NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Henrikson NB, Ivlev I, Blasi PR, et al. Screening for Skin Cancer: An Evidence Update for the U.S. Preventive Services Task Force [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2023 Apr. (Evidence Synthesis, No. 225.)
Screening for Skin Cancer: An Evidence Update for the U.S. Preventive Services Task Force [Internet].
Show detailsLiterature Search
We reviewed 20,320 abstracts and assessed 522 full-text articles for inclusion (Appendix A Figure 1). For KQ1, we included three studies (10 articles);104, 120–128 for KQ2, we included six studies (7 articles);129–135 KQ3 was comprised of two studies (3 articles);136–138 and KQ4 included nine studies (9 articles).139–147
Seventeen studies were newly identified in this update.121, 124, 130, 132–135, 138–147
A list of included studies and a list of excluded studies with reasons for exclusion are available in Appendix C and Appendix D, respectively. We determined all included studies were of fair or good quality (Appendix A Table 2).
KQ1. What Is the Effectiveness of Routine Skin Cancer Screening With Visual Skin Examination by Clinicians in Reducing Skin Cancer Morbidity and Mortality or All-Cause Mortality? Does the Effectiveness of Screening Vary by Subgroups (e.g., Age, Sex, Skin Type, Race/Ethnicity, Socioeconomic Status, or UV Exposure)?
Summary of Results
Based on ecologic data and non-randomized study data from three evaluations of two related screening programs in Germany, the included evidence (both screening programs rated fair quality) does not suggest a melanoma mortality benefit at the population level over 4 to 10 years’ followup after screening (n=NR in one study; 1,791,615 in the other two). At ten-year followup of the Skin Cancer Research to Provide Evidence for Effectiveness of Screening in Northern Germany (SCREEN) pilot, melanoma mortality in the SCREEN region compared to the rest of Germany does not support an ongoing mortality benefit to routine skin cancer screening.125 Similarly, at 5-year followup of the German National Screening Program, no mortality benefit was observed based on national population statistics. Rather, increases in mortality, not decreases, were observed for multiple European countries, including Germany.124
One non-randomized study (n=1,431,327) of melanoma mortality in individuals with documented skin cancer screening provided through German statutory health insurance found an absolute decrease in mortality suggesting a screening benefit at four-year followup, but this difference was attenuated on multivariate analysis and adjustment for lead time bias.121 No included studies reported all-cause mortality, SCC mortality, BCC mortality, or skin cancer morbidity.
Three included publications from the two German screening programs reported age- and sex-specific melanoma mortality. Similar to the results reported in the main study populations, across these analyses, there was no evidence of a melanoma mortality benefit to screening at the population level.
Detailed Results
Overview of Included Studies
Publications using data from two population-based screening programs in Germany met inclusion criteria (Tables 3 and 4). The first fair-quality study, also included in the previous review, was the SCREEN skin cancer screening pilot, which was conducted in the Schleswig-Holstein state in northern Germany from 2003–2004 (n=360,288 screened). Ten-year followup data on melanoma mortality from this program is included in this report.125 The second screening program is the German national skin cancer screening program, implemented in 2008, which includes skin cancer screening covered through statutory health insurance following the promising 5-year mortality data observed in the SCREEN program.126 One evaluation used a nonrandomized design (n=1,431,327) to examine administrative data from a German health insurance company with documented skin cancer screening as provided through the German skin cancer screening program.121 Another evaluation (2 publications) reporting melanoma mortality rates in Germany and surrounding countries during the first 5 years of the screening program, which included approximately 3 million screened individuals, also met inclusion criteria. 120, 124 All but one included publication used ecologic analyses to examine melanoma mortality rates in screened geographic areas compared to areas without population screening programs.
Table 3.
Study Characteristics of Included Studies (KQs 1–3).
Table 4.
Intervention Characteristics, Included Studies (KQs 1–3).
Subgroup data (KQ1a) on melanoma mortality by age and sex is provided for both Germany and for the SCREEN program. One study analyzed the annual percent change (APC) in melanoma mortality by sex and age in Germany compared to eight other European countries in order to estimate the impact of the German skin cancer screening program on melanoma mortality by age and sex subgroups.120
Summary of Included Screening Programs
SCREEN (Germany) (2003–2004)104, 122, 125, 126, 128
The SCREEN study was conducted to determine the feasibility of a population-based skin cancer screening program in the German primary care health system104 (Table 4). In 2001, pilot intervention activities occurred on a small scale with 200 physicians and 6,000 screened patients in the Schleswig-Holstein state of northern Germany. Between 2003 and 2004, following the pilot, the SCREEN project implemented population-based skin cancer screening in Schleswig-Holstein. All residents aged 20 years or more and insured with national statutory health insurance were eligible to participate. The screening program included three main components:
- Provider education and training. In 2003, nondermatologists (general practitioners in primary care, obstetricians and gynecologists, and urologists; n=1,673) and dermatologists (n=116) participated in an 8-hour training course focused on detecting skin cancer. Content included training in the epidemiology and etiology of skin cancer, training and practice in standardized whole body visual examination, strategies for actively recruiting patients for screening, and program documentation and referral procedures.104 Training participation rates were 64 percent for nondermatology providers and 98 percent for dermatologists in the region.
- Public outreach. An outreach campaign encouraged residents of Schleswig-Holstein aged 20 years and older to seek skin cancer screening by a nondermatology physician through the program. Communication channels included health insurers, physicians, and print, digital, and telephone mass media campaigns.
- Clinician skin exam. Screening exams were conducted from July 2003 to June 2004 via whole-body visual skin exam conducted by a nondermatology provider, although dermatologists also participated in the program and conducted initial screenings. Suspicious lesions were either referred to dermatology or handled by the screening dermatologist. Physicians were reimbursed about $20 USD per screening exam. All tentative clinical diagnoses were followed by biopsy and histopathologic evaluation. Approximately three quarters (77.4%) of screening exams were conducted by nondermatology providers, and 22.6 percent were conducted by dermatologists. Among the 73,710 individuals referred to dermatology after screening by nondermatology providers, 36.8 percent were lost to followup and did not see a dermatology provider for a second clinical exam.104 All confirmed skin cancers were reported to the state tumor registry.
German National Screening Program (2008–2013)120, 124, 135, 148, 149
Following the completion of the SCREEN study,126 Germany implemented a nationwide routine skin cancer screening covered by statutory health insurance (Table 4). The implementation was not designed as a research study and did not include a comparison group; though a 5-year evaluation was required.149 Screening included a total skin exam by either a participating primary care clinician or a participating dermatologist. Participating providers completed a standardized 8-hour training program similar to that used in the SCREEN study. Topics included benefits and risks of early screening; etiology of skin cancer and risk factors; training in implementing the screening program; patient communication; and discussion of case examples.123, 150 Screening included a visual examination of all visible skin and mucous membranes (oral, vaginal, anal) and was offered free of charge to all enrollees aged 35 years or older. Screening physicians were reimbursed for screenings.149 During 2008–2009, screenings were conducted approximately equally by general practitioners or dermatologists.148
In Germany, health insurance is mandatory, and the administration of health insurance is implemented by non-governmental insurance companies.151 Documentation of clinician skin cancer screening, a reimbursable service for clinicians, is noted in individual health records through billing codes.
Population Summary
SCREEN Program
Of a total population of 2.8 million in the Schleswig-Holstein region, 1.9 million individuals aged 20 years and older comprised the eligible screening population (Table 5). During the project period, 360,288 people received clinician visual skin exams, representing 19.1 percent of the eligible population in the region. Screening participation rates varied by age, with 20 to 22 percent of adults aged 35 to 69 years participating in screening compared with 14.9 percent of adults over age 70 years.
Table 5
Population Characteristics, Included Studies (KQs 1–4).
