European Journal of Dermatology

The incidence of cutaneous melanoma is increasing all over the world [1]. Although the clinical diagnosis of melanoma is reputedly difficult, nevertheless early diagnosis and treatment are generally considered the most effective means of ameliorating the outcome [2]. Many factors have been demonstrated to influence prognosis, but tumor thickness is the major determinant [3]. Early diagnosis is probably the best current method to improve mortality rates and delayed recognition of melanoma is considered the principal factor impeding early diagnosis. However, relatively few studies on this topic have been performed [4-10]. It is generally accepted that a correlation between delay and prognosis exists although not as strong as supposed [8]. The aim of our work was to investigate the factors related to early detection of melanoma and assess the possible correlations and effects of these factors upon delay.

Consecutive patients referred with cutaneous melanoma were enrolled in this study. All patients suffered from histologically confirmed melanoma diagnosed in our clinic between September 1994 and December 2000. They were interviewed by a trained dermatologist using a standardized questionnaire consisting of thirteen questions concerning melanoma awareness and detection, the patient's demographic data, type of physician consulted for their diagnosis and tumoral characteristics. Each patient was interviewed and examined by our medical staff. The questionnaire included three groups of factors:

tumoral factors: site of presentation, histological subtypes (superficial spreading melanoma, nodular melanoma, lentigo maligna melanoma), association with pre-existing nevus, Breslow thickness, size in vivo, presence of pigmentation;

behavioural data: causes of delay, source of knowledge of melanoma existence and significance, persons or sources calling attention to a suspicious lesion and other personal factors which might have influenced the period of time between the first observation and treatment of the tumor.

To minimize sources of error in recalling anamnestic dates, only three types of delays were indicated. Patients who were not able to respond exactly to the questionnaire were not included in the study. For this reason only 216 out of the 270 patients interviewed were enrolled in the study.

Delay in diagnosis of melanoma was defined as the interval in months from the patient's report of first noticing a suspected lesion to the date of the removal of a histologically confirmed melanoma. A sizable proportion of patients are referred to as "coincidentally diagnosed" because their tumors were detected by a physician during a visit for unrelated problems or during routine clinical examination. These patients could not be classified as having delayed in seeking medical attention.

Patients were considered as having self-detected melanoma when they reported that they themselves had suspected a lesion and for this reason they had gone to a physician.

delay of patient (patD) defined as the interval from the patient's initial observation or awareness of change in a lesion to the date of the initial medical consultation;

medical delay (medD) defined as the interval from the physician's observation of the patient's lesion to the appropriate suspected diagnosis;

medical referral delay (refD) defined as the interval from the appropriate diagnosis of suspicion to the excision of a histologically confirmed melanoma. This delay regards the time for the patient's correct treatment and comprises the interval in days from the medical diagnosis to the surgical treatment of the tumor.

Only patients agreeing to be subjects were enrolled in this study and informed consent was obtained.

Data evaluation was performed using the statistical package for Mackintosh. The information obtained was computerized and analysed with parametric and non parametric tests. The Spearman rho's coefficient was used to investigate a linear correlation between two continuous variables. For calculation purposes only, in order to evaluate the correlations, in situ melanomas were all standardized to a 0.01 mm thickness.

To compare the distribution of qualitative variables, the chi square test was used. The Mann-Whitney U and Kruskal-Wallis test was used to test the equality of medians of two respectively multiple independent variables.

Of the 216 patients enrolled in the study 125 were males (57.9%) and 91 females (42.1%).

The mean age was 53.23 ± 15.40 years (range between 16-92 years) with no significant difference between inter-gender comparison.

160 patients were married (74%) or co-habiting, of these 64% were males and 36% females; 56 patients (23.93%) lived alone (41% males, 59% females).

46 patients had attended primary school (21%), 56 patients had attended junior high school (26%), 93 patients had attended high school (43%), 21 patients had graduated from university (10%).

The leg was the most common site involved (31%), followed by the back (28.7%), arms (17.6%), chest (13.9%) and head and neck (8.8%). As expected, melanomas were more reported on legs in women and on backs in men.

