Training general practitioners in melanoma diagnosis: a scoping review of the literature

Abstract

Background

General practitioners (GPs) play a key role in early melanoma detection. To help GPs deal with suspicious skin lesions, melanoma diagnostic training programmes have been developed. However, it is unclear whether these programmes guarantee the acquisition of skills that will be applied by GPs in their daily clinical practice and maintained over time.

Objectives

This scoping review aimed to examine and compare educational programmes designed to train GPs in melanoma diagnosis using clinical (naked eye) examination alone or dermoscopy±clinical examination, and sought to inform on the long-term sustainability of the GPs’ acquired skills.

Eligibility criteria

Studies eligible for inclusion evaluated educational programmes for teaching diagnosis of melanoma to GPs. MEDLINE, EMBASE and Cochrane databases were searched for relevant articles from 1995 to May 2020.

Results

Forty-five relevant articles were found assessing 31 educational programmes. Most programmes that improved the diagnostic accuracy and long-term performances of the GPs, that is, increase in confidence, decrease in dermatologist referral for benign skin lesions and improvement in the benign/malignant ratio of excised skin lesions, trained the GPs in clinical diagnosis, followed by dermoscopy. To maintain long-term performances, these programmes provided refresher training material.

Conclusion

This review shows that studies generally report positive outcomes from the training of GPs in melanoma diagnosis. However, refresher training material seemed necessary to maintain the acquired skills. The optimal form and ideal frequency for these updates have yet to be defined.

Strengths and limitations of this study.

• Systematic review was conducted following the guidelines of the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews checklist.

• It thoroughly evaluates educational programmes on melanoma diagnosis for general practitioners.

• Specifically, the review examines the long-term effect of educational programmes and the value of providing regular refresher training sessions after the training.

• This review led inevitably to some publication bias as only English-language peer-reviewed articles were included.

Introduction

Early melanoma detection is essential to reduce the morbidity and mortality of patients with melanoma.¹ Given the increased incidence of this aggressive skin cancer, primary care physicians (PCPs) play a key role in early melanoma diagnosis.^2–4 PCPs include a number of healthcare professionals who provide first and continuing medical care to a patient. In this review, we focus on general practitioners (GPs) who take care of patients in community settings and are, in most countries, the first point of contact for any patient with a health issue.

To improve the diagnostic accuracy of melanoma by GPs, specific educational training programmes have been developed. At first, training courses focused on melanoma diagnosis by clinical (naked eye) examination alone. A systematic review,⁵ published in 2011, reported on 20 studies that evaluated 13 educational interventions in clinical melanoma diagnosis for PCPs. All the evaluated interventions improved diagnostic accuracy and melanoma management. Later, educational programmes that included dermoscopy training were created and then evaluated for primary care. To date, dermoscopy has been the most widely used non-invasive in vivo technique in clinical practice to assess skin tumours.⁶ It involves the use of a handheld device that allows the observation of skin structures invisible to the naked eye. However, the sensitivity and specificity of the technique are operator-dependent (trained vs untrained physicians).⁷ Ninety-two percent sensitivity and 95% specificity can be achieved for melanoma diagnosis by a trained dermatologist combining visual inspection and in vivo dermoscopy.⁸ In primary care, dermoscopy has also been shown to be an effective tool for the triage of suspicious pigmented skin lesions when performed by properly trained PCPs.^{9 10} Yet, the minimum training required to reach competence is still unknown.¹¹

Previously published reviews^{5 11–14} on training programmes in melanoma diagnosis for GPs focused on the content, teaching method, outcome measures and study-by-study efficacy of the evaluated educational interventions. However, they did not assess whether the GPs’ acquired skills were measured in the short or long term. Yet, given the increasing burden of melanoma on general practice, it is crucial to know whether these programmes are capable of teaching GPs easily applicable and sustainable skills in melanoma diagnosis and management. This scoping review aimed to explore educational programmes training GPs in melanoma diagnosis using clinical (naked eye) examination alone and diagnosis using dermoscopy±clinical examination. Educational programmes were examined regarding training content, teaching method, training duration, availability of refresher training material and outcome measures. This review also specifically sought to inform on the long-term sustainability of the skills acquired during these training programmes.