Among screened participants, the mean age was 49.7 years. Almost three-quarters of screened participants were females (73.6%). Nearly half of participants were judged to have at least one risk factor for melanoma based on family history of melanoma (6.1%), personal history of melanoma (1642 people, 2.6%), presence of atypical nevi (51.9%), multiple melanocytic nevi (56.2%) or congenital moles (20.9%). Other risk factors included UV-damaged skin (72.0%); actinic keratosis (31.2%), personal history of KC (16.9%); X-ray damaged skin (2.6%) or immunosuppression (2.7%).104 Data on other specific population subgroups (e.g., race and/or ethnicity) was not reported.
German National Screening Program
The total number of screened residents from the study period 2008–2013 is not reported in the included studies. However, an earlier publication describing the German program estimated that the population enrolled in the German health insurance plan DAK-Gesundheit was approximately 6.1 million members in an 18-month period in 2008–2009; in that time period approximately 920,000 people received screening.148
Characteristics of the screened population were not consistently reported. The German population during the screening period 2008–2012 was 51.2 percent female with a mean age of 43 years (including all ages) (Table 5).135 During that period, the mean quarterly screening participation was 30.8 percent, with screening participation significantly higher in females (age-and federal state- adjusted participation rates were 29.7% for males and 31.9% for females).148 Screening participation was highest in people over age 65, with state-adjusted rates among males between the ages 65 and 79 exceeding 40 percent.148
The good quality nonrandomized study by Datzmann and colleagues included enrollees in AOK PLUS, a health insurance plan that administers German national statutory health insurance and insures approximately 25 million people.121 The included sample was enrollees 35 years or older between 2010–2016 (n=1,431,327). The sample was mean age 63.9 years and 55.7% female. Neither race and ethnicity nor skin type was reported. Overall the screened and unscreened groups were similar, except for receipt of systemic therapy within 30 days (11.6% in screened vs 21.8% in unscreened group), suggesting a stage shift associated with screening. The study team identified all prevalent cases of melanoma during the study period (n=4,552) and limited their analysis to incident melanoma cases diagnosed during 2013–2016 with no evidence of melanoma in the previous 3 years (n=2475). People with documented skin cancer screening as identified through billing codes in the previous two years before diagnosis were considered to have received the screening intervention. The observation period was four years.
The outcome of interest was melanoma mortality. The team measured both the absolute risk difference as a difference in numbers of deaths during the observation period. They also conducted multivariable modeling and estimated both unadjusted and adjusted hazard ratios. Adjustment variables included age, sex, comorbidity, health-seeking behavior (estimated by receipt of flu vaccine), personal history of melanoma; and approximated stage categories of documented metastasis or receipt of systemic anticancer therapy. They also conducted sensitivity analyses to assess the potential for lead time bias by removing patients with survival of less than 360 days. Lead time bias associated with screening is when survival can appear longer because of detection, not because of intervention.152
Quality
For this key question, two included studies were rated fair quality and the Datzmann study was rated as good quality. Followup time was sufficient to observe skin cancer incidence and mortality over time, screening procedures were well described, populations were clearly defined, sources of outcome data were strong (use of national and regional health statistics), and sample sizes were large.
Quality concerns were primarily related to the known limitations of ecologic studies,153–155 including the lack of controlled intervention (e.g., selection bias with voluntary screening programs) and the lack of individual-level data on screening participants, including–importantly—previous screening participation and the presence of skin cancer risk factors. There is adequate biologic plausibility for an effect of screening to be observed at the population level, and the included studies’ followup times are sufficient to observe melanoma mortality. However, the included studies can neither directly compare individual-level changes in mortality among people receiving versus not receiving skin cancer screening, nor account for confounding through randomization or adjustment for environmental factors or socioeconomic, risk factor, or demographic variables. As such, our ability to infer a causal relationship between skin cancer screening and melanoma mortality is limited and should be viewed cautiously.
Another potential quality concern has been raised in a related study, where the authors examined the cause of death statistics for Schleswig-Holstein during the years of the SCREEN program evaluation (2007–2010), which overlaps with the years the mortality benefit was observed (2004–2009).126 They concluded that potential for misclassification existed during that time period, when an unusually high number of deaths were coded as “malignancies of ill-defined, secondary, and unspecified sites” (ICD-10 codes C76–C80) and an unusually low number of deaths were coded as due to malignant melanoma. They argue that this change would be sufficient to explain the mortality benefit observed in those years.156
Detailed Results by Outcome
No included studies reported all-cause mortality, SCC mortality, or BCC mortality.
Melanoma Mortality in the SCREEN Program
During the SCREEN study period from 2003 to 2004, 1,169 incident melanoma cases were reported to the state cancer registry, 585 of which were detected via the SCREEN study.104 Of these 585, 31 percent were melanoma in situ and 69 percent were invasive melanoma. The SCREEN study also detected 1,961 basal cell carcinomas, 392 SCC and 165 other skin cancers.
Incidence of melanoma, BCC, SCC, and melanoma in situ increased during the SCREEN pilot (Figures 3 and 4). Melanoma age-adjusted incidence rates (per 100,000 individuals) increased 27 percent from 14.2 (95% CI, 13.3 to 15.1) during 2001–2003 to 18.0 (95% CI, 16.6 to 19.4) during the active screening period (2003–2004). Similarly, melanoma in situ age-adjusted incidence rates (per 100,000 individuals) increased 48 percent between the pre-screening period versus during-screening period. SCC age-adjusted incidence rates (per 100,000 individuals) in the pre-screening period versus during-screening period were 11.2 (95% CI, 10.6 to 11.8) and 12.9 (95% CI, 12.0 to 13.8), respectively, a 15 percent increase. BCC age-adjusted incidence rates (per 100,000) increased 29 percent in the prescreening period versus during-screening period from 60.5 (95% CI, 59.0 to 62.1) to 78.4 (95% CI, 75.9 to 80.8).104
Figure 3
Incidence of BCC, SCC, and Melanoma Before and After Implementation of National Screening Programs, Germany.
Figure 4
Age-Standardized Annual Incidence for Melanoma and Keratinocyte Skin Cancer Before and After the Initiation of the German National Screening Program.
In the previous evidence review,106 included data from the SCREEN study126 suggested a 49 percent mortality reduction in the screening region compared to the surrounding regions at 5 years of followup from the end of the program (2003–2004 program; evaluation through 2009). However, updated data with longer followup to 2013 suggests that the mortality improvement previously reported appears to attenuate over time (Table 6, Figure 5). It should be noted that Germany as a whole had a fairly stable melanoma mortality rate between 1998 and 2010 (between 1.9 and 2.1 per 100,000), with a marginal increase between 2011 and 2013 (2.2 to 2.3 per 100,000). The SCREEN region’s age-standardized mortality rate fluctuated more: higher than Germany's overall rate in the years preceding the SCREEN program (1998–2002); similar to Germany’s overall rate during the SCREEN program (2003–2004); and decreasing to below the German rate around 2008–2010 when compared to Germany as a whole.125 An additional analysis conducted of observed versus expected melanoma mortality in the SCREEN region (state of Schleswig-Holstein, northern Germany) compared to the state of Saarland (western Germany) from 2003–2008 found that observed melanoma deaths in the SCREEN region were lower than would be expected (age- and sex- adjusted standardized mortality rate 0.59, 95% CI, 0.40 to 0.83) (Table 7).157 This is consistent with the 49 percent mortality rate reported in earlier publications.126
Figure 5
Melanoma Mortality, Overall Population, Schleswig-Holstein vs. Germany, 1998-2013 (In table form in (Table 6). *Period of active screening
Table 6
Age-Standardized Melanoma Mortality per 100,000, SCREEN Study Region and Germany, 1998–2013, Total and by Sex (KQ1, KQ1a, depicted graphically also in Figures 5 and 6).
Table 7
Observed vs. Expected Mortality, SCREEN Study Region and Saarland Region, Germany, 2003–2008 (KQ1).
Melanoma Mortality in the German Screening Program
Based on German incidence rates used as the comparator in the 10-year followup of the SCREEN program, age-standardized melanoma incidence rates increased markedly at the time when the German national program was introduced. Age-standardized melanoma incidence rates increased from between 14.0 and 14.9 per 100,000 during 2003–2007, to between 17.7 and 18.2 per 100,000 during 2008–2011.