The most common histogenic type was the superficial spreading melanoma (SSM) and occurred in 190 patients (87.96%) followed by lentigo maligna melanoma (LMM) (6.48%) and nodular melanoma (NM) (5.56%).

The mean age of patients was higher for LMM (75.43 ± 7.16 years) and lower for SSM (55.82 ± 14.85 years).

Nevus associated melanomas were present in 18.06% of cases. 16 cases (7.41%) of amelanotic melanomas were observed. The mean measured Breslow thickness for all the lesions was 0.75 ± 1.14 mm (male 0.80 ± 1.30 mm; female 0.68 ± 0.88 mm) with no difference between sexes. "In situ" melanomas accounted for 45 cases.

Pigmented melanomas had a mean Breslow thickness of 0.64 ± 1.03 mm, amelanotic melanoma of 2.13 ± 1.53 mm. In the different sites the mean thickness was as follows: head and neck 0.97 ± 1.33 mm; arms 0.54 ± 0.76 mm; lower limbs 1.03 ± 1.57 mm, chest 0.34 ± 0.31 mm, back 0.70 ± 0.91 mm.

The mean depth of invasion was significantly less for SSM (0.68 ± 1.15 mm) than for NM (2.69 ± 1.28 mm).

The tumor thickness measured was < 0.76 mm in 77.8% of the patients; > 0.76-1.5 mm in 7.9%; > 1.5-4 mm in 12.5%; and > 4 mm in 1.8%.

A negative correlation between thickness and age was found (p < 0.02; r = - 0.15).

There was a positive correlation between size and thickness for superficial spreading melanomas (p < 0.003 r = 0.21).

Tumoral lesions were self-detected in 102 (47.22%) patients: 45 males (36%) and 57 females (64%); diagnosed by a physician in 64 patients (29.63%) (a general practitioner in 14 cases and a dermatologist in 50 cases); by relatives in 32 cases (14.81%). Other modalities account for 18 cases (8.33%).

Self detected cases had a mean Breslow thickness of 1.05 ± 1.5 mm, physician detected cases had a mean Breslow thickness of 0.42 ± 0.62 mm; of these, general practitioners detected melanomas with a mean thickness of 0.67 ± 1.02 mm, dermatologists 0.35 ± 0.5 mm; relatives 0.56 ± 0.5 mm, other 0.51 ± 0.27 mm.

As regards the modality of discovery of the tumor according to the site of localization, the values observed are indicated in the Table I.

Coincidentally diagnosed cases were observed for 55 patients (25.46%): 40 males (73%) and 15 females (37%).
The mean depth of the 55 coincidental lesions (0.32 ± 0.35 mm) was significantly less (p < 0.001) than the 102 self-detected lesions (1.05 ± 1.5 mm).

Mean age of patients with self-detected diagnosed melanomas was lower but not significantly different than in coincidentally diagnosed ones (54.8 ± 14 vs 62.2 ± 15.7 years). Melanoma was more often self-detected in females than in males (63% vs 36%) (p < 0.01). 46% of patients had self-detected lesion located on lower limbs, 16.7% on the chest, 15.7% on back, 9.8% on head and neck, 11.8% on the arms. On the contrary, coincidental cases were located on the back in 42% of patients, limbs in 18%, arms in 24%, chest in 12%, head and neck in 4% of cases.

The level of education and sources of knowledge did not differ among patients with self-detected and coincidentally diagnosed melanomas. Lesions in men were detected more frequently by family members (19% vs 8%) (p < 0.05) while melanomas in women tended to be self-detected more often (63%).

Of the 32 cases where the partner detected melanoma, 24 cases were in male patients.

It was observed that relatives first discovered a suspicious lesion on back of the men in 18 cases (75% of familial observation), on the contrary, a familial-discovered suspicious lesion on the back of a woman was observed in only 1 case (12.5%).

Before the diagnosis of melanoma, 129 patients (60%) had knowledge of the word "melanoma" and its true meaning (a malignant skin tumor). The sources of information were mass media (46.3%), a physician (5%) or a relative (8.33%). 87 patients (40%) had never heard the word "melanoma" nor knew its meaning.

