Efficacy of a general practitioner training campaign for early detection of melanoma in France

Authors


  • Funding sources This study was supported by grants from the Ligue Contre le Cancer, the Groupement Régional de Santé Publique de Champagne-Ardenne and the Société Française de Dermatologie.
  • Conflicts of interest None declared.
  • Plain language summary available online

Summary

Background

To date, no strategy for improving early diagnosis of melanoma has been evaluated on a population basis in France.

Objective

To evaluate the efficacy of a general practitioner (GP) awareness and training campaign in a pilot French geographical region (Champagne-Ardenne), including 1·34 million inhabitants, 1241 GPs, 56 dermatologists and a population-based melanoma registry.

Methods

All GPs received repeated awareness postal mailings in 2008 and 398 (32·1%) attended training sessions organized by 27 dermatologists. The pre- (2005–7) and post-campaign (2009–11) periods were compared for the following: primary endpoint – the world-standardized incidence of very thick melanomas (VTM) (Breslow thickness ≥ 3 mm); secondary endpoints – the mean Breslow thickness; the proportions of VTM and of thin (< 1 mm) melanomas among invasive cases; and the ratio of in situ/all melanoma cases. Similar measures were performed in the control area of Doubs/Belfort territory (655 000 ha), where no similar campaign was carried out.

Results

The incidence of VTM decreased from 1·07 to 0·71 per 100 000 habitants per year (= 0·01), the mean Breslow thickness from 1·95 to 1·68 mm (= 0·06) and the proportion of VTM from 19·2% to 12·8% (= 0·01). The proportion of thin and in situ melanomas increased from 50·9% to 57·4% (= 0·05) and from 20·1% to 28·2% (= 0·001), respectively. No significant variation was observed in Doubs/Belfort territory.

Conclusion

These results strongly support the efficacy of such a campaign targeting GPs and provide a rationale for a larger public health campaign in France, including training of GPs by dermatologists and encouraging patients to ask their GP for a systematic skin examination.

Melanoma incidence and mortality rates have increased sharply in the past three decades in most developed countries.[1-4] Despite recent advances in the treatment of metastatic diseases,[5-9] early diagnosis and decreasing incidence of thick tumours are the best ways to reduce mortality.[10] Although melanoma, of the five major detectable cancers (breast, colorectal, prostate, cervical cancer and melanoma), is the only one that can be first detected by a simple visual, noninvasive examination, to date systematic screening in the general population has not been recommended in most countries, including France (www.has-sante.fr).[11, 12] Indeed, recent studies have concluded that evidence that systematic screening may improve health outcomes is lacking. Recently, a pilot systematic screening study in Schleswig-Holstein (Germany) suggested for the first time that skin cancer screening may reduce melanoma mortality,[13] and a large screening programme targeting 45 million individuals is under way in Germany.[14]

Although they have not come out in support of systematic, national screening, the French health authorities emphasized in 2005 that there was strong evidence to ‘highlight the usefulness of early diagnosis within the public and health professionals’. A possible role for general practitioners (GPs) in an early diagnosis strategy has been considered and national French recommendations for GPs have been published (www.has-sante.fr). However, to date, no systematic training programme for GPs has been set up and no strategy for improving early diagnosis has been developed or evaluated on a population basis in France.

In a preliminary study in the Haut-Rhin département of France, we demonstrated that a systematic training programme for GPs in a limited geographical area could significantly improve awareness and the ability of participants to diagnose skin tumours.[15] In more recent population-based studies in six regions of France, we found that the percentage of melanomas diagnosed in a GP setting had increased from 26% in 2004 to 42% in 2008,[12, 16] but that GPs still frequently detected very thick melanomas (VTM).[16, 17] In addition, thinner melanomas were detected by GPs who were used to performing active detection and had been specifically trained.[16, 18]

This prompted us to organize and evaluate a GP training campaign for the early detection of melanoma in one of these six regions (i.e. Champagne-Ardenne). This region was chosen to enhance the feasibility of this pilot campaign, because it had previously been demonstrated that the proportion of thick melanomas was larger in this region than in the other five regions of north-eastern France,[19] and because a regional melanoma registry was available for evaluation.[19]

The primary objective was to obtain a decrease of the incidence of VTM, defined as melanoma with a Breslow thickness ≥ 3 mm, as previously reported.[17] The secondary objectives were a decrease in the proportion of VTM, a decrease in the mean Breslow thickness and an increase in the proportions of thin melanomas (i.e. melanoma with a Breslow thickness < 1 mm) and in situ melanomas.