An evaluation of the German national program examined change in incident melanoma diagnoses based on hospital discharges. The multivariable fixed effects model suggested an association of the German program with increased diagnoses in the unadjusted model (coefficient 0.276, SE 0.02, p<0.001). Although this effect was attenuated with the addition of covariates (coefficient 0.181, SE 0.02, p<0.001), an independent association of the screening program with new diagnoses appeared to remain. This finding suggests that the German program was effective in increasing new skin cancer diagnoses after 2008, while not in the surrounding European countries.124
Two included studies reported data on melanoma mortality in Germany related to the implementation of national skin cancer screening.120, 121, 125 The first, a study by Datzmann and colleagues, observed a total of 325 melanoma deaths during the 4-year observation period, and found a higher proportion of melanoma deaths in the unscreened group compared to the screened group (171 deaths, 9.5% of the screened group; 154 deaths, 22.8% of the unscreened group; unadjusted HR 0.37, p<0.05) (Table 8).121 On adjusted analyses, the association was attenuated but remained statistically significant (adjHR 0.62, p<0.05). The sensitivity analyses to assess potential lead time bias were similarly attenuated on both unadjusted (HR 0.50, p<0.05) and adjusted estimates (adjHR 0.75, NS).
Table 8
Melanoma Mortality for First-Onset Melanoma 2013–2016 Among Screened Enrollees in AOK Plus Health Insurer in Saxony, Germany.
A separate analysis of melanoma mortality from 2000 to 2012 between Germany and 22 other European countries found that the unadjusted mean annual melanoma mortality rate per 100,000 increased, not decreased, between the time period 2000–2007 (before the German national skin cancer screening program began in 2008) and 2008–2012 (Table 9). Although point estimates were not reported, similar increasing melanoma mortality trends were observed in many of the other European countries.124 This evidence suggests that there is no observable benefit to national skin cancer screening.
Table 9.
Melanoma Mortality Overall and by Sex and Age Group; Germany and Surrounding Countries, 1980–2012.
Adjusted mortality rates were not provided, but the authors fit a multivariable fixed effects model using a hypothetical control group comprised of the included 22 European countries (excluding Germany) to estimate the screening program impact on melanoma mortality (Table 10). This model included population-level demographic variables (age, sex, physician density, education), and accounts for the years in Germany in which the SCREEN program took place (2003–2004) using dummy variables.124 There was an association between mortality and the German program in the unadjusted model (coefficient 0.242, p<0.01), but with the addition of covariates there was no significant independent relationship between the screening program and mortality (coefficient 0.077, not significant). The authors concluded there was no significant overall effect of the German skin cancer screening program on melanoma mortality.124
Table 10
Unadjusted Melanoma Mortality Rate per 100,000 Before and After Implementation of National Skin Cancer Screening, Germany, 2000–2012 (KQ1).
Specific Population Results (KQ1a)
Three included publications from the two German screening programs reported age- and sex-specific melanoma mortality.120, 124, 125 Data on race and/or ethnicity-specific or other specific populations was not reported.
In the SCREEN study, mortality rates were lower in females than males throughout the followup period (Figure 6). Female mortality rates for Schleswig-Holstein dipped below that of Germany in 2005 (the year after the SCREEN pilot ended) and then again in 2007–2010. No statistical tests of significance were reported. Female mortality rates appeared fairly steady, fluctuating between 0.9 and 2.2 melanoma deaths per 100,000 for Schleswig-Holstein and 1.5 and 1.7 per 100,000 for Germany ((Table 6).125 Similar to the trend among females, for males the mortality rate was fairly steady, varying between 2.3 and 3 per 100,000 for Germany. The rate for Schleswig-Holstein males varied the most, with one dip in mortality in 2003 (the first year of the SCREEN program) and the lowest mortality observed in 2008 (1.1 per 100,000), then increasing steadily thereafter. Mortality rates increased in 2005 (just at the end of the SCREEN program) and appeared to be increasing toward the last 3 years of the followup period, approaching that of Germany as a whole (Table 6).125
Figure 6
Melanoma Mortality, Males and Females, Schleswig-Holstein vs. Germany, 1998-2013 (In table form in (Table 6). *Period of active screening
The German National Screening Program publications reported findings on the impact of age on melanoma mortality. Although the fixed effects multivariable model in the Kaiser and colleagues study’s model did not support an overall impact of the screening program, the proportion of the population over the age of 65 years did appear independently linked to improved mortality (coefficient 0.094, SE 0.03, p<0.001) (data not shown).124
The fluctuation in melanoma mortality reported in the SCREEN program appeared primarily in people aged 65 years or over (Table 11). Melanoma mortality rates were highest in both males and females in this age group, in both the SCREEN region and in Germany as a whole (Figures 7 and 8). In females aged 65 years or over, the mortality rate in the SCREEN region was highest in 2001 (13.2 per 100,000, two years before the SCREEN program began) and lowest in 2008–2009 (3.8 per 100,000; five years after the SCREEN program), increasing thereafter.125 Similarly to females, in males aged 65 years or older, the mortality rate in the SCREEN region was also highest in 2001 (17.7 per 100,000; two years before the SCREEN program began), decreased to its lowest point in 2009 (5.7 per 100,000, 5 years after the end of the SCREEN program), then steadily increased to approach the German national mortality rate by the end of the followup period.125 Additionally, in both males and females, melanoma mortality was similar in both the SCREEN region and in Germany by the end of the followup period in all age groups.125
Figure 7
Melanoma Mortality, Males by Age Group, Schleswig-Holstein vs. Germany, 1998-2013 (In table form in Table 11). *Period of active screening
Figure 8
Melanoma Mortality, Females by Age Group, Schleswig-Holstein vs. Germany, 1998-2013 (In table form in Table 11). * Period of active screening
Table 11
Unadjusted Melanoma Mortality Rates per 100,000, Germany and Schleswig-Holstein, by Age Group and Sex (KQ1a depicted graphically also in Figures 7 and 8).
In the analysis of the annual percent change in melanoma mortality in Germany compared to eight other European countries, there was little evidence of differential annual percent change in melanoma mortality in either males or females that could be attributable to the German program (Table 9).120 Overall, in males, all countries except the Czech Republic saw a slight annual increase in melanoma mortality between 1980 and 2012. However, for the period 2008–2012 (the German program began 2008), the annual percent change was flat for all countries, including Germany.120 For females in Germany, there was no significant annual percentage change for either the entire period 1980–2012 or the screening program period 2008–2012. Other countries saw more annual fluctuations in the percent change in female melanoma mortality over the longer observation period. The annual percent change among females remained flat for all countries, including Germany, for 2008–2012.120
Annual percent change in mortality data were only reported for both sex and age for the entire period 1980–2012. These data suggest that several countries, including Germany, observed a significant increase in melanoma mortality in males aged 60–74 years (Germany 1.4 APC, 95% CI, 0.6 to 2.1) and 75 years and older (Germany's was 1.6 APC, 95% CI, 1.1 to 2.1) compared to males under age 60 years. Germany's annual percentage change increase was smaller in the two age groups of males aged 60–74 years and 75 years or more compared to other countries observing a significant increase in the annual percent change.120 For females, a similar pattern was observed (Germany: age 60–74 APC −0.4, 95% CI −0.8 to 0; age 75+ APC 1.0, 95% CI 0.8 to 1.3), although the annual percent change increases were smaller than for males.120
KQ2. Does Routine Skin Cancer Screening Lead to Higher Rates of Detection of Precancerous Lesions or Earlier Stage Skin Cancer Compared to Usual Care (e.g., Lesion-Directed Skin Examination)? Do Rates of Earlier Skin Cancer Detection Vary by Subgroups (e.g., Age, Sex, Skin Type, Race/Ethnicity, Socioeconomic Status, or UV Exposure)?
