Cribado para la reducción de la morbilidad y la mortalidad en el melanoma maligno
Resumen
Antecedentes
El cribado para el melanoma maligno puede reducir la morbilidad y la mortalidad por la enfermedad mediante la detección más precoz, ya que el pronóstico se asocia estrechamente con el espesor de la lesión en el momento del diagnóstico. Sin embargo, también puede haber daños con el cribado de individuos que no manifiestan inquietudes acerca de las lesiones cutáneas, como el sobrediagnóstico de lesiones que nunca habrían causado síntomas si hubieran seguido siendo no detectadas. El sobrediagnóstico provoca daños por el tratamiento innecesario y las consecuencias psicosociales de ser asociado con un diagnóstico de cáncer. En cualquier tipo de cribado, los beneficios deben superar los efectos perjudiciales. El cribado para el melanoma maligno actualmente se practica en muchos países, y se registra un aumento significativo de la incidencia de la enfermedad, mientras que la mortalidad permanece en gran medida inalterada.
Objetivos
Evaluar los efectos sobre la morbilidad y la mortalidad del cribado para el melanoma maligno en la población en general.
Métodos de búsqueda
Se hicieron búsquedas en las siguientes bases de datos hasta mayo de 2018: el Registro Especializado de Piel Cochrane (Cochrane Skin Specialised Register), CENTRAL, MEDLINE, Embase y LILACS. También se buscó en cinco registros de ensayos, se verificaron las listas de referencias de estudios incluidos y otros estudios relevantes en busca de referencias adicionales a ensayos controlados aleatorizados (ECA), se utilizó el seguimiento de citas (Web of Science) de artículos clave y se les consultó a los investigadores preguntados acerca de estudios e informes de estudios adicionales.
Criterios de selección
ECA, incluidos ensayos aleatorizados grupales, sobre el cribado del melanoma maligno comparado con ningún cribado, independientemente de la modalidad o el contexto del cribado, en cualquier tipo de población y en cualquier grupo etario donde no se sospechara que las personas tuvieran melanoma maligno. Se excluyeron los estudios en personas con predisposición genética para el melanoma maligno (p.ej., lunar atípico familiar y síndrome de melanoma) y los estudios realizados exclusivamente en personas con melanomas anteriores.
Obtención y análisis de los datos
Se utilizaron los procedimientos metodológicos estándar previstos por Cochrane. Los resultados primarios de esta revisión fueron: la mortalidad total, el sobrediagnóstico del melanoma maligno y las consecuencias psicosociales y para la calidad de vida.
Resultados principales
Se incluyeron dos estudios con 64 391 participantes. El primer estudio fue un ensayo aleatorizado de una intervención desarrollada para aumentar la tasa de rendimiento del autoexamen minucioso de piel. El grupo de intervención recibió materiales de instrucción, como indicaciones y ayudas, un video instructivo de 14 minutos y una sesión de asesoramiento breve y una llamada telefónica de seguimiento breve a las tres semanas con un educador en salud, orientados a aumentar el rendimiento del autoexamen minucioso de la piel. El grupo control recibió una intervención dietética con un seguimiento similar. El ensayo incluyó a 1356 personas, que se reclutaron de 11 centros de atención primaria de los EE.UU. entre 2000 y 2001. La media de edad de los participantes fue de 53,2 años, y había un 41,7% de hombres. Este estudio no informó ninguno de los resultados primarios o los siguientes resultados secundarios: mortalidad específica del melanoma maligno, tasas de falsos positivos (biopsias de piel/excisiones con resultado benigno) o tasas de falsos negativos (melanomas malignos diagnosticados entre las rondas de cribado y hasta un año después de la última ronda). A todos los participantes se les pidió que completaran entrevistas telefónicas de seguimiento a los dos, seis y 12 meses después de la asignación al azar.
El segundo estudio fue un estudio piloto de un ECA en grupos sobre el cribado poblacional para el melanoma maligno en Australia. Este ensayo piloto incluyó a 63 035 adultos mayores de 30 años. El programa de tres años incluyó educación de la comunidad, un componente de formación y de apoyo para los médicos y servicios gratuitos de cribado de la piel. La edad media de las personas que asistieron a las clínicas de revisión de la piel (que fueron realizadas por médicos de atención primaria en lugares de trabajo, lugares comunitarios y hospitales locales, e incluyeron sesiones diurnas y nocturnas) fue de 46,5 años, y el 51,5% eran hombres. El estudio incluyó comunidades enteras, dirigidas a participantes de más de 30 años de edad, pero no se informaron datos sobre la edad y el sexo de toda la población del estudio. La duración del estudio fue de tres años (1998 a 2001), y los resultados se midieron en consultorios de cribado durante ese tiempo. No hubo seguimiento adicional para ningún resultado. El grupo de control no recibió ningún programa. Debido a la falta de financiación, no se realizó el ensayo aleatorizado en grupos planificado en 560 000 adultos. En el momento de redactarse esta revisión, no había datos publicados ni no publicados sobre los resultados predefinidos disponibles y no se esperaban resultados de mortalidad del estudio piloto.