Material and methods

To carry out this literature review, a scoping review seemed the most appropriate research method. Indeed, the studies we identified, which provided evidence on the efficacy of educational programmes in melanoma diagnosis for GPs, showed a wide range of study designs and heterogeneous outcome measures. This observation made it impossible to formally assess the quality of these studies and to perform a meta-analysis, leading to a narrative synthesis of our research results. To conduct this scoping review, the framework developed by Arksey and O’Malley,¹⁵ subsequently refined by Levac et al,¹⁶ and the guidelines of the PRISMA-ScR¹⁷ (Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews) checklist were followed.

Eligibility criteria

Studies eligible for inclusion in this review (box 1) evaluated educational programmes teaching either clinical diagnosis of melanoma or diagnosis using dermoscopy±clinical examination and designed primarily for PCPs including GPs. The population of interest included qualified GPs and GP trainees. Specialists and GPs working in hospital settings and/or specialised clinics were excluded. Studies that included training programmes for PCPs other than GPs were not eligible. Studies where no participant training in melanoma diagnosis was proposed and studies evaluating exclusively non-melanoma skin cancer detection were also excluded. Studies evaluating teledermoscopy and computer-aided diagnosis of melanoma were not assessed as they do not require specific education in melanoma recognition by the participants. Only studies assessing the type of educational programme and its short-term and/or long-term efficacy on the skills acquired by the GPs were included. Finally, only peer-reviewed and English-language articles were included.

Box 1. Inclusion and exclusion criteria for article selection.

Inclusion criteria:

Articles

• Articles written in English.

• Study articles and descriptive articles of educational programmes.

Population

• Qualified GPs and GP trainees working in community settings.

Intervention

• Studies evaluating educational programmes in clinical (naked eye) diagnosis and/or diagnosis of melanoma using dermoscopy.

Outcome(s)

• Studies assessing the type of educational programme and its short-term and/or long-term efficacy on the skills acquired by the GPs.

Exclusion criteria:

Articles

• Articles not subject to peer review and written in languages other than English.

Population

• Studies involving specialists, medical students, non-GPs, GPs working in hospital settings and/or specialised skin cancer clinics.

Intervention

• Studies evaluating exclusively non-melanoma skin cancer.

• Teledermoscopy studies.

• Studies on computer-aided diagnosis of melanoma.

Outcome(s)

• No method of measuring outcomes was ruled out.

Abbreviations: GPs, general practitioners.

Data sources and study selection

MEDLINE, EMBASE and Cochrane databases were searched for relevant articles published from 1995 to May 2020. Studies were selected for inclusion independently by three authors (EH, MB and IT), with IT providing the final decision in the event of disagreement. The studies were not assessed for bias, as the risk of bias assessment was reported as not applicable to scoping reviews in the 2018 PRISMA-ScR guidelines.¹⁷

To extensively cover the literature on the subject, four categories of terms were identified: (1) GPs, family doctor, general medicine, family practice and PCPs; (2) education and continuing medical education; (3) melanoma, malignant melanoma, cutaneous melanoma and skin neoplasms; and (4) diagnosis and early detection of cancer. In MEDLINE, the following Medical Subject Headings were used: general practitioners OR family practice OR primary care physicians OR general practice AND melanoma AND diagnosis. No limits were defined. In EMBASE, Emtree terms were exploded: general practitioner, family doctor, primary care, family physician, primary care physician, melanoma, diagnosis and education. In the Cochrane database, the following terms were searched: melanoma AND diagnosis AND general practitioners OR family medicine AND dermoscopy (see online supplemental table 1 for search strategies). In addition, the reference lists of included studies were screened as a source of further relevant articles.