Summary of Results
Based on nonrandomized observational data from four evaluations of three skin cancer screening programs (all fair-quality, n=2,344,210),130, 132, 133, 135 and one good-quality physician-focused skin cancer examination initiative (n=595,799),134 routine clinician skin examination does not appear to be associated with increased detection of KC, melanoma, or skin cancer precursor lesions compared to usual care or lesion-directed examination.
Three studies reporting heterogeneous categories of stage at detection suggested a similar lack of association between screening and stage at melanoma detection. Routine clinician skin examination was not associated with earlier detection in two studies, one using AJCC stage categories of melanoma in situ, stage I/II, and stage III/IV133 and the other using SEER-comparable stages (presence of lymph node and distant metastasis).135 However, in another study, the distribution of melanoma in situ at detection favored the screened group.134
There is mixed evidence on the association between clinician skin examination and lesion thickness at detection based on three nonrandomized studies reporting clinically relevant lesion thickness categories, with one study suggesting no association with lesions <1mm at detection but some association for lesions <2mm at detection,130 and one study finding association for lesions <1mm only and not for lesions thicker than 1mm.134 The third, a case-control study, found higher odds of having received a clinician skin examination in people with melanomas detected at <0.75mm thickness compared to unaffected controls, but not for thicker lesions; though people with lesions ≥3.00mm had significantly lower odds of having received an examination compared to controls.129
Detailed Results
Overview of Included Studies
Per our inclusion criteria, all studies included evaluations of visual skin examinations conducted by primary care physicians or dermatologists and compared precursor lesion detection or stage at skin cancer detection between groups receiving either routine skin cancer screening or usual care.
Six nonrandomized studies with data on approximately 2.9 million individuals and 53,329 skin cancer or precursor lesions met inclusion criteria (Tables 3–5),129–133, 135 one of which was carried forward from the previous review.129 One of the six studies was conducted in the United States.134 Study populations ranged from 497 to 34,295 skin cancer or precursor lesion cases. The outcomes assessed included precursor lesions (2 studies),130, 132 stage at melanoma detection (3 studies),133–135 and stage at KC detection (1 study).133 Three studies reported thickness at melanoma detection,129, 130, 134 and one study reported the odds of having received a clinical skin examination in people with and without skin cancer.129
We included four studies that reported analyses of three skin cancer examination programs that included outreach to patients. These three programs varied in their implementation and comparison group composition (Tables 3 and 4). Two included publications used data from the German National Skin Cancer Screening Program to compare people with skin cancer who had documented skin cancer screening to those without documented screening.133, 135 One study compared skin cancer detection in people who had participated in a community-based screening program conducted in Trento, Italy, in 2001–2004 to skin cancer detection in the general population of the same city through 2013.130 Lastly, a study conducted in Belgium compared skin cancer detection in two communities where different public outreach strategies were used for single 4- to 5-day screening events: in one community, people were invited to receive whole body examination; in the other community, people were invited to have suspicious skin lesions examined.132 Screening participation rates were reported for two studies and were overall quite low: 12.4 percent in the German screening program;135 and 17.9 percent in the total body screening group in Belgium compared to 3.3 percent for the lesion directed screening group in the Belgian study.132
The intervention in the U.S.-based study was a physician-focused decision support intervention and did not include direct outreach to patients.134 In this quality improvement initiative in academic primary care clinics, full-body skin examination was added to the list of preventive care recommendations in the electronic health record of patients aged 35 years or older. Primary care physicians were offered training in diagnosing skin cancer and encouraged to participate in the initiative by medical center leadership.131, 134
Finally, a case-control study conducted in Queensland, Australia, identified cases among those with incident melanoma (n=3,762) and matched unaffected controls randomly selected through electoral rolls (n=3,824). The authors measured the association between self-reported whole-body physician skin examination during the three years before either the melanoma diagnosis for cases or referent date for controls and assessed the odds of having received a clinician skin exam within strata of melanoma lesion thickness at diagnosis.129
Population Summary
Screening population sizes were not always reported, but when reported they ranged from 1,328 to 533,393 in screened groups and 248 to 1,489,074 for unscreened/usual care groups (Table 5).
The largest population was for the German National Screening Program,135 and the smallest was for the Belgian study.132 In total, 53,329 detected skin cancer or precursor lesions were reported in this included body of evidence (n=11,182 melanoma cases; 41,686 KC cases, and 461 precursor lesion cases).
Overall, the intended screening populations were broadly defined adult populations (Table 5). The Italian study also included adolescents aged 15 years and older (mean age 40.2 years, range 15–84 years).130 The screening populations were majority female, except for the Australian case-control study,129 which used age- and sex-matching to select control participants. The sex distribution of the case group was not reported, whereas the control group was 57 percent male.129 Only the U.S.-based study reported race and/or ethnicity for the screened group (88.4% White).134
Only the Belgian study reported measures of socioeconomic status, reporting the highest education level attained for both study groups.132 There was a statistically significant different distribution of educational categories between groups, with the total body exam (screening) group trending slightly toward a lower educational level than the lesion-directed screening group (45.8% with no more than a high school education in the total body exam group compared to 38.9% in the lesion-directed screening group; p<0.01 for trend).132
Only one study reported skin cancer risk factors for screening populations.132 In the Belgian study, distributions of Fitzpatrick skin type and nevus count were similar between the study groups (e.g., nevus counts of less than 25 were observed in 57.1% of the total body exam screening group and 58.0% of the lesion-directed exam group; p=0.96 for trend).132 Family history of skin cancer was also reported at similar frequencies between the two groups (10.7% of total body exam screening group and 13.1% of lesion directed exam group, p=0.17).132 Additionally, personal history of skin cancer was reported in similar small proportions of both study groups (2.4% in the total body exam screening group and 2.0% in the lesion directed exam group; p=0.84).
Demographic characteristics of detected cases were reported for three studies.130, 134, 135 Two of these reported age and sex distributions of detected cases,130, 135 and one noted that 99.2% of melanomas were diagnosed in patients who identified their race as Non-Hispanic White.134 Risk factor characteristics were not reported for any detected case populations.
Quality
The single U.S.-based study was rated as good quality,134 and the five other nonrandomized studies were rated as fair quality.129, 130, 132, 133, 135 Quality concerns included unclear reporting of demographic and risk factors of study groups, unclear reporting of completeness of data and outcome assessment methods, handling of missing data, and the use of self-reported data for outcomes. In addition, one of the publications using German national screening data used a previously-published algorithm for estimating stage at melanoma detection from claims data, rather than direct observation of medical records, which could introduce misclassification errors.133 Staging conventions, stage categories, and melanoma thickness categories were heterogeneous across studies. None of the studies reported adjusting for lead time bias when reporting KQ2-relevant outcomes.
In the U.S.-based study, screened and unscreened groups were overall similar in age, sex, and insurance status, but some statistically significant differences were observed between screened and unscreened populations. The screened population was reported to be somewhat older (median age 60 years vs. 57 years, p<0.001), and race and/or ethnicity differed between groups (e.g., 6.8% of the screened population was Black, compared to 7.2% of the unscreened population, p<0.001).134 Further, insurance coverage appeared more generous in the screened population compared with the unscreened population (3.0% Medicaid in the screened group compared to 6.3% in the unscreened group; p<0.001). Outcomes, however, were adjusted for age, sex, and insurance status.134
Detailed Results
Melanoma
Melanoma detection rates. Five included publications reported data on overall melanoma detection rates associated with routine screening in three European skin cancer screening programs and in one United States health system-based skin cancer screening initiative (Table 12).130, 132–135 Across all three European programs, overall melanoma detection rates were similar between screened populations compared to usual care or lesion-directed examination populations. One evaluation of the German program using AOK PLUS data from 2005–2012 found melanoma detection rates were similar in the group receiving routine screening (0.31% case detection rate) compared to those receiving usual care (0.13% case detection rate).135 A separate evaluation of the German program reported numbers of skin cancers detected but did not provide sufficient data to calculate the difference in detection rates.133 In the Belgian study, melanoma case detection rates were 0.5 percent in the total body exam group vs. 0.4 percent in the lesion-detected exam group, p=0.87.132 In the Italian screening program, the melanoma detection rate was 0.4 percent in both the screened group (initial screening and during the followup period after screening) and in the unscreened group.130 In the U.S.-based study, melanoma detection rates were higher in the group with documented skin exam compared to the group without (0.25% in screened group, 0.14% in unscreened group; p<0.001).134
Table 12
Skin Cancer or Precursor Lesion Detection Rates, Screened vs. Unscreened (KQ2).