El riesgo de sesgo de estos estudios fue alto para el sesgo de rendimiento (cegamiento del personal de estudio y los participantes) y alto o poco claro para el sesgo de detección (cegamiento de la evaluación de resultados). El riesgo de sesgo en otros dominios fue incierto o bajo. No se pudo evaluar la certeza de la evidencia de los resultados primarios según lo planificado debido a la ausencia de datos.
Conclusiones de los autores
El cribado de la población general de adultos para el melanoma maligno no está apoyado ni refutado por la evidencia actual de ECA. Por lo tanto, no cumple los criterios aceptados para la aplicación de los programas de cribado de población. Esta revisión no investigó los efectos del cribado de los pacientes con antecedentes de melanoma maligno o en pacientes con una disposición genética para el melanoma maligno (p.ej., síndrome de lunar atípico familiar y síndrome de melanoma). Se necesita un ensayo aleatorizado realizado rigurosamente, que evalúe la mortalidad global, el sobrediagnóstico, las consecuencias psicosociales y el uso de recursos, para determinar los efectos beneficiosos y perjudiciales del cribado para el melanoma maligno.
PICO
Resumen en términos sencillos
Cribado para melanoma maligno (un tipo de cáncer de piel)
Pregunta de la revisión
Se examinó la evidencia acerca del efecto del cribado para el melanoma maligno (un tipo de cáncer de piel) en los pacientes en quienes no era presunto este cáncer, es decir, los pacientes sin un lunar o lesión sospechosa (área de la piel con una apariencia extraña en comparación con la piel circundante), en comparación con ningún cribado. Se incluyó cualquier tipo de cribado (p.ej., autoexamen de piel, por un profesional de la salud) de cualquier individuo en quien no era presunto un melanoma maligno, independientemente de la edad o el sexo. Se incluyeron estudios en pacientes con riesgo alto de desarrollo de melanoma maligno, pero no los que tenían antecedentes de melanoma.
Antecedentes
El melanoma maligno es un tumor de la piel que puede causar la muerte al extenderse a otras partes del cuerpo; el número de tumores está aumentando, mientras que en muchos países el riesgo de morir por la enfermedad no ha aumentado de manera similar. El cribado para el melanoma maligno se realiza mediante autoexamen visual de la piel o la inspección visual por un médico u otro profesional de la salud. El cribado tiene la posibilidad de reducir las muertes por melanoma. Sin embargo, también existen daños potenciales del cribado de los pacientes sin síntomas de melanoma, como encontrar melanomas que nunca habrían causado síntomas si hubieran permanecido no detectados (es decir, sobrediagnóstico), cirugía innecesaria y estrés psicológico posible. Es importante establecer la base de la evidencia para el cribado.
Características de los estudios
Dos estudios cumplieron los criterios de inclusión. El objetivo del primer estudio, realizado en los EE.UU., fue investigar cómo aumentar la frecuencia en que los pacientes realizan el autoexamen de la piel. A los 1356 participantes se les realizaron entrevistas telefónicas de seguimiento completas a los dos, seis y 12 meses después de la asignación al azar. La edad promedio de los participantes fue de 53,2 años; el 41,7% eran hombres.
El segundo estudio incluyó 18 comunidades de Australia (63 035 adultos) que se asignaron o no a un programa extrahospitalario de cribado de melanoma de tres años. El estudio no presentó información sobre la media de edad ni la proporción de hombres y mujeres en la población total del estudio, pero la edad promedio de los que asistían a los consultorios de cribado de la piel fue de 46,5 años, y un 51,5% eran hombres. El estudio duró tres años; los resultados se midieron en las clínicas de cribado durante este tiempo. No hubo seguimiento adicional. La finalidad del estudio fue investigar si era posible realizar un ensayo más grande, que fue interrumpido por falta de financiación.
El primer estudio fue financiado por el National Cancer Institute (US); el segundo, por el Queensland Cancer Fund and Queensland Health (Australia).
Resultados clave
No hubo información de ninguno de los estudios sobre los efectos del cribado en las muertes totales, el sobrediagnóstico del cribado ni la calidad de vida de los participantes. Tampoco se informaron los siguientes resultados: muertes por cáncer de piel y tasas de falsos positivos/negativos (es decir, diagnosticar una lesión cutánea como melanoma cuando no está presente/no reconocer un melanoma cuando está presente). Por lo tanto, no se sabe si el cribado para el melanoma maligno tiene algún beneficio, o si el beneficio posible sería superado por los efectos perjudiciales. El cribado general de la población adulta para el melanoma maligno no está apoyado ni refutado por la evidencia de ensayos bien diseñados hasta mayo 2018 y, por lo tanto, no cumple con los criterios aceptados para la implementación de los programas de cribado.
Fiabilidad de la evidencia
No se pudo evaluar la confiabilidad de la evidencia para los resultados primarios ya que no se evaluaron.
