Skip to content

Melanoma screening and overdiagnosis

TL;DR — Melanoma incidence has risen several-fold in light-skinned populations while mortality has been flat or falling, and the two competing explanations for that divergence — real UV-driven disease increase versus overdiagnosis of indolent lesions — remain unresolved after twenty years of argument. The USPSTF has now three times concluded that evidence is insufficient to recommend visual skin examination screening in asymptomatic adults (I statement in 2009, 2016 and 2023) (Bibbins-Domingo 2016, PMID 27458948; Mangione 2023, PMID 37071089). The most-cited quantitative overdiagnosis estimate — 59% of White women and 60% of White men diagnosed in 2014 in the US — comes from a natural-experiment design using Black patients as the comparison series (Adamson 2022, PMID 35293957), and a scoping review of 35 studies found the three published estimates spanning 29–60% (Bjørch 2024, PMID 37793786). The one population program with long follow-up, Germany's, produced a mortality decline after its Schleswig-Holstein pilot that did not survive national roll-out: an ecological comparison of 15 German federal states against nine neighbouring countries found pooled annual mortality change of −1.8% in Germany versus −2.2% in the controls, P = .42 (Hübner 2026, PMID 42268621). The deepest problem is that the field cannot yet identify which thin melanomas are lethal, so screening's benefit and harm cannot be separated by better epidemiology alone.

The divergence that starts the argument

Series Direction, 1975–present Source
US White female incidence, 1975–2014 IRR 4.01 (95% CI 3.65–4.41) Adamson 2022, PMID 35293957
US White male incidence, 1975–2014 IRR 5.97 (5.47–6.52) PMID 35293957
US Black female incidence, same period IRR 1.21 (0.97–1.49) PMID 35293957
US Black male incidence, same period IRR 1.17 (0.77–1.78) PMID 35293957
US White female mortality MRR 1.02 (0.96–1.09) PMID 35293957
US Black female mortality MRR 0.76 (0.63–0.90) PMID 35293957
US non-Hispanic White incidence 2001→2015 20.7 → 28.2 per 100,000 Thrift 2020, PMID 31346623
European invasive melanoma incidence 1995–2012 AAPC 4.0% men / 3.0% women Sacchetto 2018, PMID 29395684
European in situ incidence, same period AAPC 7.7% men / 6.2% women PMID 29395684
European thin invasive melanoma AAPC 10% men / 8.3% women PMID 29395684

The pattern is consistent across registries: the increase is concentrated in in situ and thin invasive lesions, while thick-melanoma incidence rises slowly or not at all and mortality is flat or declining (Sacchetto 2018, PMID 29395684; Huang 2023, PMID 37296344). Global registry synthesis reports age-standardised incidence 3.4 and mortality 0.55 per 100,000 in 2020, with incidence rising and mortality falling (PMID 37296344), and analysis of WHO death-certification data finds melanoma mortality falling 40–50% in Australia over recent decades (De Pinto 2024, PMID 38391175).

The overdiagnosis case

Definitional anchor. Overdiagnosis is diagnosis of a cancer that would not have caused symptoms or death; it requires a silent disease reservoir plus activity that detects it, and Welch's cross-cancer synthesis lists melanoma among the tumours where population data suggest it is occurring (Welch 2010, PMID 20413742).