Stage at melanoma detection. The three studies reporting data on stage at melanoma detection used heterogeneous stage categories (Table 13, Appendix E Figure 3). Two of these were analyses of German national screening program data. One analysis used AJCC stage categories, including in situ melanoma,133 and one analysis reported lymph node and distant metastases at detection.135 The U.S.-based study reported distributions of in situ versus various thickness categories of invasive melanoma.134
Table 13
Stage or Thickness at Melanoma or KC Detection, Screened vs. No Routine Skin Cancer Screening (KQ2).
Findings were inconsistent between two studies including in situ melanoma at detection. In the German study using AJCC stage categories (n=1536 melanoma cases), there was no association between screening and detection of in situ melanoma.133 In the U.S.-based study (n=994 melanoma cases), in situ melanoma made up a larger proportion of cases in the screened group compared to the unscreened group (48.3% of all melanomas detected at in situ stage in screened group vs 34.6% in unscreened group, adjusted hazard ratio [adjHR], 2.6 [95% CI, 2.1 to 3.1]; p<0.001).134
Findings were more consistent across two studies reporting stage at invasive melanoma only (i.e., excluding melanoma in situ); neither study found an association between skin cancer screening and stage at invasive melanoma detection. In the German study using AJCC stage categories (n=1536 melanoma cases), there was no difference between screened and unscreened groups at AJCC combined stages I/II, or combined stages III/IV at detection.133 In the German study using AOK PLUS data from 2005–2012 (n=3504 melanoma cases), lymph node metastasis and distant metastasis were observed at similar rates between screened and unscreened groups. For example, lymph node metastases at detection were similarly detected in persons with documented whole-body screening (5.9%), and those without such documentation (8.5 percent).135
Thickness at melanoma detection. Findings were inconsistent between three studies reporting data on heterogeneous categories of melanoma thickness at detection from three countries: the United States, Italy, and Australia (Table 13, Appendix E Figure 4).129, 130, 134 In the U.S.-based study, there was a higher adjusted hazard ratio in the screened group of detection at thickness ≤1mm (AJCC 7th edition sub-category T1) (adjHR1.8, 95% CI, 1.5 to 2.2; p<0.001), but not in the greater than 1mm category (AJCC sub-category T2-T4) (adjHR 1.0, NS).134 In the Italian study of routine screening compared to usual care, the proportion of melanomas detected at <1mm thickness was similar between groups (70.4% in the screened group and 57.7% in the usual care group, p=0.242), but was higher for screen-detected melanomas detected at <2mm thickness (AJCC sub-category T3-T4; 92.6% of in the screened group and 75.9% of melanomas in the usual care group, p=0.043).130
In the Australian case-control study, 28.3 percent (1083 out of 3,824) of controls reported receiving a clinical skin exam by a physician within the previous 3 years compared to 35.3 percent (1328 out of 3,762) of melanoma cases. The odds of having had a clinical skin exam by a physician decreased as thickness increased in an inverse linear pattern: 7 percent decreased odds for lesions 0.76 to 1.49 mm (95% CI, 0.79 to 1.10); 17 percent decreased odds for lesions 1.50 to 2.99 mm (95% CI, 0.66 to 1.05); and 40 percent decreased odds for lesions ≥3.0 mm (95% CI, 0.43 to 0.83).129
Keratinocyte Cancer
Keratinocyte skin cancer detection. Four included studies provided data on KC rates associated with routine screening from three screening programs (Table 12).130, 132, 133, 135 Two of these used data from Germany during the German National Screening Program at different periods. One found a similar KC detection rate in the screened population compared to those who were unscreened (2.5% vs 1.2%, RR [95% CI] 2.16 [2.11 to 2.21]).135 The other German study found similar numbers of KC cases in both screened and unscreened groups.133 The Belgian study, a comparison of total body exam compared to lesion-detected exam as a result of community skin examination events in two neighboring communities, reported similar detection rates of both BCC and SCC in both the total body exam group compared to the lesion-directed exam group (BCC 1.8% detection rate vs 2.8%, p=0.28; SCC 0.1% vs 0%, p=0.99).132 The Italian study only reported BCC (12 [0.3%]) and SCC (1 [0.03%]) identified in the screening population making comparisons to the unscreened population impossible.130
Keratinocyte skin cancer stage at detection. Only a single study reported the stage of KC detection (Table 13, Appendix E Figure 1). In the German program (n=10,844 KC cases), similar distributions of KC stage in each group were reported (99.9% of KC cases detected at stage I/II in screened group; 99.8% in unscreened group).133
Precursor lesion detection. Two studies – the Belgian study of one-time total body examination compared to one-time lesion-directed examination, and the Italian study of total body examination compared with no screening – reported rates of detection of skin cancer precursor lesions (Table 12, Appendix E Figure 2).130, 132 In the Belgian study, rates of actinic keratoses and atypical nevi were similar in both groups. Actinic keratoses was detected in 7.9 percent of the total body exam group and 7.8 percent of the lesion-directed exam group (p=0.90). Atypical nevi were detected at 15.1 percent of the total body exam group and 17.3 percent of the lesion-directed group (p=0.33).132 As was the case with the KC data, in the Italian study the number of dysplastic nevi were only reported in the screening population (11 out of 3635 [0.3%]) and not the unscreened population.130
Specific Population Results (KQ2a)
No studies reported KC detection rates or KC stage at detection comparisons stratified by specific population groups.
The Australian case-control study (n=7,586), reported age- and sex-specific adjusted odds of having reported a clinical skin exam by a physician in melanoma cases according to increasing thickness compared to controls (Table 14).129 The odds of having received a clinical skin exam followed similar patterns as the main result for age and sex subgroups, with thinner melanomas associated with higher odds of having received a clinical skin exam. Males with the thinnest melanomas (0.01mm to 0.75mm) were more likely to have received a clinical skin exam (adjOR 1.54, p<0.01), while males with the thickest melanomas (≥ 3mm) were less likely to have received a clinical exam (adjOR 0.62, p<0.05). The trend was present but not statistically significant among females.129 Within both age strata (age 20–49 and age 50–74, males and females combined) the highest odds of having received a clinical skin exam were observed for people with the thinnest melanomas (0.01mm–0.75mm), while people with the thickest melanomas (≥3mm) had the lowest odds of clinical skin exam.129
Table 14
Stage or Thickness at Melanoma Detection, Specific Populations, Routine Clinician Skin Examination Compared to Usual Care or No Screening (KQ2a).
The Italian study reported melanoma detection and thickness by sex and age in the screening population (n=3,635) (Table 14).130 During the initial screening period, the majority of those with melanoma <1mm thick were females (7 of 10), whereas the majority of those with melanoma >1mm thick were males (3 of 4). This pattern was mirrored in the followup period (n=3,618).
The U.S.-based study reported melanoma detection rates and thickness at detection among those age 65 and older (Table 15). Findings were similar to those in the full study population. For example, melanoma detection rates were higher in the screened population (0.33%) compared to the unscreened population (0.21%; adjHR 1.6 [95% CI, 1.3 to 2.0]; p<0.001). In addition, in situ melanomas made up a larger proportion of melanoma cases in the screened group (45.8%) compared with the unscreened group (38.5%; adjHR 1.9 [95% CI, 1.4 to 2.6]; p<0.001). There was a higher adjusted hazard ratio in the screened group for detection at thickness ≤1mm (adjHR 1.9 [95% CI, 1.3 to 2.6]; p<0.001) but not for thickness >1mm (adjHR 1.0 [95% CI, 0.7 to 1.6]; p=0.90).134
Table 15
Melanoma Detection Rates and Thickness at Melanoma Detection, Population Age ≤65, Screening Compared to No Screening (KQ2a).
