduce bias and limit the generalizability of the findings. Although some studies reported reduced patient wait times following training interventions, this outcome is influenced by multifactorial system-level factors, including access to care, and may not be directly attributable to training alone. In contrast, outcomes such as adequacy and diagnostic yield provide a more direct measure of the impact of procedural education. Our findings largely apply to thyroid FNA training programs, as only one study addressed RFA. This underscores that only preliminary evidence is available regarding methods and outcomes of RFA training.

Conclusions

This systematic review evaluated evidence on training programs for thyroid biopsy and ablation, highlighting practical components such as didactic instruction, hands-on practice, and clinical supervision. Overall, studies showed improvements in learner confidence, procedural accuracy, knowledge acquisition, and reductions in nondiagnostic or unsatisfactory biopsy rates. Findings highlight the positive effect of training programs for procedural thyroid care and underscore the need for the development and application of novel technologies, such as simulators, that can further support technical skill development.

Availability of Study Materials

The study materials that support the findings of this publication are available from the corresponding author upon reasonable request. Access will be provided for academic, non-commercial purposes, and in compliance with relevant ethical and confidentiality considerations

Highlights.

Teaching Points

• Multi-modal training approaches, including didactics and practical sessions, are predominant components of educational programs for thyroid nodule procedures.

• Training programs enhance learners’ confidence, knowledge, and technical skills in performing thyroid FNA and ablation.

• There is a significant gap in the use of novel technologies—such as simulation and AI-based tools—to enhance training in thyroid nodule procedures, including biopsy and ablation.

Clinical Relevance.

Effective training programs for thyroid FNA and ablation, including multi-modal approaches, can lead to improved confidence and technical skills supporting the care of patients with thyroid nodules.