Data extraction

Two authors (EH and MB) reviewed all included articles and independently collected data. Extracted data included authors, year of publication, origin of the article, study design, number of participating GPs, type of educational programme, type of outcome measures and short-term and/or long-term evaluation of these outcomes. The type of educational programme included training content, teaching method, training duration and refresher training material (if provided). To facilitate comparison with data found in previous reviews, all these data were reported into categories adapted from those presented by Fee et al.¹⁴

Table 1 gives the definition of the different categories. The training content was subdivided into six components: epidemiology, clinical diagnosis, clinical algorithm, dermoscopic diagnosis, dermoscopic algorithm and management. The teaching method was considered as live, in the form of scientific literature, e-learning or self-assessment. The refresher training material specified the material available for participants to refresh their skills after the training. The outcome measures were expressed either in terms of competence or in terms of performance, according to the assessment approach of continuing medical education programmes proposed by Moore.¹⁸ Finally, since the limits between short-term and long-term evaluation of a medical educational programme are not standardised, arbitrary limits have been chosen based on the observations made during this literature review.

Table 1.

Definition of study categories

Criteria	Categories	Definition
Training content	Epidemiology	Background information on rates of melanoma cancer, risk factors, localisation and evolution of melanomas
	Clinical diagnosis	Naked eye melanoma recognition
	Clinical algorithm	Use of a pre-existing algorithm as a learning tool to aid for clinical diagnosis
	Dermoscopic diagnosis	Recognition of melanoma using dermoscopy
	Dermoscopic algorithm	Use of an algorithm as a learning tool to aid for dermoscopic diagnosis
	Management	Determination of a plan of action for a skin lesion, that is, reassurance, follow-up or lesion excision
Teaching method	Live	Presentation by a speaker to a group of participants
	Scientific literature	Use of educational books, posters, letters, CD-ROMs or videos
	E-learning	Interactive online tutorials including audio and visual information
	Self-assessment	Learning by the participant himself using educational material
Refresher training material	Teledermatology feedback	Feedback from a dermatologist on the image and clinical history of a suspicious lesion at a distance, using remote internet-based technologies
Outcome measures	Competences	Acquired skills, which are evaluated in a training setting on clinical and/or dermoscopic photographs of skin lesions
	Diagnostic accuracy	Ability of the participants to discriminate between melanoma and benign lesions
	Knowledge	Report of conceptual understanding
	Appropriate management	Determination of the right plan of action for a skin lesion
	Performances	Changes in real-life practice measured in a clinical setting, that is, changes in the benign/malignant ratio of excised lesions, the number of total-body skin examinations performed, confidence of the general practitioners, changes in referral rates to a dermatologist and decrease in the incidence of advanced melanomas
Evaluation	Short-term	Measurement of outcomes immediately or up to 3 months after the training
	Long-term	Measurement of outcomes at ≥6 months after the training

Patient and public involvement

No patient and public involvement was required for this review.

Results

In total, 325 articles were identified from the electronic database searches, as shown in the PRISMA flowchart (figure 1).¹⁹ At the end of the study selection process, 45 relevant articles, which assessed 31 educational interventions, were included in the review analysis.

Figure 1.

Preferred Reporting Items for Systematic reviews and Meta-Analyses flowchart of the study selection process.¹⁷ GP, general practitioner.

Study designs

Thirty-six interventional studies with a range of study designs were found: 11 randomised controlled trials (RCTs),9 10 20–29 19 diagnostic accuracy studies,^30–48 3 cohort studies^49–51 and 3 case–control studies.^52–54 Five of the 31 training programmes were assessed twice.25 26 28 32 49

Four systematic reviews were identified: one on the training of PCPs in clinical melanoma diagnosis,⁵ two on the training of PCPs in dermoscopy for melanoma diagnosis^{12 13} and one on the use of dermoscopy in primary care.¹¹ A scoping review on the training of PCPs in dermoscopy¹⁴ was also included. The final three articles were descriptive articles of the educational programmes and study protocols.^55–57

Educational programmes

The educational programmes in melanoma diagnosis for GPs varied in terms of content, teaching method and outcome measures. The characteristics of these training programmes are summarised in table 2.

Table 2.