The reasons for non attendance to a physician or/and for delay were different. 111 patients (51.39%), 57 males (51.3%) and 54 females (48.7%) referred to carelessness (minimize the seriousness of the problem, fear and anxiety, no time/busy, decided to wait and see, symptomless).

46 patients (18.52%) referred a delay for time of appointment with no differences between sexes, 8 (3.7%) patient delayed for holidays, 2 cases were delayed for unspecified problems, 55 cases (25.46%) were coincidentally diagnosed.

The mean time elapsed between first detecting a suspicious lesion and the time of tumor treatment (total delay) was 8.11 ± 10.84 months; between first noticing a suspicious pigmented lesion to initial consultation (patient delay) was 6.11 ± 9.75 months; the first visit to a physician and suspected diagnosis for surgical treatment (medical delay) was 1.53 ± 5.34 months; and from physician suspicion and definitive treatment was 0.46 ± 0.63 months.

The factors considered for delay and its components (totD, patD, medD, refD) in melanoma diagnosis and treatment were represented in the Table II.

Because the site of the tumor could influence early recognition of melanoma, we considered data by sites. The total delay for lesion recognition on the head-neck region was significantly longer (p < 0.001) (12.58 ± 16.26 months) than the delay for the back region (4.40 ± 5.73 months); lower limbs had a delay of 10.56 ± 11.75 months, chest 9.4 ± 11.71 months, and arms 6.56 ± 10.12.

The same is of value for physician delay but only for the head-neck region vs the back region.

Lentigo maligna melanoma had the higher total delay (11.7 ± 18.26 months) followed by superficial spreading melanoma (8.05 ± 10.3 months) and nodular melanoma (4.87 ± 6.5 months) (p < 0.001)

For this factor, the significant component of delay was referred to patient delay only (p < 0.001).

There was a significant (p < 0.04) positive correlation (r = 0.19) between Breslow thickness and the total delay. This delay was significant referring only to patient delay (p < 0.03).

There was a significant positive correlation (p < 0.03; r = 0.14) between size of the tumor and total delay. This delay was significant referring only to patient delay (p < 0.001)

Patients who self-detected a suspicious lesion had a mean total delay of 11.7 ± 12.4 months which was higher than the mean delay of physician and dermatologists (6.13 ± 8.7 months and 2.9 ± 6.9 months respectively). The delay of a suspicious lesion first observed by a relative or other were 6 ± 6.4 months and 7.5 ± 11.2 months respectively.

Amelanotic melanoma had a significantly delayed diagnosis (12.72 ± 12.25 months) in comparison with pigmented melanomas (7.42 ± 10.71 months) (p < 0.001) and this delay was particularly due to physicians (4 ± 9 months vs 1.34 ± 5 months) (p < 0.04).

Sex significantly influenced the total delay. In fact, women had a mean total delay of 10.27 ± 12.4 months, men 6.5 ± 9.3 months (p < 0.01). The delay was referring only to patient delay.

A significant negative correlation was observed between total (p < 0.01) and physician delay (p < 0.02) and age (p < 0.01; r = - 0.17).

There were no differences among causes of patient delay and mean age, anatomic site of lesions, level of education, knowledge of the problem, civil status or pigmentation. 51% of the patients delayed the consultation of a physician because of carelessness (anxiety, fear, no time/too busy). They tended to have a longer patient delay and a higher Breslow thickness (0.99 ± 1.41) (p < 0.001).

There was a significant correlation between the causes of delay and Breslow thickness (Table IV).

Regarding physician delay, 22 cases (10.19%) were observed in which the practitioner or the specialist delayed diagnosis or treatment, 7 cases (3.24%) were due to a further control of a lesion then excised, 187 patients were immediately sent to surgical excision. The mean delay of these cases was 11.7 ± 10.2 months and 4.7 ± 2.73 months respectively. No correlation between physician delay and anatomic location of the lesion was observed. The thickness of lesions in which there was a delay by physician was 0.72 ± 1.13 for cases sent to surgical excision, 1.16 ± 1.36 when the delay was due to a misunder-standing. This difference is not significant. On the contrary, pigmentation of the lesion significantly delayed the time of diagnosis by the physician (4 ± 9 months vs 1.34 ± 5 months for the pigmented melanomas) (p < 0.04). 176 (88%) pigmented melanomas and 11 amelanotic melanomas were correctly removed, 24 pigmented (12%) and 5 amelanotic melanomas (31%) were delayed. This difference is significant (p < 0.03).

