Self-reports of mole counts and cutaneous malignant melanoma in women: Methodological issues and risk of disease

Christopher Bain, Graham A. Colditz, Walter C. Willett, Meir J. Stampfer, Adele Green, Ben R. Bronstein, Martin C. Mihm, Bernard Rosner, Charles H. Hennekens, Frank E. Speizer

Research output: Contribution to journalArticlepeer-review

42 Scopus citations


The relation of the presence of moles (nevi) on all four limbs to risk of cutaneous malignant melanoma was explored among 98 incident cases aged 32-59 years at diagnosis and 190 age-matched controls drawn from the Nurses' Health Study, a prospective cohort of female nurses in the United States. Cases diagnosed during follow-up from 1976 to 1982 were included in this study. Participants reported counts of all moles and raised moles alone on postal questionnaires. Distributions of moles were similar for right and left sides on upper and lower limbs for cases and controls. Counts declined with increasing age for all women, from a median of 15 for the youngest tertile of controls (aged 36-46 years) to three for the oldest (aged 54-62 years). Cases had more moles than did controls (medians of 23 and 9, respectively, for total moles on all four limbs): The presence of any mole on a limb gave relative risks for melanoma ranging from 2.2 (95% confidence interval (Cl) = 1.2-4.0) for one or more moles on an arm to 2.9 (95% Cl = 1.6-5.3) for one or more moles on the lower limb. For raised moles, relative risks were 1.7 (95% Cl = 1.0-2.7) for arm, 2.1 (95% Cl = 1.3-3.5) for lower limb, and 3.5 (95% Cl = 2.0-6.3) for leg (below knee). The highest site-specific risk (i.e., for any moles on the same limb as the melanoma vs. no moles on that limb) was for moles on the lower limb (relative risk = 5.0 (95% Cl = 1.8-13.5)). There were positive and significant trends in overall and site-specific risk with increasing numbers of moles on all limbs when absolute mole counts were considered, e.g., for total moles on all four limbs combined, x for trend = 4.0, one-sided p<0.001, with relative risk for more than 100 moles versus none of 6.0. Inclusion of sun exposure and other constitutional factors in logistic regression analyses did not alter these observed relations between the presence of moles and risk of melanoma.

Original languageEnglish
Pages (from-to)703-712
Number of pages10
JournalAmerican journal of epidemiology
Issue number4
StatePublished - Apr 1988


  • Epidemiologic methods
  • Melanoma
  • Nevus


Dive into the research topics of 'Self-reports of mole counts and cutaneous malignant melanoma in women: Methodological issues and risk of disease'. Together they form a unique fingerprint.

Cite this