Characteristics of educational programmes in melanoma diagnosis for GPs

Article	Study design	Study participants	Training content						Teaching method	Training duration	Refresher raining material
Author, year, locationref.			Epidemiology	Clinical diagnosis	Clinical algorithm	Dermoscopic diagnosis	Dermoscopic algorithm	Management
Marra et al, 2020, The Netherlands53	Case–control study	185 (83; 102)*		X		X (optional)		X	E-learning	2 hours
Sawyers et al, 2020, Canada48	DA study	33 GPs				X	TADA step I		Live	3.5 hours
Augustsson et al, 2019, Sweden54	Case–control study	43 GPs (27;16)*				X	Pattern analysis		Live	1 day	PDF files of the course
Seiverling, 2019, USA47	DA study	59 GPs		X		X	TADA		Live	75 min
Beecher et al, 2018, Ireland46	DA study	23 GP trainees	X	X				X	Live, literature	1 hour
Secker et al, 2017, The Netherlands45	DA study	293 PCPs including ? GPs		X		X		X	Live, literature and e-learning	1 day
Rogers et al, 2016, USA44	DA study	16 GPs				X	TADA		Live	1 day
Badertscher et al, 2011, 2015, Switzerland29 56	RCT	78 GPs (39;39)*		X		Lumio			Live	1 day	Teledermatology feedback
Gulati et al, 2015, UK51	Cohort study	967 GPs	X	X				X	E-learning	PD
Koelink et al, 2014, The Netherlands10	RCT	53 GPs	X	X		X	Seven-point checklist		Live	4-hour clinical diagnosis; 6-hour dermoscopy
Grange et al, 2013, France50	Cohort study	398 GPs	X	X					Live, Literature	2.5 hours	CD-ROM+regular information sheets
Markova et al, 2013, USA28	RCT	46 GPs (20;26)*			ABCD(E)				E-learning
Mikkilineni et al, 2001 and 2002, USA42 43 (Weinstock et al, 199657)	DA study	7 GPs	X	X				X	Live Literature	2 hours
Eide et al, 201341  (Shaikh et al, 201255)  INFORMED	DA study	54 PCPs including 9 GPs (20%)	X	X	ABCD(E)+ugly duckling sign	X	Three-point checklist	X	E-learning, self-assessment	2-hour web-based learning	Unlimited e-learning access
Bourne et al, 2012, Australia40	DA study	3 GPs		X		X	BLINCK Three-point checklist Menzies method		Live, literature	UD
Shariff et al, 2010, UK39	DA study	94 GPs		X					Literature	PD
Grimaldi et al, 2009, Italy38	DA study	13 GPs		X		X	ABCD rule		Live, e-learning, sself-assessment	PD	Self-assessment, e-learning, training sessions
Menzies et al, 2009, Australia37	DA study	63 GPs		X		Yes+sequential digital dermoscopy	Menzies method		Live, literature, e-learning	2-hour workshop	Unlimited e-learning access
Peuvrel et al, 2009, France36	DA study	210 GPs	X	X	ABCD(E)			X	Live, literature (CD-ROM)	2 hours	CD-ROM
Youl et al, 2007, Australia49	Cohort study	16 GPs									Self-assessment, paper-based training sessions
Raasch et al, 2000, Australia23	RCT	46 GPs (23;23)*	X	X	ABCD(E)	X		X	Literature	PD
Argenziano et al, 2006, Italy and Spain9	RCT	73 GPs (36;37)*		X	ABCD	X	Three-point checklist		Live	1 day
Dolianitis et al, 2005, Australia35	DA study	35 GPs				X	Menzies method Seven-point checklist ABCD rule Pattern analysis		Literature, e-learning, self-assessment	PD
Carli et al, 2005, Italy34	DA study	41 GPs	X	X	ABCD(E)			X	Live	4 hours
de Gannes et al, 2004, Canada27	RCT	27 GPs (10;17)*	X	X					Literature (video format)	12 min	Unlimited access to the 12 min video
English et al, 2003, Australia26	RCT	468 GPs (245;228)*		X+polaroid instant camera				X	Literature	>6 hours
Del Mar and Green,1995, Australia20	RCT	93 GPs (48;45)*								1 hour
Brochez et al, 2001, Belgium33	DA study	146 GPs	X	X				X	Live, literature	2 hours
Harris et al, 1999 and 2001, USA30 32	DA study	232 GPs	X	X	ABCD Seven-point Glasgow checklist			X	Literature, e-learning	1 hour
		17 GPs
Westerhoff et al, Australia 200024	RCT	74 GPs				X	Menzies method		Live, literature	1-hour live
Bedlow et al, 2000, UK31	DA study	17 GPs		X					Live, literature	UD
Gerbert et al, 1998 and 2002, USA22 25	RCT	52 GPs (26;26)*	X	X				X	Live, literature, e-learning	>3 hours
Dolan et al, 1997, USA21	RCT	82 PCPs, including 16 GPs (46;36)*	X	X					Live	2 hours
Girgis et al, 1995, Australia52	Case–control study	41 GPs (24;17)*	X	X				X	Live, literature	>6 hours

*Number of participants in the intervention group and control group.