  • Biopsy rates track incidence. Across nine SEER areas 1986–2001 in people aged ≥65, skin biopsy rates rose 2.5-fold (2,847 → 7,222 per 100,000) while melanoma incidence rose 2.4-fold (45 → 108 per 100,000); 1,000 additional biopsies were associated with 6.9 (95% CI 3.1–10.8) extra melanoma diagnoses after adjustment, and the extra cases were confined to in situ (4.4, 2.1–6.8) and local disease (2.3, 0.0–4.6) with no association with advanced melanoma. Mortality changed little (Welch 2005, PMID 16081427).
  • Geography follows physicians, not sunlight. Across 727 continental US counties with 235,333 melanomas diagnosed 2012–2016, UV daily dose was uncorrelated with melanoma incidence (r = 0.03, P = .42) while median household income was (r = 0.43, P < .001); counties with no dermatologists and primary-care shortages had the lowest incidence. Incidence and melanoma mortality were barely associated (r = 0.09, P = .05) against r = 0.96 for lung cancer incidence and mortality in the same counties (Adamson 2022, PMID 36190719).
  • The magnitude estimate. Using mortality trends in Black patients as a marker for improvements in care and therefore for true occurrence, an estimated 59% (45–70%) of White women and 60% (32–75%) of White men diagnosed with melanoma in 2014 were overdiagnosed (Adamson 2022, PMID 35293957).
  • Independent synthesis. A scoping review of 35 empirical studies (29 register, 3 cohort, 1 case-control, 1 survey, 1 RCT) found annual incidence increases of 0.39–6.6% with little or no mortality increase; the three studies that estimated overdiagnosis directly ranged 29–60%, and every study of skin screening found increased detection of predominantly thin and in situ melanoma (Bjørch 2024, PMID 37793786). Ecological evidence from Denmark reaches the same conclusion (Olsen 2024, PMID 38752346).
  • Melanoma in situ specifically. A 2026 review of MIS argues incidence increases have outpaced invasive melanoma without proportional mortality change, and that quantitative studies place the overdiagnosed proportion of MIS above 60–80% in some settings (Greco 2026, PMID 42279200).

The counter-case

The counter-argument is not that overdiagnosis is absent but that it is not the dominant driver and not the main clinical problem (Paiva 2022, PMID 35895044; Adamson 2022 reply, PMID 35895049).

  • Increases occur in thick and advanced disease too. In 58,000+ California melanomas 1998–2012, incidence rose in almost every stratum defined jointly by thickness/stage and small-area socioeconomic status, with the largest relative increases in regional, distant and ulcerated disease — and specifically among men in the lowest-SES neighbourhoods, where access-driven overdetection is least plausible (Clarke 2017, PMID 28736233).
  • Registry granularity complicates the picture. In 2,310 primary cutaneous melanomas over 50 years in Olmsted County, incidence rose 11.1-fold since the 1970s and continued rising in the last decade, driven by women aged 40+, while melanoma-specific mortality fell from 26.7% in the 1970s to 1.5% in the 2010s; the authors argue the age–sex pattern implicates factors beyond overdiagnosis (Reinhart 2024, PMID 38957842).
  • A sun-exposure alternative to the scrutiny hypothesis. A third reading holds that the incidence rise is largely real and UV-driven but that most of the new sun-caused melanomas are non-lethal — consistent with the Icelandic sunbed episode, where incidence rose sharply in young women while melanoma mortality in the young stayed near zero (de Gruijl 2022, PMID 36191966). Iceland's national registry from 1957–2021 shows invasive incidence peaking 2002–2006 and declining thereafter, MIS peaking earlier and re-rising in women since 2017, and melanoma mortality falling since 2012 (Thomas 2025, PMID 39444324).

The two readings are not distinguishable by incidence data alone. They differ on whether the excess thin lesions are misclassified benign biology or genuine UV-induced neoplasms with low lethality, and both predict the same registry curves.

Pathological indeterminacy is upstream of the epidemiology

Overdiagnosis arguments assume a reproducible diagnostic label. It does not exist in the diagnostic middle.

Interpretation class (M-Path) Intraobserver reproducibility Accuracy vs expert consensus
I — nevus / mild atypia 76.7% 92% (90–94)
II — moderate atypia 35.2% 25% (22–28)
III — severe atypia / melanoma in situ 59.5% 40% (37–44)
IV — pT1a invasive melanoma 63.2% 43% (39–46)
V — ≥pT1b invasive melanoma 82.6% 72% (69–75)

Source: 187 US pathologists, 240 cases, 8,976 independent interpretations, two phases ≥8 months apart (Elmore 2017, PMID 28659278). At population level an estimated 82.8% (81.0–84.5) of melanocytic biopsy diagnoses would be verified by an expert consensus panel, with 8.0% (6.2–9.9) overinterpreted and 9.2% (8.8–9.6) underinterpreted (PMID 28659278). Second opinions help but do not solve it: misclassification was 52.8% (51.3–54.3) with general pathologists and no second opinion, falling only to 36.7% (33.1–40.7) with universal second opinions by subspecialty dermatopathologists, and in situ and thin invasive melanoma were "relatively intractable to all examined strategies" (Piepkorn 2019, PMID 31603483). The absence of a biomarker distinguishing indolent from progressive early melanocytic lesions is the acknowledged root cause (Kutzner 2020, PMID 32841508).