Delay of patients was not related to knowledge of melanoma problems. Patients who had knowledge of melanoma problems had a mean delay of 5.3 ± 6.15 months, patients who had never heard the word "melanoma" had a mean delay of 7.3 ± 11.3 months.

Level of education and civil status per se were not correlated with delay in diagnosis.

Factors related to diagnostic delay are complex and often intermingled. In order to lessen the possible bias and to simplify anamnestic recording of data by the patients, we considered only three simple types of delay. In fact, we think that patients would have had some difficulty in remembering exactly the time in months of more complex anamnestic recording.

For the same reason we have reduced the time of investigation to within the last five years and included only patients with sufficiently certain reliability in their recording.

To our knowledge, this is the first study which tries to define diagnostic delay in melanomas within the specialist structure of the Italian healthcare system. The patients were recruited both from a specialist university hospital setting and from routine practices. We think that this double system might represent the normal way of melanoma screening in our district area. On the whole we referred to a population of 500,000. Our results are complex and sometimes difficult to explain, if at all. Three main subgroups of items have been explored. As in other studies [5, 9, 11] we confirm that the major component of delay is the patient (6.1 months), then the physician (1.5 months). The total delay amounted to 8.1 months. Some factors are interconnected with patient delay. Site, sex and the type of the lesion significantly influenced this delay. However, in particular, accessibility of the tumor to view did not seem to influence early diagnosis and treatment.

Lesions on visible sites of the body had a mean delay significantly higher than lesions on less visible sites. These results are in contrast with some studies [4, 12] but in agreement with others [5, 9]. As suggested, this datum implies "that the failure to appreciate the implications of change in a pigmented lesion rather than lack of visibility is a more important factor in patient delay" [5].

Curiously, we observed that melanoma of the back of a married man had a shorter delay than melanomas of the back of a single man or woman with the tumor in the same site. That implies and strengthens the assumption that different and more complex factors are involved in patient delay and visibility. Sex may be one of them, as found by some studies [5] with men that delayed less than women although conflicting results have been reported [9, 10].

It may be that men pay little attention to their skin as suggested [8] but other considerations besides patient characteristics may influence sex delays of patients. For example civil status (in our study significant only if we consider the site of tumor) or the role of relatives. Blum and Kaskel pointed out the importance of the latter in encouraging men to undergo a medical visit [9, 13].

The importance of behavioral/psychological components such as minimizing the seriousness of melanoma was well stressed in the past although not related to a specific gender difference [4].

Delay and extent of previous knowledge of melanoma did not seem to be a function of educational/socio-economic status [4, 7, 9].

Studies about delay and behavioral factors are very complex because of the many and not completely recognized variables implicated [4, 6] but until now the data observed have failed to demonstrate a direct correlation between awareness of melanoma and age, educational or socio-economic status.

Delay is significantly influenced by other variables. Probably the most important and intriguing factor related to delay is the Breslow thickness. Although many studies were not able to find any correlation between delay and thickness, Richard et al. first demonstrated this relation [8]. These Authors reported a statistically significant positive correlation between tumor thickness and the interval of time between the date the patient noticed or became aware of the lesion and the date the patient believed the lesion was potentially dangerous (D1-D2 interval). This correlation was observed only in a subset of patients who put the time interval of attention at less than 5 years, in all histogenic types, and in SSM when the types were considered in isolation. The correlation was lost when NM were studied in isolation. Our results are very similar in a less numerous population. We have observed a positive correlation between thickness and patient delay only. Although the definition and the assessment of delay are a possible drawback of the studies on this topic, nevertheless we think that our delay referred to as by the patient only, may be similar to the D1-D2 interval proposed by Richard and considered as referred to the patient at all times.