ABCD, Asymmetry, uneven Borders, uneven Colours and Diameter >6 mm; ADCD(E), Asymmetry, uneven Borders, uneven Colours, Diameter >6 mm and Evolution; DA, diagnostic accuracy; GPs, general practitioners; INFORMED, Internet curriculum for melanoma early detection; PCPs, primary care physicians; PD, participant-dependent; RCT, randomised controlled trial; TADA, triage amalgamated dermoscopic algorithm; UD, undisclosed.

Training content

Of the 31 educational programmes, 15 involved the training of GPs in clinical diagnosis, 5 involved dermoscopic diagnosis alone and 11 involved the training of GPs in both of these melanoma diagnostic methods. Twelve (80%) of the clinical diagnostic training programmes also involved learning of epidemiology and 11 (73%) involved learning of management guidelines for suspicious lesions. Only seven (47%) programmes teaching clinical diagnosis used an algorithm to teach melanoma recognition, with the ABCD(E) rule⁵⁸ (Asymmetry, uneven Borders, uneven Colours, Diameter >6 mm and Evolution) being most commonly taught. Of the dermoscopic training programmes, 12 (80%) included learning of at least one dermoscopic algorithm (Menzies method,24 35 37 40 three-point checklist,^{9 40 41} seven-point checklist,^{10 35} triage amalgamated dermoscopic algorithm (TADA),^{44 47} the ABCD rule,^{35 38} BLINCK⁴⁰ and pattern analysis^{35 54}). The Menzies method,⁵⁹ the ABCD rule,⁶⁰ pattern analysis and seven-point checklist⁶¹ were designed originally for trained physicians and were later tested as effective when used by non-experts.⁶² Other algorithms, such as the three-point checklist⁶³ and the TADA,⁴⁴ were initially created for use by PCPs. In addition, two educational programmes included training on other diagnostic tools, such as sequential digital dermoscopy imaging³⁷ and polaroid instant camera photography.²⁶

Teaching method, training duration and refresher training material

Live training courses and the use of educational books, posters or videos (literature) were the two preferred teaching methods of clinical diagnostic training programmes. Five training programmes also used an e-learning approach.25 28 30 51 53 The most common teaching method used in dermoscopic training programmes was live training. This approach was combined with literature and/or e-learning in six programmes. Three programmes also used self-assessment. Overall, the teaching method did not appear to have influenced the programme outcomes. Duration of training varied from 75 min to 1 day, was not specified in two studies^{31 40} and was participant-dependent in six studies using self-assessment methods. Six dermoscopic diagnostic training programs29 37 38 41 49 54 and three programmes in clinical diagnosis^{27 36 50} provided regular refresher training material such as unlimited e-learning access or self-assessment training sessions.

Training outcomes

Table 3 summarises the outcome measures of the studies. In the selected studies, the GPs’ competences were generally measured in the short term and their performances measured in the long term after the training.

Table 3.

Outcome measures of the melanoma diagnostic educational programmes for general practitioners

(A) Studies with short-term outcomes

Studies

Measured competence in training setting

Measured performance in clinical setting

Timing after the training

(First author and year of publication)

