Self screening for risk of melanoma: validity of self mole counting by patients in a single general practice

Introduction

Malignant melanoma has a fairly low average annual incidence (about 1 in 10000)1 and has a poor prognosis if identified late.2 Identifying individuals at high risk of melanoma in a primary care setting, in addition to population approaches, is a potentially important part of the strategy to reduce mortality from melanoma.3

The number of benign moles is one of the best simple indicators of risk of melanoma, associated with high relative risks and population attributable risk.3 4 5 6 7 8 9 10 11 A case-control study in Australia used data from one half of the cases to try to predict melanoma in the other half: the number of benign moles was the best predictor of risk, and prediction was not greatly improved by the addition of other risk factors.7 Site specific counts of moles (on limb or trunk) are also associated with high risk of melanoma,4 5 6 7 8 9 10 11 and mole count at one site of the body correlates with the total mole count.6 12 13 The presence of atypical moles or the atypical mole syndrome may further increase risk of melanoma.14 15 16

Is it feasible to identify individuals with many moles (therefore at high risk of melanoma) in general practice? Although total body mole counts are likely to be too time consuming in primary care, site specific counts might be feasible to identify high risk individuals. Site specific self counts of moles by nurses predict risk of melanoma6 and self reported mole counts of “more than average (20)” provide as good an indicator of risk of melanoma as mole counts by trained interviewers on limbs.17 Interviewers' counts of raised moles on one arm, however, do not correlate well with dermatologists' counts.12 These observations have been documented in few studies, none in a British primary care setting, and no evidence exists for the comparative accuracy of untrained mole counting for different body sites.

Other risk factors for melanoma are freckling, tendency to burn, hair colour, probably sunburn in childhood, and possibly sunburn in adulthood and regular sunbathing or beach holiday.10 11 16 18 19 20 21 22 23 24 25 26 The case-control studies in Britain, however, used hospital controls. How common each of these risk factors is in British general practice lists is not known.

What kinds of practice should have greater priority for screening for risk of melanoma? Estimation of risk is especially relevant for young families with children and in higher social classes. Sunburn in childhood may be particularly important,7 8 18 22 24 27 so it is probably vital to identify children who are at additional risk because they have many moles, which are likely to increase in number with age.28 29 Higher social classes have more than twice the risk of melanoma,8 18 probably partly because of sunburn.18

We therefore chose a general practice population that was relatively affluent and young for a validation and prevalence study to estimate the sensitivity, specificity, and positive predictive value of patients' mole counts; to estimate the prevalence of reported changes in moles and of known risk factors for melanoma in an affluent general practice population; and to assess the extent to which current skin protection behaviour and mole counts are associated in families.

Methods

Results

Of the initial 237 subjects (from 67 families) initially sent a questionnaire, 219 (92%) took part. In all, 199 (84%) returned the questionnaire, and 212 (89%) agreed to be examined. The study population (113 male, 106 female) had a preponderance of social classes I and II (50.2% compared with 25% nationally) and of subjects under the age of 16 (43% compared with 20% nationally).

Discussion

This study concerns assessment of risk factors for melanoma in a relatively young and affluent British general practice population. The comparability with other studies of the adult median mole count and correlations between site counts and total count6 12 28 29 31 32 and the comparability of the validation data for the lower social classes and for adults alone with the data for the whole group suggest that it may be possible to extrapolate these findings to other practices and settings. To demonstrate the generalisability of these findings and for tighter confidence intervals further studies are needed.

Risk factors for melanoma include high mole counts, freckling, high social class, fair or red hair, tendency to burn, frequent severe sunburn, and probably atypical moles.4 5 6 7 8 9 10 11 18 19 21 22 23 24 25 26 33 On the basis of the nearest categories to a British study that demonstrated freckling, high mole count, and high social class as independent risk factors,8 6/109 (5.5%) adults in the present study have both many moles and freckles, and 28/109 (25.7%) have either factor. Individuals with both factors (odds ratio 21, 25/187 of the cases8) or either factor (120/187 of the cases8) had a conservatively estimated 0.5%-1% and 0.1%-0.2% 10 year chance respectively of developing melanoma.8 These estimates are probably increased twofold to threefold by either high social class or severe sunburn,8 18 a further twofold to threefold increase if the population incidence is 1-2 per 10000 per year,1 2 3 4 5 6 7 8 and a further 10-fold increase with less conservative British estimates of risk associated with moles.4 We need further research to see how much reinforcement by health professionals (if any) is needed to change the behaviour of high risk individuals after self screening, to assess if screening results in earlier diagnosis and reduced mortality, and to document costs such as increased workload and patient anxiety.34

Individuals commonly report changes in moles: 13% of the adult study population had reported some change in a mole during the previous three months and 5% a major change. If patients are asked to attend the surgery if they see changes in moles doctors should be aware of the workload implications and perhaps suggest that major changes only should be reported.

The validity of mole counts by untrained patients on different body sites has been documented in this study. High mole counts by patients on the front of the trunk (but not the arm) were reasonably sensitive and predictive of the observer's mole counts, which suggests that high risk individuals could assess their own melanoma risk. The difference between arm and trunk counts at high values may relate to the difficulty of seeing moles on the back of the arm and to freckling. At lower cut offs—for example, the 50th centile—the validation characteristics for arm mole counts improve. Further research is needed to confirm these observations in different settings and to assess possible costs and benefits of self screening. We also need to assess other screening methods—for example, arm mole counts when measuring blood pressure.

The within family correlation of sunscreen use during leisure time indicates that sun protection behaviour is associated within families. For individuals who do not protect their skin against the sun, it would be worth while emphasising that other family members should also be careful about sun protection.

The prevalence of the phenotype of atypical mole syndrome and atypical moles in this study is comparable to that in other populations.16 35 Clinical definitions of atypical moles or the atypical mole syndrome are limited by the subjective nature and number of features defining the term atypical, the number of atypical moles and other features defining the atypical mole syndrome, and changing criteria—for example, since the study began.16 36 Furthermore, most studies have not estimated the risk associated with atypical moles or the atypical mole syndrome. Before atypical moles or the associated syndrome can be useful in primary care screening, the issues of definition, risk, and training requirements need clarifying.

In conclusion, individuals with multiple risk factors for melanoma or who report changes in moles are likely to be common in practices with affluent patients, and sun protection behaviour is associated within families. Self screening with trunk mole counts may help to identify high risk individuals. Further research is needed to confirm the validity, adverse effects, and net benefits of screening for risk factors for melanoma in primary care.

We are most grateful to Professors David Mant and Ann-Louise Kinmonth for many comments on the manuscript, Professor Rona Mackie for encouragement and discussion of the project, Julia Newton for initial correspondence, to Clive Osmond for statistical advice, the staff and patients of Nightingale Surgery, Romsey, and the Royal Southants Hospital for its help and support. This project was supported by a small grant from the Wessex Dermatology Research Fund.