Authors' conclusions
Summary of findings
| Screening compared with no screening for malignant melanoma | |
| Patient or population: asymptomatic people Settings: any setting Intervention: screening Comparison: no screening | |
| Outcomes | Comments |
| Total mortality | Not measured |
| Overdiagnosis of malignant melanoma | Not measured |
| Quality of life/psychosocial consequences | Not measured |
| Mortality specific to malignant melanoma | Not measured |
| False positive rates | Not measured |
| False negative rates | Not measured |
Background
Description of the condition
The incidence of malignant melanoma in Western populations has risen many‐fold over recent decades (Garbe 2009). This is likely due in part to an increase in exposure to risk factors, mainly ultraviolet (UV) radiation from the sun and artificial sources (Waldmann 2012). However, it is also clear that some of the rise in incidence is caused by overdiagnosis due to increased disease awareness and screening, since the large increases in incidence has not always been followed by similar increases in mortality (Norgaard 2011; Welch 2005). The prognosis of malignant melanoma is closely correlated to the thickness of the lesion at diagnosis, with thinner lesions having a much lower risk of metastases and a substantially better prognosis (Breslow 1970). The vast majority of the observed increase in the incidence of invasive malignant melanoma represents thin lesions (Norgaard 2011; Welch 2005), and the increase is even more pronounced for melanoma in situ (Johnson‐Obaseki 2015). In contrast, the incidence of thick melanomas has remained largely constant in younger age groups, while some studies also suggest an increase of thick melanomas in older age groups (Norgaard 2011). The lifetime risk of dying from malignant melanoma in Western populations is strongly correlated with birth cohort, with an increasing risk in successive generations born from 1875 with a peak in cohorts born between 1936 and 1957 depending on region, followed by a gradually decreasing risk. This pattern suggests that mortality from malignant melanoma will gradually move to older age groups over time and eventually decrease, even without improved care or treatment (Autier 2015).
The risk of dying from malignant melanoma is higher in men than in women and is also correlated with skin complexion, with the highest risk in people with low skin pigmentation. In the US, the lifetime risk of dying from malignant melanoma is 0.24% among white women and 0.49% among white men, compared to 0.04% among black women and men (National Cancer Institute 2016). In Australia, malignant melanoma is substantially more common and mortality is also higher; the risk of dying from melanoma by the age of 85 years is 0.44% for women and 1.3% for men, and the risk of being diagnosed with melanoma is 4.3% for women and 7.1% for men (Cancer Australia 2016).
The most important avoidable risk factor is exposure to UV radiation from sunlight and artificial sources (Gandini 2005). Intermittent sun exposure confers an increased risk, while continuous exposure (i.e. from working outdoors) seems to be inversely associated with the risk of malignant melanoma (Gandini 2005). Exposure in childhood appears to induce a higher risk than exposure later in life (Gruber 2006). Observational studies found an association between artificial sources of UV radiation, such as solariums, and malignant melanoma (Lazovich 2016). Other risk factors include blonde or red hair, green or blue eyes, freckles, an inability to tan, a family history of malignant melanoma, and a large number of naevi and dysplastic naevi (Marks 2000). One randomised trial showed that sunscreen reduced the risk of malignant melanoma, but there were few events in the trial (Green 2011). Educational programmes, including counselling on the avoidance of intense and intermittent sun exposure and use of sunscreen, have been suggested as a way to reduce mortality from malignant melanoma through primary prevention; a Cochrane Review that evaluates this strategy is currently in progress (Langbecker 2014).
Screening for malignant melanoma is not recommended in the US (USPSTF 2016), Canada (CTFPHC 2013), Australia, or New Zealand (ACNMGRWP 2008). Germany has had a national screening programme for malignant melanoma since 2008 (Katalinic 2015), and opportunistic screening (i.e. when someone asks their doctor or health professional for screening, or screening is offered by a doctor or health professional outside of an organised screening programme) is increasingly used in many Western countries (Lakhani 2014). In Australia, the annual skin screening rate ranged from 10% to 50% of the adult population depending on how skin screening was defined (Balanda 1994; Borland 1995; Girgis 1991; Heywood 1994; Janda 2004), and the corresponding rate in the US was 14% to 20% (Federman 1997; Federman 2006; Ford 2004; Saraiya 2004). Several professional societies, who may have inherent vested interests, recommends skin screening. In Europe, a campaign involving dermatologists in over 30 countries (EUROMELANOMA) recommended "visiting your dermatologist regularly for a skin check‐up" and conducting self‐examination every month (EADO 2016). In the US, the American Cancer Society recommended a skin self‐examination every month (American Cancer Society 2017) and the American Academy of Dermatology runs a skin screening programme wherein over 2.5 million skin screens have been conducted since 1985 (American Academy of Dermatology 2017).
Description of the intervention
Screening for malignant melanoma can be performed through visual self‐examination of the skin or visual inspection by a general practitioner, dermatologist, or other health professional, which can be followed by dermatoscopy of identified lesions. Other methods to assist in diagnosing malignant melanomas are evolving and might also be used for screening, for example, teledermatology, mobile phone applications, and spectroscopy‐based techniques (Dinnes 2015). The heightened sensitivity that these new methods might confer may increase both the major benefit (a mortality reduction) and the major harm (overdiagnosis) from the intervention. Among general practitioners, the sensitivity of visual inspection has been estimated to be 72% to 84% and specificity to be 70% to 71% (Brochez 2001). However, sensitivity and specificity do not take overdiagnosis into account; therefore, they are less informative in a screening context, where overdiagnosis is a higher concern than for diagnostic tests for symptomatic conditions. A suite of Cochrane Reviews are currently evaluating the accuracy of tests to assist in diagnosing malignant melanoma (Dinnes 2015).