KQ3. What Are the Harms of Skin Cancer Screening and Diagnostic Followup? Do the Harms of Screening Vary by Subgroups (e.g., Age, Sex, Skin Type, Race/Ethnicity, Socioeconomic Status, or UV Exposure)?
Summary of Results
We identified only two small fair-quality nonrandomized studies that explicitly addressed the harms of skin cancer screening. One was conducted in Germany (n=45)136 and assessed cosmetic acceptance of shave biopsy in a screened population at 6-month followup; lesions suspected of melanoma were excluded. The other was conducted in the United States (n=187)137, 138 and assessed psychological wellbeing at 5 and 8 months after screening.
In the German study, 27 patients rated 7 percent (4 out of 56) of shave sites as having poor cosmetic outcomes at 6-month followup (median score 1.5, IQR [1–2], excellent to good). In the U.S.-based study of adults who underwent skin cancer screening by trained primary care providers (n=187), participants at 5- and 8-month followup assessment scored within the normal range on measures of anxiety, depression, or none to minimal psychological impacts of screening.
Detailed Results
Overview of Included Studies
Two fair-quality nonrandomized studies136, 138 (n=232) met inclusion criteria, one of which was brought forward from the previous review (Tables 3–5).136 One German study examined the cosmetic harms of shave biopsy excision.136 The other study set in the United States reported on the potential psychological harms137, 138 of skin cancer screening. The U.S. study was an evaluation of the same physician-focused decision support intervention that was included for KQ2134 and was newly identified since the previous review. No included studies for KQ1 reported harms data. No included studies reported on procedure-related adverse events beyond 30 days (e.g., scar revisions) in screened populations.
One small study was conducted in Germany (n=45) in 2000 (Table 16).136 This study assessed patient-reported cosmetic acceptance of deep shave excisions at 6-months of followup after excision. Only razor blade excisions of macular melanocytic nevi less than 15 mm in diameter were included in this study; lesions suspected of melanoma were excluded. Participants were identified during routine skin cancer screening before the implementation of the German National Screening Program.
Table 16
Cosmetic Harms of Routine Skin Cancer Screening or Diagnostic Workup (KQ3).
The other study was conducted in a United States academic medical center (n=187)137, 138 (Table 17). Primary care providers who had completed an online training program for detecting and diagnosing skin cancer conducted skin cancer screenings. This study used various scales to estimate patient-reported psychological harms (e.g., anxiety, depression, physical and social consequences) and health-related quality of life at 5 and 8 months after screening. The study compared results between two screened patient groups—patients classified as those with “full body exam” and those with “partial body exam.” Another comparison examined psychological indicators among patients who underwent biopsy following screening and those without biopsy at 5 months after screening. Although only people with skin cancer screening documented in their electronic medical record were included, group allocation was based on patients’ self-reported screening and experience (i.e., recall of screening, the level of undress, and body parts examined).
Table 17
Psychological Measures and Quality of Life Among Individuals Who Reported Full-Body Skin Cancer Screening and Those Who Reported Partial Body Examination (KQ3).
Population Summary
The German study included persons aged 15–54 years (mean age 32 years);136 the other study included adults aged 35 years and older (Table 5).138 An estimated 44.4 percent of total participants in both studies were females.
In the U.S.-based study,137, 138 89.8 percent of participants were White, and 5.3 percent were Black. In this study, 20.9 percent of participants had a personal history of skin cancer, 58.6 percent had a family history of melanoma, and 40.1 percent had received previous skin exams.
Quality
Both included studies for this key question were rated fair quality. Both were nonrandomized cross-sectional studies with a very small number of participants and no unscreened comparator group. In the German study,136 only 60 percent (27 out of 45) of recruited persons evaluated cosmetic outcomes of the shave sites. The results do not assess cosmetic results from excisional biopsies needed for melanoma diagnosis, which are more invasive procedures. Further, this study reports the number of shave sites, rather than the number of patients dissatisfied with the shave sites’ appearance. The U.S.-based study137, 138 did not assess for psychological or quality of life outcomes at baseline. The authors compared outcomes between two groups of patients based on the patients’ recall of whether part of their body was examined or whether their whole body was examined for skin cancer five months prior to completing the survey.
Detailed Results by Outcome
A fair-quality study of routine outpatient cancer screening (n=45 patients, 56 deep shave excisions) assessed patients’ perceptions of the cosmetic acceptance of deep shave excisions of macular melanocytic nevi with the razor blade technique at 6-months of followup (Table 16).136 Patients used a four-point scale (1=excellent, 2=good, 3=moderate, or 4=poor) to evaluate cosmetic outcomes with no prespecified judgment criteria. The median patient evaluation score was similar (1.5) and IQR [1–2] (excellent to good). Patients judged 7 percent (4 of 56) of shave sites as having poor cosmetic outcomes, 4 percent (2) as moderate outcomes, 39 percent (22) as good outcomes, and 50 percent (28) as excellent outcomes.
A fair-quality study137, 138 estimated the potential for psychological harms and health-related quality of life after skin cancer screening among 187 adults at 5-months of followup and 126 (67.4% of 187) at 8-months of followup (Table 17). This study also estimated the potential for psychological harms associated with skin biopsy among 186 patients at 5-months following skin cancer screening.137 This study assessed patients’ anxiety and depression with the Hospital Anxiety and Depression rating scale (HADS), general anxiety with the Spielberger State-Trait Anxiety Index—form 6 (STAI-6), positive and negative psychological impacts of screening with the Psychological Consequences Questionnaire (PCQ), and health-related quality of life with the 12-Item Short-Form Health Survey (SF-12).
Participants had undergone skin cancer screening and were classified as having “full body examined” or only “partial body exam.” Participants in both groups scored within the normal range on various scales, with no indication of anxiety, depression, or other negative psychological impacts from screening at 5- and 8-month followup assessment.138 Participants with skin biopsy (n=23) and without biopsy (n=163) also scored within the normal range on psychological scales. Overall, there were no meaningful differences in psychological indicators between biopsied and non-biopsied patients.137
Specific Population Results (KQ3a)
We did not identify evidence on the potential harms of screening for specific populations.
KQ4. What Is the Association Between Detection of Precancerous Lesions or Earlier Stage Skin Cancer and Morbidity and Mortality Due to Skin Cancer or All-Cause Mortality? Does This Association Vary by Subgroups (e.g., Age, Sex, Skin Type, Race/Ethnicity, Socioeconomic Status, or UV Exposure)?
Summary of Results
In three included nonrandomized studies (two good-quality, one fair-quality; n=407,133) reporting melanoma-specific mortality and three nonrandomized studies (one good-quality, two fair-quality; n=473,660) reporting all-cause mortality, progression of stage at detection was consistently and positively associated with increased risk of melanoma mortality. Risk estimates varied according to referent groups used. Compared to in situ disease at detection, adjusted hazard ratios for melanoma mortality were 5.8 (95% CI, 5.3 to 6.3) for localized, 31.5 (95% CI, 28.9 to 34.2) for regional, and 169.6 (95% CI, 154.2 to 186.6) for distant stage in one U.S.-based study (n=185,219).146 Two studies using localized stage at detection as the referent group found a similar pattern of increasing melanoma mortality risk with increasing stage.
In two studies (one good-quality, one fair-quality; n=135,490), melanoma mortality was higher for males than for females. Three studies (n=708,814) that examined melanoma mortality risk with respect to racial and ethnic groups found a higher risk among Black, Hispanic, and Asian American, Native American or Pacific Islander (AANAPI) adults compared with White adults. One of these found similar odds of melanoma mortality risk for White and Black persons within each stage of detection. In the other two studies with overlapping populations, melanoma mortality risk was higher among Black, Hispanic, and AANAPI adults with melanoma AJCC Stage I and SEER localized stages compared to White adults. One of these studies also demonstrated a higher risk of melanoma mortality among Hispanic persons with regional or distant melanoma stages compared with White adults.