What the screening evidence actually shows

Evidence Design Result
SCREEN, Schleswig-Holstein 2003–04 Population program, 360,288 of 1.88 million eligible (19% participation) 3,103 malignant skin tumours; invasive melanoma incidence +34% during the program; melanoma mortality 5 years later 0.79 vs 2.00/100,000 expected (men), 0.66 vs 1.30 (women) (Breitbart 2012, PMID 22074699)
German national program from 2008 Ten-year review Evidence for efficacy judged present but weak, no RCT, mortality reduction had not appeared after national roll-out; screening intensity lower than in SCREEN (Hübner 2018, PMID 30411137)
German national program, 2009–2022 Ecological comparison, 15 federal states vs 9 neighbouring countries; ~79.1 M vs ~164.8 M inhabitants Pooled APC −1.8% (95% CI −2.3 to −1.4) Germany vs −2.2% (−2.8 to −1.6) controls; P = .42 (Hübner 2026, PMID 42268621)
Skin self-examination Case-control, 650 cases / 549 controls, Connecticut SSE practised by 15%; adjusted OR 0.66 (0.44–0.99) for melanoma incidence, OR 0.37 (0.16–0.84) for lethal melanoma — never replicated in a trial (Berwick 1996, PMID 8847720)
3D total-body photography + sequential digital dermoscopy RCT, 314 high-risk adults, Brisbane, 24 months Excisions per person 5.73 vs 3.99 (P = .02); fewer melanomas found in the intervention arm (24 vs 43) (Soyer 2025, PMID 40136310)
Same trial, economics Prespecified cost-effectiveness analysis Incremental cost US$945 per person (738–1,157) over 24 months; QALYs identical (1.84 both arms); US$40 per additional malignant tumour excised (Lindsay 2025, PMID 40136266)

No randomised trial has ever tested whether population melanoma screening reduces melanoma mortality. The USPSTF's 2016 statement rested on "a single fair-quality ecologic study with important methodological limitations" — SCREEN — and its 2023 update reached the same I statement (PMID 27458948; PMID 37071089). The one adjacent recommendation the USPSTF does make positively is behavioural counselling about UV exposure: B for ages 6 months to 24 years with fair skin, C for selective counselling above 24, and an I statement for counselling adults about skin self-examination (Grossman 2018, PMID 29558558).

Modelling fills the trial vacuum and returns opposing answers depending on assumed test performance. A Markov model of White men aged 50–75 found biennial screening from age 60 reduced melanoma mortality 20% at US$26,503/QALY and from age 50 reduced it 30% at $67,970/QALY — but results were sensitive to the assumed sensitivity gap between systematic screening (50%) and usual care (20%), a parameter with essentially no empirical basis (Adamson 2020, PMID 31705474). A whole-disease model in Veneto, Italy found one-time GP-based screening dominant over usual care, with the proportion of melanomas screen-detected and population adherence as key drivers (Buja 2020, PMID 31711359).

Detection tools do not settle the question

Dermoscopy improves accuracy over naked-eye inspection when performed by trained users: meta-analysis of clinical-setting studies found dermoscopy superior to naked-eye examination for diagnosing primary melanoma (Vestergaard 2008, PMID 18616769), and the Cochrane review of dermoscopy with and without visual inspection quantifies the same direction of effect (Dinnes 2018, PMID 30521682) — see clinical diagnosis and dermoscopy. But a more accurate test applied to a reservoir of indolent lesions increases the count of overdiagnosed cases; test accuracy and overdiagnosis are orthogonal problems. The European interdisciplinary guideline recommends sequential digital dermoscopy and whole-body photography specifically in high-risk patients rather than as a population strategy, and requires dermoscopy to confirm any clinical melanoma diagnosis (Garbe 2025, PMID 39700658).