DA

Knowledge

AM

B/M ratio

TBSE

DA

Confidence

Decrease in VTM incidence

RR

Immediately

At 1 month

At 3 months

Dermoscopic diagnostic training programmes

Sawyers (2020)48

+

X

Seiverling (2019)47

+

X

Secker (2017)45

+

+

X

Rogers (2016)44

+

X

Bourne (2012)40

+

X

Dolianitis (2005)35

+

X

Westerhoff (2000)24

+

+

X

Clinical diagnostic training programmes

Beecher (2018)46

+

+

+

X

X

Carli (2005)34

+

+

X

Mikkilineni (2001, 2002)42 43

+

+

+

+

X

Brochez (2001)33

+

X

Harris (2001)32

+

+

+

+

X

Harris (1999)30

+

–

+

+

X

Raasch (2000)23

–

+

3-month period

Bedlow (2000)31

+

+

X

Gerbert (1998, 2002)22 25

+

+

+

+

X

Dolan (1997)21

–

+

–

X

X

Girgis (1995)52

–

+

+

X

X

(B) Studies with long-term outcomes

Studies

Measured competence in training setting

Measured performance in clinical setting

Timing after the training

DA

Knowledge

AM

B/M ratio

TBSE

DA

Confidence

Decrease in VTM incidence

RR

Dermoscopic diagnostic training programmes

Shariff (2010)39

–

At 11 months

Youl (2007)49

+

+

6-month period

Argenziano (2006)9

+

+

16 -month period

Clinical diagnostic training programmes

Gulati (2015)51

–

+

8-month period

Peuvrel (2009)36

+

+

15-month period

de Gannes (2004)27

–

–

–

At 6 months

English (2003)26

–

21-month period

Del Mar (1995)20

+

24-month period

(C) Studies with short-term and long-term outcomes

Studies

Measured competence in training setting

Timing after the training

Measured performance in clinical setting

Timing after the training

DA

Knowledge

AM

Immediately after

At 1–3 months

B/M ratio

TBSE

DA

Confidence

Decrease in VTM incidence

RR

Dermoscopic diagnostic training programmes

Augustsson (2019)54

+

X

At 6 months (here competence measure)

Badertscher (2015)29

+

X

–

+

At 12 months

Koelink (2014)10

+

+

X

+

+

At 8 months

At 12 months

At 19 months

Eide (2013)41

+

+

+

X

+

+

+

At 6 months

Youl (2007)49

+

+

6-month period

Raasch (2000)23

–

3-month period

Grimaldi (2009)38

+

X

X

+

6-month period

Menzies (2009)37

+

+

X

+

+

6-month period

Clinical diagnostic training programmes

Marra (2020)53

+

X

+

+

+

At 10 months

Grange (2013)50

+

+

X

+

3-year period

Markova (2013)28

–

At 12 months

Mikkilineni (2001, 2002)42 43

+

+

X

To note that Menzies et al³⁷ and Carli et al³⁴ provided only descriptive statistics of their study results. For the other studies, a p value <0.05 was considered significant, except for Peuvrel et al³⁶ (p<0.001).

+=significant improvement; –=no significant improvement.

AM, appropriate management; B/M, benign/malignant; DA, diagnostic accuracy; RR, decrease in dermatologist referral rates for benign lesions and increase in referral rates for malignant lesions; TBSE, total-body skin examination; VTM, very thick melanomas.

Eight clinical diagnostic training programmes and seven dermoscopic training programmes only assessed the short-term efficacy of their programme (table 3A). For these studies, the competences most often evaluated were diagnostic accuracy and appropriate management measured in a training setting. The most evaluated short-term performance, measured in a clinical setting, was the GPs’ confidence in diagnosing melanoma. With the exception of two studies, all showed a positive impact of their intervention.^{21 23} Four clinical diagnostic training programmes and three dermoscopic training programmes (one teaching dermoscopy alone⁵⁴) only assessed long-term performances (table 3B). The most evaluated long-term performances, measured in daily clinical practice, were the GPs’ diagnostic accuracy and the benign/malignant ratio of excised lesions. Three studies^{9 36 49} reported improvement in the GPs’ performances in melanoma diagnosis. The other studies reported no improvement.