Screening can be organised as programmes where all eligible people in a community are personally invited to screening or as public campaigns where the eligible population is encouraged to participate, for example, through the mass media or advertising.
How the intervention might work
Screening for malignant melanoma has the potential to reduce mortality from the disease through earlier detection, as prognosis is closely associated with the thickness of the lesion at the time of diagnosis. Screening might also result in less‐invasive surgery and less use of adjuvant therapy if the incidence of late‐stage disease is reduced (Welch 2011). For cancer screening to be effective, it must detect more cancers at an early stage and must lead to a lower incidence of late‐stage disease over time (Keen 2015; Vainio 2002). If a decrease in late‐stage disease does not occur, the increase in early‐stage disease may represent detection of lesions that are histologically malignant but would never have caused symptoms or death if they had remained undetected (i.e. overdiagnosis) (Biesheuvel 2007; Welch 2011).
Overdiagnosed malignant melanomas differ from false‐positive findings in that they fulfil the histological criteria for malignancy (Welch 2011). It is not possible to know which specific individuals are overdiagnosed as practically all lesions are removed once they are diagnosed and treated and overdiagnosed individuals will therefore be considered as 'cured' (Biesheuvel 2007). The same is true for the mortality benefit (i.e. it is not possible to know which specific individuals who have avoided death from malignant melanoma due to screening as many fortunately survive also without screening due to treatment (Welch 2011). Overdiagnosis leads to overtreatment, which means that healthy people are exposed to unnecessary surgery and possibly adjuvant therapy. Overdiagnosis also constitutes unnecessary labelling of healthy people with a cancer diagnosis, which may result in psychological harm (Welch 2011). To evaluate the balance between benefits and harms of screening, it is important to consider both rare harms with major effects for few people and common harms with less‐serious effects in many people (Harris 2014; UKNSC 2015).
As it is not always possible to distinguish between benign naevi and malignant melanomas with certainty through visual inspection or dermatoscopy alone, a number of unnecessary biopsies or local excisions of benign lesions will result from screening for malignant melanoma (i.e. false‐positive findings) (Harris 2014; Welch 2005). This may lead to psychological stress in addition to the physical consequences of the excisions (Brodersen 2013). In contrast, malignant lesions may also be missed at screening (false negatives), which may lead to false reassurance and delayed contact with health professionals and thus delayed diagnosis and treatment (Goldenberg 2016). Screening for malignant melanoma may also lead to the discovery of other skin conditions, non‐malignant as well as malignant, and result in treatment for these conditions. The consequences of this can be both beneficial and harmful.
Disease‐specific mortality in cancer screening trials is an outcome prone to bias from misclassification of the cause of death (Gøtzsche 2013; Prasad 2016). Knowledge of the diagnosis increases the risk that the cause of death is falsely attributed to the disease in question although the true cause was another condition (termed sticky‐diagnosis bias) (Black 2002). Conversely, a death can be falsely attributed to another cause, usually because some time has elapsed since diagnosis or because the connection is not always clear (termed slipper‐linkage bias) (Black 2002). Total mortality is free from these and other biases and is therefore the most reliable outcome in cancer screening. The downside is that very large trials are needed to reliably detect a difference, as the effect of cancer screening is small in absolute numbers at the population level (Prasad 2016).
Why it is important to do this review
Screening for malignant melanoma is currently practised in many countries, apparently without support from randomised trials. This is problematic since data from randomised trials demonstrating that benefits outweigh harms is considered mandatory before the introduction of screening programmes for cancer (UKNSC 2015; WHO 2008). If screening for malignant melanoma can contribute to reducing morbidity and mortality from this disease, it is important to clarify gaps in the evidence base so that this can be remedied. Screening for malignant melanoma may cause overdiagnosis of malignant melanomas and consequently overtreatment (Norgaard 2011). False‐positive findings occur, which is known from breast cancer screening to cause substantial long‐lasting psychological stress (Brodersen 2013). In addition, screening for malignant melanoma has a potential for opportunity costs.
A protocol for this review has been published (Johansson 2016).
Objectives
To assess the effects on morbidity and mortality of screening for malignant melanoma in the general population.
Methods
Criteria for considering studies for this review
Types of studies
Randomised controlled trials (RCTs), including cluster‐randomised trials, that compared screening for malignant melanoma with no screening, regardless of screening modality, type of population, or setting.
Types of participants
Any type of population and any age group that is not suspected of having malignant melanoma (i.e. asymptomatic people; those who do not present with a suspicious lesion). We included studies in high‐risk populations such as older men and people with light skin living in countries with high sun exposure. However, we did not include studies in people with a genetic disposition for malignant melanoma (e.g. familial atypical mole and melanoma syndrome), neither did we include studies performed exclusively in participants with previous melanomas as we considered this control monitoring rather than screening and as the benefit/harm balance may differ substantially. However, we did include studies that did not explicitly exclude participants with previous melanomas.