Risk-targeted rather than population screening

Because population screening lacks trial support while high-risk surveillance has a plausible base rate, the practical frontier is risk stratification rather than universal examination. Modelling work has assessed risk-stratified surveillance in the UK (Wilson 2018, PMID 29909870), and a protocol exists for a microsimulation platform to evaluate risk-tailored melanoma screening (PMID 41452872). Stage-specific SEER analysis of 829,391 non-Hispanic White patients 2004–2021 found higher-SES counties had higher early-stage incidence with no corresponding reduction in late-stage incidence, leading the authors to argue early- and late-stage melanoma may behave as distinct diseases (Lal 2025, PMID 40637374). That finding is the sharpest available challenge to the stage-shift logic on which screening advocacy rests.

How to read this literature

  • Incidence and mortality are not two measurements of one thing. In melanoma the correlation between county-level incidence and county-level melanoma death is r = 0.09 (PMID 36190719). Any argument that moves from an incidence curve to a disease-burden claim must say why.
  • Stratify by thickness before interpreting a trend. In situ, thin invasive and thick invasive series behave differently in every registry examined, and pooling them hides the disagreement (PMID 29395684; PMID 39444324).
  • A screening program's detection yield is not its benefit. SCREEN raised invasive melanoma incidence by 34% during its 12 months; that number is a detection measure and says nothing about avoided death (PMID 22074699).
  • An ecological mortality decline after a program is weak evidence. The USPSTF said so explicitly of SCREEN in 2016 (PMID 27458948), and the national program that followed did not reproduce it (PMID 42268621).
  • Overdiagnosis estimates inherit the assumptions of their comparison series. Adamson's uses melanoma mortality in Black patients as the proxy for improvements in care, which assumes care improved comparably in both populations (PMID 35293957).
  • Reproducibility bounds everything. An estimate of overdiagnosis cannot be more precise than the diagnostic label it counts, and that label has 25–43% accuracy in the middle of the spectrum (PMID 28659278).

Harms that a mortality endpoint does not capture

Excisions are the dominant measurable harm. In the only randomised evaluation of intensified surveillance technology, adding 3D total-body photography and sequential digital dermoscopy raised excisions per person from 3.99 to 5.73 and produced 1,527 excisions across 314 participants of which 67 (4%) were melanoma and 402 (26%) keratinocyte carcinoma — the remaining 70% were benign (PMID 40136310). Cost followed: US$945 incremental per person over 24 months with no QALY difference (PMID 40136266). Beyond excision, the harm inventory the USPSTF names includes unnecessary biopsies with cosmetic or rarely functional consequences, and overdiagnosis with its downstream overtreatment (PMID 27458948). Overdiagnosis-specific harms — insurance and employment consequences, surveillance burden, and the psychological weight of a cancer label — are argued rather than measured in this literature (PMID 32841508; PMID 42279200).

Where the two literatures fail to meet

Global burden projections extrapolate 2020 incidence rates forward to 510,000 cases and 96,000 deaths by 2040 (Arnold 2022, PMID 35353115), and UVR-attributable modelling assigns ~88% of the 330,000 cases in 2022 to ultraviolet exposure with the steepest projected growth in low- and medium-HDI settings (Oh 2026, PMID 42502459). Both use the same incidence series that the overdiagnosis literature argues is inflated by detection. Neither literature has reconciled its estimate with the other's — see OQ-1.