Material and methods

The regional project

The study was approved by the institutional review board of Reims University Hospital. It was launched in 2007 in the French north-eastern region of Champagne-Ardenne, which groups together four administrative départements (Ardennes, Aube, Marne and Haute-Marne), covers 32 975 km2 and includes 1 338 850 inhabitants according to the 2006 census. It also included at the time of the study 1241 GPs with clinical activity and 56 dermatologists. All GPs and dermatologists in the study area were first contacted by mail and subsequently informed of the objectives and implementation of the campaign directly by mail and indirectly through all the associations for continuing medical education registered in the study area. Training sessions were organized throughout the entire region and GPs were invited to participate as auditors in one of them. Dermatologists were invited to participate as teaching experts in the same sessions, after special training in the use of specific educational media.

Training sessions and immediate evaluation

Thirty-one training sessions involving 27 trainer dermatologists (generally two for each session) were conducted in the study area between December 2007 and January 2009, with most taking place in 2008. These standardized two and a half hour awareness and training courses included five parts. The first was an initial evaluation of knowledge (pretest). It consisted of 12 short questions, including three theoretical questions on risk factors for melanoma, the ABCD rule and the prognostic value of Breslow thickness; and nine clinical pictures, including one seborrhoeic keratosis, one pigmented basal cell carcinoma, three benign naevi and four melanomas. For each clinical picture, participants had to choose one of the following answers: benign cutaneous lesion, carcinoma or melanoma. The second part focused on epidemiological data regarding melanoma in France and the Champagne-Ardenne region, in order to increase the global awareness of GPs and to emphasize the importance of early diagnosis and the potential key role of GPs. The third part focused on risk factors of melanoma and recognition of at-risk subjects in clinical practice. The fourth part focused on clinical positive and differential diagnosis of melanoma. For this part, a total of 48 clinical pictures were shown, either separately, or as part of six clinical case reports derived from real-life situations in daily general practice (e.g. a 45-year-old fair-skinned man, consulting for a certificate of fitness for sport, with numerous naevi and a melanoma in his back, or a 55-year-old black man consulting for suspicion of onychomycosis, with a large melanonychia with periungual spread and nail dystrophy, consistent with melanoma). The last part of the session consisted of a final evaluation, or post-test, using the same 12 questions as for the pretest.

At the end of the training session, participants received a CD-ROM containing all the session material and a poster promoting early melanoma detection, to be put up in their waiting rooms. The session material was also made available on a dedicated website.

Three questionnaires were collected from participating GPs during the training sessions: the individual answers to the pretest, the individual answers to the post-test, and a global evaluation of the session by the GP. Questions included in this last questionnaire are shown in Table 1.

Table 1. Immediate global evaluation of training sessions by participating general practitioners (GPs)
 Yes (%)No (%)No response (%)
Do you think that this session has improved your ability to recognize at-risk subjects for melanoma?98·71·30·0
Do you think that this session has improved your ability to distinguish melanoma from benign lesions?69 (Yes, very much)29·6 (Yes, moderately)0·50·8
Do you think the active participation of GPs could improve early diagnosis of melanoma?94·92·23
Would you like to receive additional information on melanoma on a regular basis?87·310·52·2

Additional action targeting nonparticipating general practitioners

In addition to training sessions, regular information about melanoma was sent to all GPs in the study area by mail, whether they participated in a training session or not. This consisted of five issues of a quarterly, one-sheet publication with a short, informative text and related colour photographs on the following themes: early diagnosis of melanoma; primary prevention in children; at-risk subjects for VTM (targeting in particular men, the elderly and patients living alone);[17] early diagnosis and excision of lentigo malignant melanoma; and diagnosis and management of rapidly growing melanoma.[20, 21]

Evolution of incidence and Breslow thickness in the Champagne-Ardenne region and control area

The primary endpoint for the evaluation of efficacy was defined a priori as the incidence of VTM in the Champagne-Ardenne region. VTM was defined as melanoma with a Breslow thickness ≥ 3 mm, as previously reported in different studies,[17, 22] including a recent one in north-eastern France comprising both the target and control areas of the present study.[17] The incidence rates were evaluated using data from the regional melanoma registry of Champagne-Ardenne (Observatoire du Mélanome en Champagne-Ardenne),[19] and compared between the 3-year period preceding the campaign (2005–7) and the 3-year period following the campaign (2009–11). The secondary endpoints were the evolution of the mean Breslow thickness, the proportion of VTM among incident invasive cases, the proportion of thin melanomas (i.e. melanoma with a Breslow thickness < 1 mm) among incident invasive cases and the ratio of in situ/all melanoma cases.