Regarding all-cause mortality, the same pattern was observed over three large nonrandomized studies using varying referent groups. In one study (n=185,219), the risk for all-cause mortality was adjHR 1.5 (95% CI, 1.5 to 1.5) for localized, 3.9 (95% CI, 3.8 to 4.1) for regional, and 15.8 (95% CI, 14.9 to 16.7) for distant disease, compared to in situ melanoma at detection.
No included studies addressed KC mortality by stage at detection. We did not identify studies that evaluated the association between stage at diagnosis and skin cancer morbidity.
Detailed Results
Overview of Included Studies
A total of nine fair- or good-quality nonrandomized studies with data collected between 1975 and 2016 (n=1,326,051) were included139–143, 145–147 (Tables 18–23). All nine studies were newly identified since the prior recommendation; however, some studies had overlapping populations. Seven studies (n=1,037,610) reported the association between stage at diagnosis and melanoma mortality,139, 140, 142–146 and three studies141, 146, 147 (n=473,660) reported the association between the stage of melanoma at diagnosis and all-cause mortality in adults. No included studies evaluated the association between stage at diagnosis and skin cancer morbidity. No studies contributed data for KC mortality.
Table 18
Study Characteristics, Included Studies (KQ4).
Table 19
Melanoma Stage at Diagnosis and Melanoma Mortality (KQ4).
Table 20
Melanoma Stage at Diagnosis and All-Cause Mortality (KQ4).
Table 21
Melanoma Stage at Diagnosis and Melanoma Mortality, by Race and/or Ethnicity (KQ4a).
Table 22.
Melanoma Stage at Diagnosis and Melanoma Mortality, by Race and/or Ethnicity (KQ4a) From Qian 2021 Study (n=398,034).
Table 23
Melanoma Stage at Diagnosis and Melanoma Mortality, Sex and Age Group (KQ4a).
No studies included in the 2016 review were carried forward for this KQ due to the change in our inclusion criteria to focus on stage at detection rather than independently on melanoma thickness, which is included in the AJCC staging criteria.
Population Summary
Studies used large databases with patient information from the United States (SEER, National Cancer Database), Australia (Queensland Cancer Registry), Norway (data from the Norwegian Malignant Melanoma Registry matched with data from other sources), and Sweden (Swedish Cancer Registry). The six U.S.-based studies used the SEER and National Cancer Database data collected between 1975 and 2016 (median data collection period 22 [range 11 to 41] years) with the longest period in the U.S.-based Qian 2021144 study. Studies not conducted in the United States used data that were collected between 2003–2005 in Sweden,147 2008–2012 in Norway,145 and 1995–2008 in Australia.142
Although the patient populations across the included United States studies overlap, these studies differed in the outcomes they presented (i.e., all-cause mortality, melanoma-specific mortality), melanoma staging systems they used (AJCC or SEER), referent stages the authors used to estimate the risk of melanoma mortality, and subgroup analysis (Table 18).
The weighted average age across all included studies was 59.0 years. Across included studies, 45.4 percent of all participants were female. All six U.S.-based studies provided information on participants’ race and/or ethnicity.139–141, 143, 144, 146 Most participants in these studies (96.0%) were white, 0.7% were Black, 0.8% were Asian American or Pacific Islanders, 0.2% were American Indian or Alaska Natives, and 3.0% of participants were of Hispanic ethnicity. Participants’ personal, family, or environmental risk factors for skin cancer were rarely reported.
Quality
Among the nine included studies, six139–141, 143, 144, 147 were rated fair- and three142, 145, 146 were rated good-quality nonrandomized studies. The included studies used large national-level databases with the data systematically collected over many decades (i.e., SEER, Swedish Cancer Registry, Queensland Cancer Registry [Australia], Norwegian Malignant Melanoma Registry). The followup time was sufficient to observe mortality.
Quality concerns were primarily related to the limitations of nonrandomized studies using retrospectively collected data, which include the incompleteness and inaccuracy of the collected data, incompleteness of individual-level data, and unreported handling of missing data. For example, miscoding and missing data in the SEER database resulted in a restaging of a large number of diagnoses.158, 159 In addition, the included studies either did not report how they handled missing data or reported omitting missing data from their analyses.
Detailed Results by Outcome
Melanoma Mortality and Stage at Diagnosis
Overall, seven fair- and good-quality studies reported an association between the stage of melanoma at diagnosis and melanoma mortality using either the AJCC or SEER stages (Table 19).139, 140, 142–146 Four studies contributed data only for specific populations (see Specific Population Results subsection).139, 140, 142, 144
All three studies143, 145, 146 with estimates across all participants demonstrated a consistent and statistically significant increase in the risk of melanoma mortality with disease progression. Two large U.S.-based studies using SEER staging with overlapping populations used different referent categories.143, 146 A good-quality Ward-Peterson 2016 study (n=185,219) compared the risk for melanoma mortality at localized, regional, and distant melanoma with in situ melanoma diagnosed between 1982 and 2011.146 Using in situ melanoma as the reference category, the adjusted HR of risk for melanoma mortality was 5.8 (95% CI, 5.3 to 6.3) for localized, 31.5 (95% CI, 28.9 to 34.2) for regional, and 169.6 (95% CI, 154.2 to 186.6) for distant stages. The fair-quality Mahendraraj 2017 study (n=213,827) also demonstrated a statistically significantly increased risk of melanoma mortality at regional (OR, 3.8 [95% CI, 3.5 to 4.1]) and distant (OR, 7.5 [95% CI, 6.3 to 8.9]) stages compared to the reference group of local disease.143 Another smaller study using SEER staging of 8,087 Norwegian adults demonstrated a trend similar to the U.S. studies—increasing risk for melanoma mortality within disease progression. Using the localized stage as the reference category, the adjusted risk of melanoma mortality was adjHR 4.00 (95% CI, 3.26 to 4.90) for regional disease and 16.82 (95% CI, 12.88 to 21.95) for distant disease.145
All-Cause Mortality and Stage at Diagnosis
Two fair-quality studies and one good-quality study (n=473,660) reported an association between the stage of melanoma at diagnosis and all-cause mortality in adults (Table 20). The studies used the AJCC,141 SEER,146 and tumor, node, metastasis (TNM)147 sub-stages.
Two U.S.-based studies with overlapping populations (Farrow 2020 and Ward-Peterson 2016) used different staging systems. The Farrow 2020 study (n=268,668)141 used the AJCC’s stage IV as a referent stage and demonstrated that the risk for all-cause mortality was statistically significantly lower in patients with earlier stages. The risk of all-cause mortality was 62 percent lower in persons with stage III (adjHR, 0.38 [95% CI, 0.36 to 0.40]), 81 percent lower among those with stage II (adjHR, 0.19 [95% CI, 0.18 to 0.20]), and 91 percent lower in persons diagnosed with stage I (adjHR, 0.09 (95% CI, 0.08 to 0.10]), compared to persons diagnosed with stage IV melanoma between 2004 and 2015. The Ward-Peterson 2016 study used the SEER stages of in situ, localized, regional, and distant melanoma for adults diagnosed between 1982 and 2011. This study demonstrated similar results—disease progression is statistically significantly associated with a higher risk of all-cause mortality. In this study, the risk of all-cause mortality was 50 percent higher in persons with the localized stage (adjHR, 1.5 [95% CI, 1.5 to 1.5]), 290 percent higher in persons with the regional stage (adjHR, 3.9 [95% CI, 3.8 to 4.1]), and 1,480 percent higher among persons diagnosed with distant melanoma (adjHR, 15.8 [95% CI, 14.9 to 16.7]), compared with the reference group of in situ melanoma.