Open questions

  • How much of the projected rise in melanoma incidence to 2040 is disease and how much is detection? The burden and overdiagnosis literatures use overlapping incidence series without reconciliation (PMID 35353115; PMID 35293957).
  • Can a molecular or morphometric marker separate indolent from progressive thin melanocytic lesions? Its absence is named as the root cause of both over- and underdiagnosis (PMID 32841508; PMID 28659278).
  • Would a randomised trial of risk-targeted screening in defined high-risk strata be feasible and ethical given the failure of population programs to show a mortality difference (PMID 42268621)?
  • Are early- and late-stage melanoma partly distinct diseases, as the failure of high early-stage incidence to reduce late-stage incidence suggests (PMID 40637374)?
  • Why did SCREEN's mortality signal not reproduce at national scale in the same health system — lower screening intensity, regression to the mean, or an artefact of the pilot's ecological design (PMID 22074699; PMID 30411137; PMID 42268621)?

References

  1. Bibbins-Domingo K, et al. Screening for Skin Cancer: US Preventive Services Task Force Recommendation Statement. JAMA. 2016;316:429-35. PMID 27458948
  2. Mangione CM, et al. Screening for Skin Cancer: US Preventive Services Task Force Recommendation Statement. JAMA. 2023;329:1290-1295. PMID 37071089
  3. 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
  4. Bjørch MF, et al. Overdiagnosis in malignant melanoma: a scoping review. BMJ evidence-based medicine. 2024;29:17-28. PMID 37793786
  5. Hübner J, et al. Population Skin Cancer Screening and Melanoma Mortality Rates. JAMA dermatology. 2026;162:775-779. PMID 42268621
  6. Thrift AP, et al. Melanoma Incidence Among Non-Hispanic Whites in All 50 US States From 2001 Through 2015. Journal of the National Cancer Institute. 2020;112:533-539. PMID 31346623
  7. Sacchetto L, et al. Trends in incidence of thick, thin and in situ melanoma in Europe. European journal of cancer (Oxford, England : 1990). 2018;92:108-118. PMID 29395684
  8. Huang J, et al. Global Incidence, Mortality, Risk Factors and Trends of Melanoma: A Systematic Analysis of Registries. American journal of clinical dermatology. 2023;24:965-975. PMID 37296344
  9. De Pinto G, et al. Global trends in cutaneous malignant melanoma incidence and mortality. Melanoma research. 2024;34:265-275. PMID 38391175
  10. Welch HG, et al. Overdiagnosis in cancer. Journal of the National Cancer Institute. 2010;102:605-13. PMID 20413742
  11. Welch HG, et al. Skin biopsy rates and incidence of melanoma: population based ecological study. BMJ (Clinical research ed.). 2005;331:481. PMID 16081427
  12. Adamson AS, et al. Association of UV Radiation Exposure, Diagnostic Scrutiny, and Melanoma Incidence in US Counties. JAMA internal medicine. 2022;182:1181-9. PMID 36190719
  13. Olsen CM. Ecological evidence for melanoma overdiagnosis in Denmark. The British journal of dermatology. 2024;191:318-319. PMID 38752346
  14. Greco ME, et al. Overdiagnosis of Melanoma In Situ. Journal of clinical medicine. 2026;15. PMID 42279200
  15. Paiva ML, et al. Melanoma Overdiagnosis Is Not the Key Problem. JAMA dermatology. 2022;158:1082-1083. PMID 35895044
  16. Adamson AS, et al. Melanoma Overdiagnosis Is Not the Key Problem-Reply. JAMA dermatology. 2022;158:1083-1084. PMID 35895049
  17. Clarke CA, et al. Continued Increase in Melanoma Incidence across all Socioeconomic Status Groups in California, 1998-2012. The Journal of investigative dermatology. 2017;137:2282-2290. PMID 28736233
  18. Reinhart JP, et al. Incidence and mortality trends of primary cutaneous melanoma: A 50-year Rochester Epidemiologic Project study. JAAD international. 2024;16:144-154. PMID 38957842
  19. DE Gruijl FR, et al. Cutaneous Melanoma: Sheep in Wolves Clothing?. Anticancer research. 2022;42:5021-5025. PMID 36191966