Epidemiological data from the départements of Doubs and Belfort territory were used for comparison with the Champagne-Ardenne region. These two administrative départements include 655 000 inhabitants. This geographical area was selected as a control area because it is located in the same north-eastern part of France but does not share a common border with Champagne-Ardenne, has similar sociodemographic characteristics, and has a tumour registry that provides regular estimations of melanoma incidence. It is noteworthy that no secondary prevention campaign was performed in this control area during the study period. As the registry of Doubs and Belfort territory is a general tumour registry collecting data on all types of cancer, delays for validation of incident melanoma cases were longer than in the Champagne-Ardenne region. Therefore, data in this control area were only available up until 2010 at the time of the analysis. However, given that no training campaign was conducted in this control area, the periods of 2005–7 and 2008–10 could be used for comparison of two 3-year periods. A comparison excluding the year 2008 similarly to the Champagne-Ardenne region was also performed.

Statistical analysis

Quantitative variables are described as mean ± SD and qualitative data as number and percentage. A Student paired t-test was used to determine if the mean differences between pre- and post-test scores (global score, theoretical questions and clinical pictures) significantly differed from zero. For these analyses, data from individual GPs could only be used if they had completed both the pre- and post-tests. World standardized incidence rates of invasive melanoma were estimated by direct standardization using the distribution by age categories of the world population. These rates were calculated for all invasive melanomas and for VTM in the whole population and by sex. The world standardized incidence rates were compared using the Mantel–Haenszel test. For all analyses, a P-value < 0·05 was considered statistically significant. All analyses were performed using SAS version 9·3 (SAS Inc., Cary, NC, U.S.A.).

Results

Immediate evaluation

In the whole study area, 1241 GPs (100%) received the information letters and 398 (32·1%) attended a training session. Pre-and post-test scores were available for 364 of 398 (91·5%) participating GPs. The Student paired t-test showed that the mean differences significantly differed from zero concerning the three scores (2·8 ± 2·3 for the global score, P < 0·0001; 1·6 ± 0·9 for the theoretical questions, < 0·0001; and 1·2 ± 1 for the clinical pictures, < 0·0001).

Table 1 shows the GPs' satisfaction rate and answers regarding the global evaluation of the session and their self-assessed improvement.

Evolution of incident melanomas in the Champagne-Ardenne region before and after the campaign

Figure 1 shows the annual number of incident invasive melanomas from 2005 to 2011, according to Breslow thickness in the intervention area. Table 2 shows the distribution of in situ and invasive melanomas according to Breslow thickness and period of diagnosis. Table 3 gives the world standardized annual incident rates of invasive melanomas according to thickness and period of diagnosis.

Table 2. Distribution of incident cases of melanoma in the Champagne-Ardenne region according to period of diagnosis and Breslow thickness
Incident cases of melanomaFirst period (2005–7), n (%)Post-campaign period (2009–11), n (%)P-value
Melanoma536 (100)621 (100)0·001
In situ 108 (20·1)175 (28·2)
Invasive428 (79·9)446 (71·8)
Invasive melanoma (mm)  0·03
< 1218 (50·9)256 (57·4)0·05
1–2·99128 (29·9)133 (29·8)0·98
≥ 382 (19·2)57 (12·8)0·01
Table 3. Annual incident rates of melanoma in the Champagne-Ardenne region according to period of diagnosis and Breslow thickness
CasesFirst period (2005–7)aPost-campaign period (2009–11)aP-value
  1. a

    World standardized incidence rates, per 100 000 inhabitants per year.

All invasive melanomas (mm)6·726·870·88
< 13·704·260·13
1–2·991·961·900·94
≥ 31·070·710·01
Figure 1.

Annual number of invasive melanoma according to Breslow thickness in the Champagne-Ardenne region.