A third study147 using TNM sub-staging of 19,773 Swedish adults found an increased risk of all-cause mortality within each T, N, and M stages. For example, the risk for all-cause mortality was statistically significantly higher among persons diagnosed with T4b (adjHR, 5.90 [95% CI, 5.17 to 6.74]) compared to those diagnosed with T1a, higher in persons diagnosed with N+ (adjHR, 2.24 [95% CI, 1.82 to 2.75]) compared to those with N0, and higher in those with M+ (adjHR, 3.17 [95% CI, 2.40 to 4.19]) compared to patients with the M0 substage.
Mortality and Lesion Thickness at Diagnosis
Lesion thickness at diagnosis was not an exposure of interest for this review, as the association of increasing tumor thickness and melanoma mortality was previously established by the systematic review to support the 2016 recommendation.106 Two included studies (n=27,860)145, 147 also reported the association between melanoma thickness and risk for either all-cause or melanoma mortality. Consistent with the prior review, both studies showed an increased risk for either all-cause or melanoma mortality with increasing melanoma thickness at diagnosis.
In the Zheng 2020 study (n=19,773), the risk for all-cause mortality was higher among persons diagnosed with T4a (adjHR, 4.37 [95% CI, 3.72 to 5.13]) compared to those diagnosed with T1a.147 In the Robsahm 2018 study (n=8,087), the risk for melanoma mortality was also statistically significantly higher among persons diagnosed with T4 (adjHR 9.68 [95% CI, 7.06 to 13.28]) compared to persons diagnosed with T1.145
Specific Population Results (KQ4a)
Three overlapping studies provided estimates of melanoma mortality risk stratified by race and/or ethnicity (Table 21). The Dawes 2016 study (n=96,953) used AJCC stages and White persons as a reference group to estimate the risk for melanoma mortality among Black, AANAPI, and Hispanic persons diagnosed with melanoma between 1992 and 2009.139 This study demonstrated that Black persons, compared to White persons, had a statistically significantly higher risk of melanoma mortality at stages I (HR, 3.04 [95% CI, 2.34 to 3.95]) and III (HR, 1.86 [95% CI, 1.21 to 2.87]), but not at stages II (HR, 1.34 [95% CI, 0.92 to 1.95]) and IV (HR, 1.03 [95% CI, 0.68 to 1.57]). Additionally, the risk for melanoma mortality was statistically significantly higher among Black, AANAPI, and Hispanic adults aged 25–49, 50–74, and ≥75 years with melanoma stage I compared with White adults. Among pediatric patients and young adults aged 0–24 years, Black (HR, 9.50 [95% CI, 1.23 to 73.62]) and Hispanic (HR, 3.75 [95% CI, 1.22 to 11.56]) patients with stage II melanoma had a higher risk for melanoma mortality than White patients. The Qian 2021 study144 (n=398,034) used SEER stages and White persons as a reference group to estimate the risk for melanoma mortality among Black, Asian or Pacific Islanders (API), American Indian or Alaska Native (AIAN), and Hispanic persons diagnosed with melanoma between 1975 and 2016 (Table 22). This study, which provided similar risk estimates as the Dawes 2016 study,139 demonstrated that Black, Hispanic, API and AIAN persons diagnosed with localized melanoma between 2010 and 2016 had a statistically significantly higher risk (HRs) of melanoma mortality compared with White persons diagnosed at the same stage. This study also found that Hispanic individuals diagnosed with regional and distant melanoma between 2010 and 2016 had a higher risk of melanoma mortality compared with White persons at the same stages. The Mahendraraj 2017 study143 (n=213,827) used SEER stages to estimate melanoma mortality risk for White (n=212,721) and Black (n=1,106) persons diagnosed with melanoma between 1988 and 2011. Overall, 19,207 (9.0%) of White and 241 (21.8%) of Black persons died of melanoma during the observation period. The risk for melanoma mortality was statistically significantly higher among persons with regional and distant stages for both White and Black persons than the reference group of localized disease (Table 21).
Two studies (n=135,490) compared the difference in the risk of melanoma mortality between females and males by stage (Table 23). The U.S.-based Enninga 2017 study140 (n=106,511) used SEER stages to compare the risk for melanoma mortality among females and males diagnosed with melanoma between 1992 and 2011. The age-adjusted risk of melanoma mortality among males was higher at the localized (adjHR 1.59, [95% CI 1.49 to 1.70]), regional (adjHR 1.37, [95% CI 1.28 to 1.47]), and distant (adjHR 1.10, [95% CI 1.01 to 1.20]) stages than among female patients. The higher risk for melanoma mortality for males persisted across all age categories (i.e., 18–45, 46–54, and ≥55) for persons with the localized and regional stages; however, the difference between males and females with distant disease for the same age groups was not statistically significant. The Australian Khosrotehrani 2015 study142 (n=28,979) compared the risk for melanoma mortality at different AJCC stages among females and males by stage. The data for AJCC stages III and IV were combined due to the small number of persons diagnosed with these stages. In this study, the risk of melanoma mortality for females of any age (15–45 years, 46–59 years, and ≥60 years) was 36 percent lower (adjOR, 0.64 [95% CI, 0.51 to 0.82]) at stage I and 29 percent lower (adjOR, 0.71 [95% CI, 0.58 to 0.87]) at stage II compared with male patients. There was no statistically significant difference in the risk of melanoma mortality between females and males with the combined III and IV stages (adjOR, 0.70 [95% CI, 0.44 to 1.10]). Female patients with stage II and aged 15–45 and those with any stage aged ≥60 years also had a statistically significantly lower risk of melanoma mortality compared with male patients.
- Literature Search
- KQ1. What Is the Effectiveness of Routine Skin Cancer Screening With Visual Skin Examination by Clinicians in Reducing Skin Cancer Morbidity and Mortality or All-Cause Mortality? Does the Effectiveness of Screening Vary by Subgroups (e.g., Age, Sex, Skin Type, Race/Ethnicity, Socioeconomic Status, or UV Exposure)?
- KQ2. Does Routine Skin Cancer Screening Lead to Higher Rates of Detection of Precancerous Lesions or Earlier Stage Skin Cancer Compared to Usual Care (e.g., Lesion-Directed Skin Examination)? Do Rates of Earlier Skin Cancer Detection Vary by Subgroups (e.g., Age, Sex, Skin Type, Race/Ethnicity, Socioeconomic Status, or UV Exposure)?
- KQ3. What Are the Harms of Skin Cancer Screening and Diagnostic Followup? Do the Harms of Screening Vary by Subgroups (e.g., Age, Sex, Skin Type, Race/Ethnicity, Socioeconomic Status, or UV Exposure)?
- KQ4. What Is the Association Between Detection of Precancerous Lesions or Earlier Stage Skin Cancer and Morbidity and Mortality Due to Skin Cancer or All-Cause Mortality? Does This Association Vary by Subgroups (e.g., Age, Sex, Skin Type, Race/Ethnicity, Socioeconomic Status, or UV Exposure)?
- Results - Screening for Skin Cancer: An Evidence Update for the U.S. Preventive ...Results - Screening for Skin Cancer: An Evidence Update for the U.S. Preventive Services Task Force
- zinc finger TRAF-type-containing protein 1 isoform 1 [Homo sapiens]zinc finger TRAF-type-containing protein 1 isoform 1 [Homo sapiens]gi|1060604633|ref|NP_001317547.1|Protein
- Homo sapiens glucosidase II alpha subunit (GANAB), transcript variant 2, mRNAHomo sapiens glucosidase II alpha subunit (GANAB), transcript variant 2, mRNAgi|1519244487|ref|NM_198334.3|Nucleotide
- Solanum tetrathecum isolate ASP033 tRNA-Leu (trnL) gene and trnL-trnF intergenic...Solanum tetrathecum isolate ASP033 tRNA-Leu (trnL) gene and trnL-trnF intergenic spacer, partial sequence; chloroplastgi|116014553|gb|DQ855051.1|Nucleotide
Your browsing activity is empty.
Activity recording is turned off.
See more...