  20. Thomas S, et al. Declining invasive and rising in situ melanoma incidence trends in Iceland: A nationwide cohort study. Journal of the European Academy of Dermatology and Venereology : JEADV. 2025;39:1278-1284. PMID 39444324
  21. Elmore JG, et al. Pathologists' diagnosis of invasive melanoma and melanocytic proliferations: observer accuracy and reproducibility study. BMJ (Clinical research ed.). 2017;357:j2813. PMID 28659278
  22. Piepkorn MW, et al. Assessment of Second-Opinion Strategies for Diagnoses of Cutaneous Melanocytic Lesions. JAMA network open. 2019;2:e1912597. PMID 31603483
  23. Kutzner H, et al. Overdiagnosis of melanoma - causes, consequences and solutions. Journal der Deutschen Dermatologischen Gesellschaft = Journal of the German Society of Dermatology : JDDG. 2020;18:1236-1243. PMID 32841508
  24. Breitbart EW, et al. Systematic skin cancer screening in Northern Germany. Journal of the American Academy of Dermatology. 2012;66:201-11. PMID 22074699
  25. Hübner J, et al. [Skin cancer screening in Germany: review after ten years]. Bundesgesundheitsblatt, Gesundheitsforschung, Gesundheitsschutz. 2018;61:1536-1543. PMID 30411137
  26. Berwick M, et al. Screening for cutaneous melanoma by skin self-examination. Journal of the National Cancer Institute. 1996;88:17-23. PMID 8847720
  27. Soyer HP, et al. 3D Total-Body Photography in Patients at High Risk for Melanoma: A Randomized Clinical Trial. JAMA dermatology. 2025;161:472-481. PMID 40136310
  28. Lindsay D, et al. Cost-Effectiveness Analysis of 3D Total-Body Photography for People at High Risk of Melanoma. JAMA dermatology. 2025;161:482-489. PMID 40136266
  29. Grossman DC, et al. Behavioral Counseling to Prevent Skin Cancer: US Preventive Services Task Force Recommendation Statement. JAMA. 2018;319:1134-1142. PMID 29558558
  30. Adamson AS, et al. Screening for Melanoma in Men: a Cost-Effectiveness Analysis. Journal of general internal medicine. 2020;35:1175-1181. PMID 31705474
  31. Buja A, et al. Cost-effectiveness of a melanoma screening programme using whole disease modelling. Journal of medical screening. 2020;27:157-167. PMID 31711359
  32. Vestergaard ME, et al. Dermoscopy compared with naked eye examination for the diagnosis of primary melanoma: a meta-analysis of studies performed in a clinical setting. The British journal of dermatology. 2008;159:669-76. PMID 18616769
  33. Dinnes J, et al. Dermoscopy, with and without visual inspection, for diagnosing melanoma in adults. The Cochrane database of systematic reviews. 2018;12:CD011902. PMID 30521682
  34. Garbe C, et al. European consensus-based interdisciplinary guideline for melanoma. Part 1: Diagnostics - Update 2024. European journal of cancer (Oxford, England : 1990). 2025;215:115152. PMID 39700658
  35. Wilson ECF, et al. A Modeling Study of the Cost-Effectiveness of a Risk-Stratified Surveillance Program for Melanoma in the United Kingdom. Value in health : the journal of the International Society for Pharmacoeconomics and Outcomes Research. 2018;21:658-668. PMID 29909870
  36. McLoughlin KG, et al. A protocol for development of a microsimulation model platform to evaluate the potential benefits, harms, and cost-effectiveness of risk-tailored melanoma screening. PloS one. 2025;20:e0339177. PMID 41452872
  37. Lal T, et al. Socioeconomic Influences on Melanoma Incidence Patterns by Stage in the US. Journal of surgical oncology. 2025;132:465-472. PMID 40637374
  38. Arnold M, et al. Global Burden of Cutaneous Melanoma in 2020 and Projections to 2040. JAMA dermatology. 2022;158:495-503. PMID 35353115
  39. Oh J, et al. Global patterns of melanoma, burden attributable to ultraviolet radiation exposure, and projections to 2050 across 185 countries: A population-based study with global modeling. JAAD international. 2026;27:149-159. PMID 42502459