As screening participants should not be invited based on a specific suspicion of malignant melanoma, we excluded studies of diagnostic tests or studies in symptomatic individuals who sought medical attention.
Types of interventions
Screening for malignant melanoma using any type of screening modality in any asymptomatic population and in any setting. We included studies that employed any screening frequency, including once‐only. Screening could be performed by any type of health professional or through skin self‐examination. The control intervention was no screening.
Types of outcome measures
Primary outcomes
-
Total mortality.
-
Overdiagnosis of malignant melanoma (i.e. excess number of malignant melanomas diagnosed in the screening group).
-
Quality of life (QoL)/psychosocial consequences (short‐term: postintervention up to six months; medium‐term: six to 12 months; and long‐term: more than 12 months).
Secondary outcomes
-
Mortality specific to malignant melanoma.
-
False‐positive rates (skin biopsies/excisions with benign outcome).
-
False‐negative rates (malignant melanomas diagnosed between screening rounds and up to one year after the last round).
-
Use of surgery defined as more than local excision (included surgery with lymph node removal).
-
Use of surgery defined as local excision.
-
Use of adjuvant therapy.
-
Incidental findings of other skin conditions (benign or malignant).
-
Use of health services for any reason.
We included studies regardless of whether they quantified our prespecified outcomes or not. We planned to include at least all primary outcomes in our 'Summary of findings' tables.
We planned to quantify total and disease‐specific mortality at five years, ten years, and for the longest follow‐up period available.
Search methods for identification of studies
We aimed to identify all relevant RCTs regardless of language or publication status (published, unpublished, in press, or in progress).
Electronic searches
The Cochrane Skin Information Specialist searched the following databases up to 2 May 2018:
-
the Cochrane Skin Group Specialised Register using the search strategy in Appendix 1;
-
the Cochrane Central Register of Controlled Trials (CENTRAL) 2018, Issue 3, in the Cochrane Library using the strategy in Appendix 2;
-
MEDLINE via Ovid (from 1946) using the strategy in Appendix 3;
-
Embase via Ovid (from 1974) using the strategy in Appendix 4; and
-
LILACS (Latin American and Caribbean Health Science Information database, from 1982) using the strategy in Appendix 5.
Trials registers
We searched the following trials registers up to 21 May 2018 using the terms: melanoma, skin cancer, skin neoplasm, screening, early detection.
-
The ISRCTN registry (www.isrctn.com).
-
ClinicalTrials.gov (www.clinicaltrials.gov).
-
The Australian New Zealand Clinical Trials Registry (www.anzctr.org.au).
-
The World Health Organization International Clinical Trials Registry Platform (ICTRP) (apps.who.int/trialsearch/).
-
The EU Clinical Trials Register (www.clinicaltrialsregister.eu).
Searching other resources
References from included studies
We checked the reference lists of included and relevant studies and reviews for further references to relevant trials. We used Web of Science for citation tracking of key articles.
Searching by contacting relevant individuals
We asked the lead authors of included studies if they were aware of any other published, unpublished, or ongoing studies, or results of studies, that would meet our inclusion criteria.
Adverse effects
We did not perform a separate search for adverse effects of the intervention. However, we searched for data in the included studies, but they did not report on adverse effects.
Data collection and analysis
Some parts of the Methods section of this review were similar to text in other Cochrane Reviews coauthored by KJJ and PCG, predominantly (Krogsbøll 2012).
Selection of studies
Two authors (MJ and KJJ) independently assessed the relevance of all titles and abstracts that were identified through the searches, and assessed full‐text copies of potentially eligible articles. When necessary, the other authors (JB and PCG) resolved disagreements through discussion. Two authors (MJ and KJJ) independently searched reference lists, and one author (MJ) undertook citation tracking (Web of Science) of included articles.
We used Covidence to assess the titles and abstracts that were identified in our searches of the listed databases (Covidence), provided reasons for exclusions, and generated a flow chart. Covidence is an online systematic review platform provided by Veritas Health Innovation Ltd, an Australian not‐for‐profit company.
Data extraction and management
Two authors (MJ and KJJ) independently extracted data from the included trials and entered them into a data extraction form using Covidence (Covidence). One author (MJ) exported the extracted data into Review Manager 5 (Review Manager 2014). We planned to evaluate the data extraction form by pilot testing using a representative sample of studies; however, there was an insufficient number of included studies to be able to do this. When relevant information was missing from the reports, we contacted the study authors.
We extracted the following data from all included trials: study design, type of screening test used, total study duration, number of participants allocated to each arm, gender of participants, number lost to follow‐up for each outcome, baseline comparability, setting, age, country, and date of study start.
We planned to extract the number of events or rates for total mortality, mortality specific to malignant melanoma, overdiagnosis, false positives, false negatives, surgical interventions defined as more than local excision, surgical interventions defined as local excision, and adjuvant therapy. For psychosocial consequences or QoL outcomes, we planned to extract the mean; standard deviation or standard error; and name, range, and direction of the scale used.
Assessment of risk of bias in included studies
We used Cochrane's 'Risk of bias' tool to formally assess the following domains, as described in Higgins 2011: sequence generation; allocation concealment; blinding of participants and personnel; blinding of outcome assessment; incomplete outcome data; selective reporting; and other biases, including the degree of contamination of the control group by searching for data on the rate of opportunistic screening in the control group.
