Evaluating Early Detection in the Diagnosis of Melanoma

In the largest population-based analysis of the role of length of time to diagnosis, Baade et al¹ in this issue of the ARCHIVES evaluate the role of time from first noticing a lesion to final diagnosis. Their results echo those of other authors^{2 - 9} (Table) and unpublished data (M.B., 1987-1989) to indicate that there seems to be no correlation between the length of time between first noticing a lesion to its ultimate diagnosis and the Breslow thickness of the lesion, the major clinical prognostic factor to date. The study by Baade et al is based on a combination of telephone interviews with 3772 Queensland, Australia, residents, with a solid response rate of 77.9% and pathological data from the population-based Queensland Cancer Registry supplying Breslow thickness and histologic findings for the lesions. By conducting reinterviews of 176 responders, these investigators confirmed a high level of reliability among their participants. They found intraclass correlation coefficients ranging from 0.85 to 0.90, impressive relationships, but nevertheless demonstrating a small level of error, which the authors adjusted for by regression calibration.

The finding that the length of time from noticing a lesion to diagnosis is not related to Breslow thickness is not a new finding but is an impressive substantiation of findings reported by most other investigators with smaller samples sizes, some with non–population-based subjects (Table). The same results were found in a case-control study¹⁰ conducted in Connecticut in the mid-1980s, but those results were never published because of concern that the potential misclassification by subjects might have muddied the results (M.B., unpublished data, 1987-1989). In fact, the subjects were likely to be as accurate as those in Queensland, and the results are consistent with the large proportion of such studies.

Although the findings are robust and the authors conclude that we should continue to educate individuals to recognize melanoma for early diagnosis, there are several issues that the authors did not raise and that should be considered in relationship to the findings. In the first place, they did not control for patient knowledge of melanoma symptoms. It is conceivable that their population included a mixture of different levels of awareness as to symptoms of melanoma and that the lack of association could be attributed to attenuation by including individuals who were attuned to early diagnosis and those who were not. Therefore, a lack of association may be the result.

In the second place, and contrary to the first argument, it seems that individuals who are aware of the signs and symptoms of melanoma have more self-efficacy and might allow a lesion to stay while watching carefully for any new signs. Therefore, these individuals could form the basis for a positive correlation between length of time from noticing a lesion and diagnosis because they are comfortable and knowledgeable and only go for a diagnosis when they are sure that the lesion should be excised. More than 90% of lesions in Queensland were found at a thin Breslow thickness and are associated with a 98% survival rate. Therefore, such patients may be the types of individuals that educational campaigns would like to target.

Indeed, Welch et al¹¹ pointed out that the biopsy rate for skin cancer (including melanoma and nonmelanoma skin cancers) has increased dramatically in the United States among the population receiving Medicare. Although the costs are not high, excessive biopsy procedures result in some morbidity and in inflation of the rates of skin cancer, with attendant public health worries that might be based on artifact rather than on a true increased incidence of melanoma. This is not a minor point and deserves further examination by those scientists interested in the public perception of health messages.

Although Baade et al¹ did not report several relevant characteristics of their population, this study is important and should serve as a milestone in the development of our understanding of the role of early detection and melanoma: that is, we still do not know much! The assumed positive relationship (the longer the delay, the thicker the lesion) has not held up. What will this mean for the future?

The authors note 2 salient reasons for the lack of association. In the first place, they note the long-held (but little analyzed) belief that “melanomas progress at different rates.” The biology of melanoma is not the same for all melanomas; some are more biologically aggressive, and some seem to be more indolent.¹² That said, it does not mean that the indolent melanomas should be left alone to progress at a slow rate. However, it behooves epidemiologists and molecular biologists to continue to develop methods to understand these differences between growth rates, so that appropriate prevention strategies can be developed. In the second place, the authors note that “The time of first recognition of an abnormality will depend on many factors and may bear no consistent relationship with any measure of the biological growth of the tumor.”^{1 (p1422)} This concept subsumes the suggestions that patient knowledge of the signs and symptoms of melanoma⁴ and the ability to recognize these (self-efficacy) are likely to be critical factors in estimating time 1 (the time between first noticing a suspicious spot and the first physician visit). In sum, this large and well-conducted population-based study provides the impetus to develop methods to more carefully evaluate the biology of the lesions and the role of patient knowledge and self-efficacy in early detection.

Correspondence: Dr Berwick, MSCO8 4603, Room 103 A, Division of Epidemiology, 1 University of New Mexico, Albuquerque, NM 87131-0001 (mberwick@salud.unm.edu).

Financial Disclosure: None reported.