Finally, three clinical diagnostic and seven dermoscopic training programmes assessed the short-term and long-term outcome of their training (table 3C). Except for one,²³ all these training programmes demonstrated improvement in the GPs’ competences, measured in a training setting in the short term. In the long term, eight training programs10 37 38 41 49 50 53 54 reported significant improvement in the GPs’ performances for the diagnosis of melanoma and benign skin lesions. This led to either a decrease in the referral rates to dermatologists^{37 41 53} or a decrease in the ratio of benign/malignant excised skin lesions.^{37 49} Among the major studies, Koelink et al¹⁰ found that their dermoscopic training programme improved the GPs’ long-term performances with up to 1.25 times greater diagnostic accuracy for skin lesions including melanomas. In a French department, Grange et al⁵⁰ observed an impressive reduction in the incidence of advanced melanomas (Breslow thickness ≥3 mm) during the 3-year period after their GP training campaign in clinical melanoma diagnosis. A very recent study by Marra et al⁵³ assessing 1662 referrals reported better quality of referrals by GPs trained in melanoma diagnosis than by untrained GPs, potentially leading to less unnecessary referrals. However, two educational programs^{28 29} were unable to maintain the GPs’ acquired performances in the long term.

Discussion

This scoping review aimed to explore educational programmes training GPs in melanoma diagnosis using clinical (naked eye) examination alone and diagnosis using dermoscopy±clinical examination. Educational programmes were examined regarding training content, teaching method, training duration, availability of refresher training material and outcome measures. This review also specifically sought to inform on the long-term sustainability of the skills acquired during these training programmes.

Types of educational programmes with positive long-term outcomes

Most reported educational programmes that improved long-term diagnostic accuracy and changed GPs’ melanoma practice patterns trained their participating GPs in dermoscopy combined with clinical diagnosis. This teaching method is supported by a recent Cochrane review⁸ in which dermoscopy alone was found to be less accurate than clinical examination followed by dermoscopy for the diagnosis of melanoma. The Cochrane review results also suggested that dermoscopic algorithms were the most useful method to train non-experts in dermoscopy. In our review, we found that five of the programmes with long-term positive impact used dermoscopic algorithms to teach melanoma diagnosis.10 37 38 41 54

Unfortunately, the substantial number of training hours necessary to become competent in dermoscopy is the main reported factor limiting its use in general practice.^{64 65} At this time, there is no evidence on the optimal length of training, even though it has been demonstrated that diagnostic accuracy of dermoscopy depends on the degree of training of the practitioner.⁶⁶ One study suggested that 1 day of live training in dermoscopy was sufficient to build the confidence of GPs with special interest in melanoma diagnosis.⁶⁷ We found two RCTs that demonstrated sustained improvement in GPs’ diagnostic accuracy, both of which proposed live training in dermoscopy over 1 day or 10 hours.^{9 10} Because the duration of training in dermoscopy is a limiting factor for most GPs, it is important to keep in mind that training in clinical melanoma diagnosis has also been shown to improve the GPs’ performances while requiring less training time (in this review, a mean duration of 2.5 hours was observed for clinical diagnostic training programmes).^{36 50}

However, we found that educational programmes teaching dermoscopy have been more likely to assure positive long-term outcomes than programmes teaching clinical examination alone. One of the reasons could be that the latter used measures of performance such as GPs’ confidence level and number of total-body skin examinations performed before and after training, which did not reflect GPs’ diagnostic ability in clinical practice. On the one hand, measuring the confidence of GPs in their own ability to diagnose melanoma is more useful in assessing the quality of a training programme than evaluating skills acquired by participants. From a pedagogical point of view, participants feel more confident when they know how to use the teaching content in daily practice, but this does not define their true diagnostic competence.¹⁸ On the other hand, the number of total-body skin examinations performed may be useful in measuring GPs’ awareness of skin tumours, but not for evaluation of GPs’ diagnostic skills.

Long-term improvement of the GPs’ performances in clinical settings

The GPs’ long-term performances measured in clinical settings were assessed for 15 educational programmes: 6 in clinical diagnosis and 9 in dermoscopic diagnosis. Ten showed a positive impact on the GPs’ performances measured over periods ranging from 6 to 19 months. The most frequent observations were a decrease in referral rates to dermatologists for benign skin lesions and an improvement in the benign/malignant ratio of excised skin lesions. The internet curriculum for melanoma early detection (INFORMED) group⁴¹ reported an increase in melanoma diagnosis during a screening campaign by GPs trained with their programme in 2016.⁶⁸ Furthermore, a decrease in the incidence of advanced melanomas was shown in a French department over a 3-year period after their training programme in clinical melanoma diagnosis.⁵⁰ Unfortunately, two educational programs^{28 29} failed to maintain the GPs’ acquired performances at 1 year after the end of the training. The reasons might be that Markova et al²⁸ chose to assess the number of total-body skin examinations performed but did not evaluate the GPs’ diagnostic accuracy and that Badertscher et al²⁹ trained their GPs to use Lumio, a polarised magnifying glass with 2× magnification instead of a standard dermoscopy device (10× magnification).