We also planned to use the Outcome Reporting Bias in Trials (ORBIT) tool to assess outcome reporting bias (Kirkham 2010). We planned to assess the randomised groups for baseline comparability. We planned to use GRADE to assess the level of confidence in individual outcomes (Schünemann 2013).
Measures of treatment effect
For total mortality, mortality specific to malignant melanoma, overdiagnosis, false positives, false negatives, surgical interventions defined as more than local excision, surgical interventions defined as local excision, and adjuvant therapy, we planned to calculate the risk ratios (RR) and the risk differences (absolute risks). We planned to calculate standardised mean differences (SMD) for QoL outcomes if different scales were used and the scales were comparable. If the same scales were used in all studies, we planned to calculate the mean difference (MD). For all measures, we planned to calculate 95% confidence intervals (CI). We planned to define false positives as the rate of biopsies and local excisions with benign results in the intervention arm in the included trials.
Unit of analysis issues
For cluster‐randomised trials, we planned to use effect estimates and standard errors from analyses that took clustering into account. When such estimates were not available, we planned to explore the possible effect of clustering in a sensitivity analysis.
Dealing with missing data
We planned to conduct analyses as intention‐to‐treat (ITT), when possible. We planned to contact authors if the reports did not contain sufficient data for this. If ITT analyses were not possible, we planned to undertake available‐case analyses and assess the possible bias resulting from dropouts and losses to follow‐up in best‐case or worst‐case analyses for all primary outcomes.
Assessment of heterogeneity
We planned to assess clinical and methodological differences between the trials before any meta‐analyses were performed, and to judge whether we could pool trial results. We planned to explore statistical heterogeneity using the I² statistic. If we had found I² values above 30%, we planned to explore causes of heterogeneity in sensitivity analyses and subgroup analyses, and we planned not to present pooled results if we encountered unexplained heterogeneity that would render the pooled results uninformative.
Assessment of reporting biases
We intended to create funnel plots if more than 10 trials were found. Otherwise, we planned to narratively evaluate outcome reporting bias for individual outcomes in any of the included trials and explore this using the ORBIT tool (Kirkham 2010).
Data synthesis
If we had judged meta‐analyses to be appropriate, we planned to use a random‐effects model if there were substantial differences between the populations included; trial designs; and the type of, or frequency and number of, screens offered.
We planned to apply the trial sequential analysis model to the dichotomous outcomes (Brok 2008), but not to the continuous outcomes because the trial sequential analysis currently assumes MDs and not SMDs, which we expected was necessary to use. It is a statistical model, similar to interim analyses in clinical trials, used to quantify the reliability of data in cumulative meta‐analyses, adjusting the P values for sparse data and multiplicity. The required information size (the number of participants required to accept or reject the hypothesis of a certain a priori anticipated effect) is calculated using the following five components.
-
Alpha = 0.05 (type 1 error).
-
Power = 0.90 (type 2 error 0.10).
-
Proportion (frequency) of participants experiencing serious adverse events and adverse events (based on observations).
-
Relative risk reduction (RRR) or increase of 20%.
-
Diversity (heterogeneity based on our observations).
Preferably, this model should be applied to trials with a low risk of bias only, but we planned to conduct analyses that also included trials with high risk of bias.
Where results were estimated for individual studies with low numbers of outcomes (fewer than 10 in total) or where the total sample size was fewer than 30 participants and an RR was used, we planned to report the proportion of outcomes in each group together with a P value from a Fisher's Exact test.
Subgroup analysis and investigation of heterogeneity
We planned to perform subgroup analyses for the following groups:
-
light versus dark skin;
-
young versus old people;
-
women versus men;
-
high‐risk versus low‐risk countries;
-
screening by specialists (i.e. dermatologists or in screening units by specially trained staff) versus usual care (e.g. general practitioners); and
-
high‐intensity versus low‐intensity screening.
Sensitivity analysis
We planned to perform sensitivity analysis for studies with high versus low overall risk of bias concerning the randomisation process and blinded outcome assessment. If results differed between studies with high and low risk of bias, we planned to rely on studies with low risk of bias.
We planned to conduct a sensitivity analysis of any included studies that were prospectively registered in trials registers.
'Summary of findings' table
We included a 'Summary of findings' table in our review summarising our three primary outcomes and three of our secondary outcomes (mortality specific to malignant melanoma, false‐positive rates, and false‐negative rates). We planned to assess the certainty of the evidence using the five GRADE domains (risk of bias, inconsistency, imprecision, indirectness, and publication bias) (Higgins 2011; Schünemann 2013).
Results
Description of studies
Results of the search
The searches of the five databases (see Electronic searches) retrieved 780 records. Our searches of other resources identified 17 additional studies that appeared to meet the inclusion criteria (14 identified through reviews of reference lists, two from citation‐tracking, and one from a trials register). Therefore, we had a total of 797 records.
We excluded 775 records based on titles and abstracts. We obtained the full text of the remaining 22 records. We excluded 16 of these studies (see Characteristics of excluded studies table). We identified no ongoing studies and no studies are awaiting classification.