No significant difference was observed between 2005–7 and 2009–11 for the overall incidence of invasive melanomas, the incidence of cases < 1 mm and the incidence of cases with a Breslow thickness from 1 to 2·99 mm. In contrast, the incidence of VTM decreased by 34%, from 1·07 to 0·71/100 000 habitants per year (= 0·01). This global decrease was the result of a 44% decrease in men, from 1·41 to 0·79/100 000 (= 0·03), and a 11% decrease in women (from 0·74 to 0·66/100 000, = 0·15). The mean and median Breslow thickness decreased from 1·95 mm and 1·00 mm (range 0·1–23·0) in 2005–7 to 1·68 mm and 0·85 mm (range 0·10–35·0) in 2009–11, respectively (= 0·06).

The proportion of VTM among invasive cases decreased from 19·2% to 12·8% (= 0·01), while the proportion of thin melanomas increased from 50·9% to 57·4% (= 0·05). The ratio of in situ/all melanoma cases increased from 20·1% to 28·2% (= 0·001).

Data in the control area

In the geographical control area of Doubs and Belfort territory, the incidence rates of melanoma remained stable between 2005–7 and 2008–10 in each group of thickness, with rates of 7·42 and 7·40/100 000 habitants per year for melanoma < 1 mm (= 0·68), and 1·02 and 1·10 for VTM (= 0·70), respectively. The mean and median Breslow thickness were 1·45 and 0·70 mm (range 0·1–9·8) in 2005–7, and 1·46 and 0·72 mm (range 0·05–18·0) in 2008–10 (= 0·87). The proportion of VTM was 12·0% in 2005–7 and 13·2% in 2008–10 (= 0·63) while the proportion of thin (< 1 mm) melanomas and in situ melanomas also remained stable [63·2% and 65·1% (= 0·61) and 20·7 and 19·8% (= 0·74), respectively]. Similar results were obtained when the year 2008 was excluded from analysis as in the Champagne-Ardenne region, for a comparison between 2005–7 and 2009–10.

Discussion

We report herein the first population-based evaluation of a secondary prevention programme concerning melanoma in France. This campaign targeted GPs and involved dermatologists as trainers. It confirmed previous observations that awareness of GPs regarding melanoma can be enhanced by specific community-level campaigns.[15]

When comparing the 3-year periods before and after the campaign, we observed a significant achievement of nearly all primary and secondary objectives, including a decrease of 34% in the incidence of VTM (primary objective), a decrease in the proportion of VTM and an increase in the proportions of melanomas < 1 mm and in situ melanomas. A borderline significant decrease in the mean Breslow thickness was also observed, from 1·95 to 1·68 mm. The incidence and proportion of melanomas of intermediate thickness (i.e. 1·01–3 mm) remained stable. No change in the incidence and distribution of melanomas according to their thickness was observed in the geographical control area of Doubs and Belfort territory, where no systematic secondary prevention campaign was conducted during this period.

It is important to note that our primary objective was achieved in a wider range in men, with a 44% decrease of the incidence of VTM, than in women (11%). This may be explained by the higher basal precampaign incidence of VTM in men and by specific messages highlighted during the campaign. Indeed, in view of previous epidemiological data concerning thick melanomas,[17, 19] both the short publication sent to all GPs by mail and the standardized training sessions attended by 32% of them included specific and emphatic messages concerning men – particularly older men – as important targets for detecting thick tumours.

Many secondary prevention campaigns have been carried out in developed countries since the 1980s.[23-28] Most of these campaigns directly targeted the general population through the media. These campaigns were regularly followed by an increase in melanoma incidence, associated in most cases with an increase in the proportion of thin melanomas. However, they generally failed to show a decrease in either melanoma-related mortality and/or in incidence of thick melanomas.[23-29] A recent American study concluded that surveillance and early detection efforts in the U.S.A. have not resulted in a substantial reduction of tumours with prognostically unfavourable thickness.[30] In this context, our results, although limited to a small geographical area, seem original and of interest. Recently, a pilot systematic screening study in Schleswig-Holstein, a German state comparable in size to the French Champagne-Ardenne region, first provided strong evidence (although not absolute proof) that a systematic skin cancer screening programme in the adult population may reduce melanoma mortality,[31] providing a rational basis for a large national screening programme, which is currently under way in Germany.[14] However, given the high direct and indirect costs of such a strategy, it is unlikely that similar programmes will be generalized in many developed countries in the near future.