To retain acquired diagnostic skills over the long term, results of a recent RCT suggested the need for ‘refresher sessions at regular intervals’.^{69 70} In our review, nine (60%) educational programmes evaluated in the long term provided refresher training material. Seven of these programmes were successful. In 2014, Grange et al⁵⁰ produced a CD-ROM containing their teaching material and sent regular information about melanoma to the GPs. The INFORMED group⁴¹ provided GPs with a free unlimited access to their web-based course. Menzies et al³⁷ gave participating GPs an educational textbook and an unlimited e-learning access. Grimaldi et al³⁸ and Youl et al⁴⁹ also ensured self-assessment refresher training sessions. Marra et al⁵³ found a 10-month sustainability of the diagnostic accuracy of their trained GPs and assumed that daily use of the obtained knowledge during the study period achieved this effect. Only Koelink et al¹⁰ who evaluated the longest post-training period (19 months) and who demonstrated sustainability of diagnostic skills did not specify whether update training modalities were provided.

However, the ideal frequency and form of updates have never been studied. An RCT, taking place in the English National Defibrillator Programme, determined that update session intervals after a medical education session should not exceed 7 months to limit the loss of acquired skills and maintain the participants’ confidence.⁷¹ In the UK, a survey among GPs with special interest in dermatology stated that self-assessment learning was the most popular for refresher sessions.⁶⁷ Nevertheless, they also showed that 36% of GPs who use dermoscopy in their clinical practice reported to have never updated their training skills. We found that the most appreciated form of self-assessment updates was the unlimited access to an e-learning course. In the future, this enthusiasm for online training could lead to the development of smartphone applications to train GPs in melanoma diagnosis. Some newly developed applications have currently been evaluated among medical students⁷² and dermatology residents.⁷³ Initial results already looked very promising.

Finally, the variability of refresher training material provided in the educational programmes and the heterogeneity of outcome measures did not allow more robust conclusions to be drawn on the most beneficial training modality for sustainable improvements in GPs’ diagnostic skills.

Limitations

This scoping review has some limitations and led inevitably to certain publication biases. We used keywords for the selection of articles, and only peer-reviewed articles were included. By limiting our research to English-language articles, some studies may also have been missed. It is also very likely that melanoma diagnostic training programmes exist in unpublished forms, for example, in university continuing medical education programmes. Moreover, we focused only on studies assessing melanoma diagnostic training methods among GPs. Therefore, we may have failed to mention some educational programmes for primary care in this review. Furthermore, educational programmes which were not evaluated in studies were overlooked. Finally, this review covers educational programmes over a 25-year period. As technology has evolved considerably over this time, some teaching methods and refresher training materials have been overshadowed by interactive online tutorials (e-learning)—all the more so with the health crisis caused by COVID-19 during which distance learning methods have developed very rapidly.

Conclusion

In conclusion, educational programmes trained GPs in melanoma diagnosis using clinical examination alone or dermoscopy±clinical diagnosis. Most reported programmes that improved the long-term diagnostic accuracy and changed routine performances of the GPs (ie, decrease in dermatologist referral of benign skin lesions and improvement in the benign/malignant ratio of excised skin lesions) trained their participating GPs in both diagnostic methods. The preferred teaching methods were live and e-learning, but the teaching method did not seem to influence the GPs’ acquired performances. It is important to note that the educational programmes that achieved long-term sustainability of GPs’ performances in daily clinical practice provided refresher training material. However, no conclusions on the most beneficial training modality to sustainably improve GPs’ diagnostic skills could be drawn given the heterogeneity of outcome measures and study designs. Therefore, the optimal form and ideal frequency of these updates have yet to be defined.