We included two studies reported in six publications. For a further description of our screening process, see the study flow diagram (Figure 1).
Study flow diagram.
Included studies
We included two studies. The first study was a randomised trial of an intervention developed to increase performance of thorough skin self‐examination (TSSE), "the Check‐It‐Out Project" (Weinstock 2007). Study participants were recruited from 11 primary care practices in the US between 2000 and 2001. All participating primary care clinicians attended a workshop prior to initiation of recruitment in their clinics. The workshop focused on early detection of skin cancer. In total, 2126 people scheduled for a routine primary care visit were interviewed by telephone prior to that visit (i.e. the baseline interview). At the time of the visit, and after seeing the primary care clinician, a health educator randomised 1356 participants into either an intervention group (688 participants) or control group (668 participants). The mean age was 53.2 years, and 41.7% were men. The intervention group received educational materials, cues, aids, and a brief counselling intervention by a health educator aimed at increasing performance of TSSE for early detection of melanoma and other skin cancers. The educational materials advocated monthly TSSE with physician consultation for any new or changing skin lesions. The control group received a diet intervention, which was intended to control for degree of contact with healthcare personnel. All participants were asked to complete follow‐up telephone interviews at 2, 6, and 12 months after randomisation. The main outcome was rate of performance of TSSE. Rate of skin surgeries was retrieved from medical records for those participants who reported a procedure in follow‐up telephone interviews.
The second study included was a cluster‐randomised pilot study performed in Australia where nine communities including 35,058 adults who were 30 years or more were randomised to a three‐year melanoma screening programme and nine communities served as controls, including 27,977 adults aged 30 years or more (Aitken 2002). The programme was implemented between 1998 and 2001. The purpose of the programme was to promote annual whole‐body skin examination performed by medical practitioners, defined as visual examination of the skin, excluding areas covered by underwear, for early signs of skin cancer. The programme also encouraged regular whole‐body skin self‐examination and presentation of suspicious lesions to a doctor. The programme had three main components: 1. a community education component; 2. an education and support component for medical practitioners aiming to improve their skills in early diagnosis and management of skin cancer as well as to encourage doctors to offer skin screening to their patients; and 3. the provision of free skin screening services to which personal invitations for screening were posted to residents aged 30 years or more. Screening clinics were provided by primary care physicians and held in workplaces, community venues, and local hospitals and included day and evening sessions. The mean age of those attending the skin screening clinics was 46.5 years, and 51.5% were men; the study did not report the mean age or proportion of men/women in the whole study (communities) population. Outcomes from the screening clinics were measured between 1998 and 2001. There was no further follow‐up for any outcomes. The originally planned outcomes were mortality from malignant melanoma in the inception cohort (i.e. all residents above 30 years of age) (primary outcome). Other outcomes included incidence of melanoma by tumour thickness, the impact of the intervention on the diagnosis and treatment of skin lesions, the proportion of the population undergoing skin screening, and cost outcome measures.
The plan was to expand the trial to include 44 communities (aggregate population of 560,000 adult men and women aged 30 years or more) (Aitken 2002), but due to lack of funding this trial was never initiated (Aitken 2017 [pers comm]).
Excluded studies
We excluded 16 publications after obtaining the full text following our review of titles and abstracts. None of these were randomised trials of the effects of screening for malignant melanoma on morbidity or mortality. The main reasons for exclusion was a non‐randomised study design or an ineligible control group (e.g. studies where the control group received instructions for skin self‐examination). The most relevant excluded study was the SCREEN (Skin Cancer Research to Provide Evidence for Effectiveness of Screening in Northern Germany) study; an observational study comparing trends in melanoma mortality in regions with and without screening programmes in Germany (Breitbart 2012; Katalinic 2012). We describe this study in detail in the Discussion and Agreements and disagreements with other studies or reviews sections.
Risk of bias in included studies
Figure 2 and Figure 3 summarise the risk of bias of the two included studies.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Allocation
In "the Check‐It‐Out Project", randomisation was performed by participant selection of an opaque envelope with the assignment enclosed (Weinstock 2007). The risk of bias was low.
In the cluster randomised pilot study, randomisation of communities was within pairs (Aitken 2002). There was no information regarding sequence generation or allocation concealment. The risk of bias was unclear.
Blinding
In "the Check‐It‐Out Project", the interviewers who retrieved data on reported skin surgeries through telephone interviews with participants were not informed of the respondent's intervention assignment prior to any of these interviews (Weinstock 2007). However, intervention assignment could arguably have been revealed during the interviews. There was no information on whether the people who retrieved data on skin surgeries from medical records were blinded to intervention status. Blinding of participants was not possible due to the nature of the intervention. The risk of performance bias would therefore be high and the risk of detection bias unclear.
In the cluster randomised pilot study, blinding of screening participants and healthcare personnel was not possible due to the nature of the intervention (Aitken 2002). Outcome assessment was planned to be registry based. We found no information on a plan for a blinded outcome assessment panel. The risk of bias would therefore be high, but no outcomes were reported.
