Skip to content

Adamson AS, et al. Estimating Overdiagnosis of Melanoma Using Trends Among Black and White Patients in the US. JAMA dermatology. 2022;158:426-431. PMID 35293957

One-paragraph summary

Using joinpoint regression of SEER data from 1975 to 2014, the authors compared melanoma incidence and mortality trends between Black and White patients in the US. The design uses mortality trends in Black patients as a marker for improvements in medical care — a population in whom incidence barely rose — and thereby estimates what mortality in White patients would have been had care not improved. That estimate stands in for the change in true cancer occurrence, and overdiagnosis is calculated as the difference between observed incidence and estimated true occurrence. Analyses were stratified by sex.

Key findings

  • Incidence, 1975–2014: incidence rate ratio 4.01 (95% CI 3.65–4.41) in White women and 5.97 (5.47–6.52) in White men, against 1.21 (0.97–1.49) in Black women and 1.17 (0.77–1.78) in Black men.
  • Mortality, same period: mortality rate ratio 0.76 (0.63–0.90) in Black women and 0.72 (0.62–0.84) in Black men — i.e. a ~25% decline; 1.02 (0.96–1.09) in White women; 1.49 (1.25–1.77) in White men.
  • Had medical care not improved, estimated mortality would have increased 60% in White women and more than doubled in White men.
  • Estimated overdiagnosis in 2014: 59% (95% CI 45–70) of White women and 60% (32–75) of White men diagnosed with melanoma.

Limitations

  • The design assumes medical care improved comparably for Black and White patients over four decades — an assumption the same literature elsewhere contradicts, since distant-stage survival remained consistently lower in Black patients throughout (Di Carlo 2020, PMID 33409455).
  • The confidence intervals are wide, particularly for men (32–75%).
  • It is an ecological estimate: it identifies a population-level discrepancy, not which individual diagnoses were overdiagnoses.
  • The counter-literature holds that increases occur in thick and advanced disease too, including among men in the lowest-socioeconomic-status neighbourhoods where detection-driven explanations are least plausible (Clarke 2017, PMID 28736233), and that registry granularity implicates factors beyond overdiagnosis (Reinhart 2024, PMID 38957842).
  • An alternative reading holds that the incidence rise is largely real and UV-driven but that most of the new sun-caused melanomas are non-lethal (de Gruijl 2022, PMID 36191966) — a hypothesis that predicts the same curves.

Why it matters

This is the most-cited quantitative estimate of melanoma overdiagnosis and the one that moved the argument from "the incidence–mortality divergence suggests overdiagnosis" to a specific number with a confidence interval. It sits directly against the global-burden projections that extrapolate the same incidence series forward to 510,000 cases by 2040 (Arnold 2022, PMID 35353115) and against UVR-attributable-fraction modelling that assigns ~88% of cases to ultraviolet exposure (Oh 2026, PMID 42502459). Neither literature has reconciled its estimate with the other's — the junction recorded as OQ-1 and D1 in OPEN-QUESTIONS.md. The companion paper by the same author, showing that US county melanoma incidence correlates with dermatologist supply (r = 0.43) and not with UV daily dose (r = 0.03), supplies the mechanism (PMID 36190719).

Cited by wiki pages

  • screening and overdiagnosis
  • epidemiology and global burden
  • risk factors and prevention
  • red flags and safety concerns
  • overview