Histotype has been advocated as being an important factor of delay by some authors [5, 8] but denied by others [9]. A direct relationship between histologic subtypes and thickness has not necessarily been demonstrated. We have found a higher total and patient delay for lentigo malignant melanoma but lower for nodular melanoma with NM having a higher mean thickness than SSM and LMM.

Probably fast evolving melanomas such as nodular melanomas draw more suspicion and attention than slow growing ones. These data underline that the Breslow thickness results not necessarily only from the delay in the diagnosis but also from the biological aggressiveness of the tumor [8].

The conflicting results obtained about this topic [4, 5, 7] may be explained considering that "the respective impact of the delay and the biological behaviour of the melanoma on the final thickness of the tumor may depend on the population studied" [8].

The importance of the correlation of thickness and delay lies in the implications on prognosis with the Breslow thickness being its major determinant. The observation about the shorter delays in thicker tumors may imply that the "inherent biological activity of a tumor may determine the prognosis rather more than public education campaigns aimed at earlier diagnosis, particularly in the case of NM" [8]. In fact, prognosis in slow growing melanomas like LMM and SSM are most influenced by early recognition and treatment campaigns. Public education campaigns have only resulted in excision of non progressing lesions in the radial growth phase. The problem seems to be to improve early detection of thick and fast growing cases.

In a recent study, Richard observed that "shortening the delay in the diagnosis by intensifying education and screening of the whole population may lead any longer to a stronger improvement of prognosis" [8]. In our opinion,
this assumption is not completely true. Undoubtedly superficial spreading histotype is most influenced by early recognition and treatment campaigns. But, in the past, data about detection of the tumor between patients with self-detected melanomas and patients with coincidentally diagnosed melanomas were compared. The former had a longer delay than the latter [8]. Moreover patients with self-detected melanomas showed a positive correlation between patient delay only and thickness. Our results are similar, in fact patients with self detected melanomas had a mean thickness and a mean delay greater than patients with coincidentally detected melanomas. All these results, besides suggesting that medical delay has little influence on prognosis, strengthen the usefulness of early detection and treatment campaigns performed by physicians in order to ameliorate the diagnosic skilfulness and to shorten the delay. Moreover, in consideration of the successes of educational campaigns and of the above-mentioned findings about delay of thick and fast growing cases, we think that further education campaigns should be carefully focused even on the meaning of evolution or growth of a lesion than other signs of ABCD rule, stressing the importance of a new or fast growing lesion.

Total delay and patient delay were correlated with sex with female patients having a mean delay longer than men.

Females had a higher proportion of self-detected melanomas than males, but their delay was longer, suggesting that other factors, probably of behavioural origin, rather than simple detection of a lesion or the knowledge of the problem are implied in a early diagnosis. Melanomas, on the contrary, were more frequently diagnosed in males and although males pay little attention to their skin, nevertheless they benefit from the intervention of relatives for a more precocious diagnosis.

As regards our data on physician delay, only three types of significant correlations are shown: namely, pigmentation, age and site.

Site correlation with physician delay concerned the same body region as that of patient delay. In this case, the concept of visual accessibility was partially explained with lower delay for the back region. Probably the longest delay, that for the head and neck region, may be explained by considering the effect of aesthetic results. We observed a negative correlation between physician delay and age. The presence of shorter physician diagnostic delay in older people is a datum already observed by Richard and it may reflect medical surveillance for other problems in this age group [8]. Although not unequivocal [9, 10] the lower delay observed in older cases may reflect medical surveillance for other problems in this group of age but the different observations about this topic may depend by the population observed or the healthcare system.

More interesting is the result regarding physician delay and amelanotic melanomas. To our knowledge the importance of the type of clinical melanoma was never considered in similar studies. The presence of a significantly delayed physican diagnosis of amelanotic melanomas probably implies the necessity of further education or updating programmes for physicians, with new educational approaches targeting the medical profession. It is well known that amelanotic melanomas represent a difficult diagnosis. Accuracy in diagnosis of these lesions is very low [14] but it is important that physicians and dermatologists in particular, have a high degree of attention and suspiciousness about every fast growing lesion. It is well known in fact that initial misdiagnosis by a physician of these cases with inappropriate reassurance exacerbates the delay [9].

Factors of delay in the diagnosis of melanoma