In the absence of population-based systematic screenings, more limited, secondary prevention campaigns are of great importance. The rationale is strong for promoting GP training and the role of primary care providers in the early detection of melanoma. Among arguments recently highlighted by Geller et al.[10] regarding the U.S. population (i.e. most Americans do not have a dermatologist, few of them are screened for melanoma, most of them make frequent visits to their GP, most patients with melanoma have seen a physician in the year before diagnosis, many melanomas are not easily visible to patients), all are transposable to the French population and probably also to the populations of many developed countries. In France, a country counting 64 million inhabitants, 3400 dermatologists and 101 500 GPs at the beginning of the century,[18] we recently brought to light an important decline in dermatologists as first medical caregivers to patients with melanoma between the years 2004 and 2008.[12, 16] This may be explained both by an annual reduction in the number of dermatologists of about 6% since 2002,[18] and by the modification of French legislation in 2004 which limited direct access to specialists, including dermatologists, and encouraged patients first to consult their referent GP.[12, 16, 18] As reported by other authors,[32-34] we observed that melanoma diagnosed in a GP setting were much thicker than those diagnosed in other healthcare settings (particularly by dermatologists).[12, 16] However, we also demonstrated in this previous retrospective study that active detection by GPs (as opposed to detection by patients) was associated with a lower Breslow thickness and was improved by GP training.[16] The present study extends these preliminary observations in France, by providing, for the first time, population-based evidence that a systematic sensitization and training campaign for GPs may decrease the incidence of VTM and improve early diagnosis.

There are some limitations to this study. Firstly, it was performed in a limited French territory, and it remains unknown whether similar campaigns elsewhere would be followed by similar trends. Secondly, our study did not include an evaluation of melanoma-related mortality, which, although delayed in time, should be the gold standard for efficacy assessment. However, Breslow thickness is a powerful predictor of patient survival and has been recognized as a valid proxy endpoint for efficacy of secondary prevention campaigns.[10, 30, 35] In particular, the incidence of VTM (which plays a major role in the global mortality related to this cancer) is probably a valuable surrogate of subsequent mortality, as opposed to the more frequently used proportion of thick tumours and/or mean Breslow thickness, which may decrease as a simple consequence of an increasing recognition of early, nonlife-threatening melanomas.[36-38] A last limitation is the nonrandomized design of the study, which does not allow any definitive conclusion concerning the causal relationship between the campaign and the evolution of melanoma incidence and thickness. However, some important arguments for causality have to be emphasized. Firstly, no similar progress was observed in the geographical control area of Doubs and Belfort territory where no similar campaign was carried out. Secondly, the annual evolution of incident invasive melanomas and their thickness in the study region clearly shows two clear-cut periods (Fig. 1): a precampaign period (2005–7) with an overall stable number of invasive melanomas and VTM, and a post-campaign period (2009–11) with a reduced (and again stable) number of VTM compared with the first period. Both periods were separated by 2008, the year of the campaign, in which an increased diagnosis of tumours of any thickness was observed (Fig. 1). This suggests not only an anticipated diagnosis of a number of melanomas in 2008, resulting in a decrease of VTM in 2009, but also a persistent effect of the campaign during the following years. This possible persistent effect clearly seems to be a strong advantage of campaigns aiming to increase the awareness and skill of physicians concerning melanoma diagnosis, compared with media campaigns targeting the general public, the effects of which have frequently been assessed as transient and without any durable impact on thick tumours.[36]

In conclusion, the proportion of melanomas diagnosed in a general practice setting has been increasing in France for several years, giving rise to arguments regarding the optimal place of GPs in skin-tumour screening.[16] Our study first provides strong evidence that a well-focused awareness and training campaign for GPs by dermatologists may result in a decrease in the incidence of thick, potentially life-threatening melanomas. These preliminary data provide a rationale for a larger campaign targeting GPs in France. They also lead to reconsidering the respective role of GPs and dermatologists, taking into account the recent and future abrupt decline in the number of dermatologists.[18] GPs could act more widely as first medical caregivers to patients presenting with suspicious skin lesions, with dermatologists as trainers for GPs and second-line experts. In the near future, a nationwide campaign for secondary melanoma prevention could combine the systematic training of GPs by dermatologists and the wide diffusion of the following message to the general adult population: ‘Be aware of skin tumours, ask your GP for a skin examination’.

Acknowledgments

We thank the dermatologists who acted as trainers for GPs, and the Association d'Enseignement Post-Universitaire en Dermato-Vénérologie de Champagne-Ardenne (AEPUDV).

Ancillary