Incomplete outcome data
In "the Check‐It‐Out Project", risk of attrition bias was low because there was a similar dropout rate across groups (Weinstock 2007). In the cluster randomised pilot study, no data on the prespecified outcomes of this review were reported; therefore, the risk of bias was unclear (Aitken 2002).
Selective reporting
In "the Check‐It‐Out Project", the risk of reporting bias was considered low because all preplanned outcomes were reported (Weinstock 2007). In the cluster randomised pilot study, no data on the prespecified outcomes of this review were reported; therefore, the risk of bias unclear (Aitken 2002).
Other potential sources of bias
Weinstock 2007 was at low risk of bias because no other sources were identified. In the cluster randomised trial, the few clusters inferred another source of bias (Aitken 2002). Trials with few clusters are at high risk of bias because there are few units of randomisation. Thus, random error is likely and the risk of bias is therefore high.
Effects of interventions
There were some data from "the Check‐It‐Out Project" on the prespecified secondary outcomes of this review: 'use of surgery defined as local excision' and 'incidental finding of other skin conditions (benign or malignant)' (Weinstock 2007). However, this trial was not designed to answer our review question and the available data, therefore, had limited relevance when estimating the effect of screening on our prespecified outcomes. Due to the lack of data on indications for skin surgeries (i.e. suspicion of melanoma, or suspicion of other malignant or benign skin lesions), it was not possible to estimate the rate of false‐positives based on the published data.
For the outcome 'incidental finding of other skin conditions (benign or malignant)', data were limited to few events; there were two severely atypical nevi (one in each group), seven squamous cell carcinoma (three in screening group, four in control group), and 10 basal cell carcinomas detected (seven in screening group, three in control group). Additionally, there were no data reported for benign skin conditions detected through screening.
There were 82 skin surgical procedures in the screening group and 46 in the control group over 12 months. Skin surgery was defined as "biopsy, cut or freeze", and there were no data on the proportion of either. Further, "skin surgeries were determined by examination of medical records of patients who reported a procedure". This may have introduced the risk of recall bias because people who have just received an intervention to increase skin self‐examinations might have a higher probability of remembering going through skin surgery than people who have not received such interventions. We did not find it meaningful to estimate the effect of screening on "use of surgery defined as local excision" based on the published data. Finally, there were no malignant melanoma detected in the screening group and only one malignant melanoma detected in the control group. In conclusion, based on the data from this trial, it was not possible to assess our prespecified outcomes.
From the cluster randomised pilot study, there were no data with comparisons between the screening and control communities for the prespecified outcomes of this review (Aitken 2002), and none are to be expected (Aitken 2017 [pers comm]). This information was retrieved after contact with study authors. One publication presented data on the number of suspicious lesions and excisions performed in people screened at the screening clinics within the screening programme; in 15,343 people screened, there were 4129 suspected lesions and 14% of all screening examinations resulted in referral for at least one suspected lesion. Thirty‐three histopathologically confirmed malignant melanomas were diagnosed as well as one Hutchinson's melanotic freckle, which constituted 2% of all excised lesions. Due to the lack of data on rates of referral and diagnoses in the control arm and from screening examinations performed due to the screening programme but not at the screening clinics, and because 64% of those referred for suspected lesions after screening were already concerned about a specific skin lesion prior to the screening examination, it was not possible to estimate the rate of false positives based on the published data. The authors compared the percentage distribution according to thickness for the screen‐detected melanomas to melanomas diagnosed in Queensland as a whole in the prescreening period from 1999 to 2002 and found a lower percentage of thick melanomas among the screen‐detected cases. However, when a cancer screening programme may lead to the overdiagnosis of predominantly small invasive cancers and in situ lesions, such percentage distributions could be highly misleading and were, therefore, uninformative.
Discussion
Summary of main results
We found no data from randomised trials to support or refute an effect of screening for malignant melanoma on morbidity or mortality (summary of findings Table for the main comparison).
We identified one randomised trial of an intervention developed to increase the frequency of TSSE that met our inclusion criteria, but did not provide any data on our primary outcomes (total mortality, overdiagnosis of malignant melanoma, QoL/psychosocial consequences). The available data on our secondary outcomes were limited by few events and poor reporting. For our outcome Incidental finding of other skin conditions (benign or malignant), there were two severely atypical nevi (one in each group), seven squamous cell carcinoma (three in screening group, four in control group) and 10 basal cell carcinomas detected (seven in screening group, three in control group). Our key secondary outcomes (mortality specific to malignant melanoma, false‐positive rates, false‐negative rates) were not measured. It was not possible to estimate an effect of screening on any of our prespecified outcomes based on data from this trial.
Additionally, we identified one cluster‐randomised pilot study for a larger, population‐based cluster‐randomised trial, but no results for our predefined outcomes were reported or were available in unpublished format, and the main trial was never initiated due to lack of funding (Aitken 2017 [pers comm]). In the pilot study, every seventh person screened had at least one suspicious lesion, while only one malignant melanoma was found for every 465 people screened and only 2% of all excised lesions constituted malignant melanomas. This indicated that many unnecessary biopsies were performed and substantial opportunity costs by screening for malignant melanoma both regarding health personal involved in the screening examinations and resources for histopathological investigation, which means economical resources were made unavailable to interventions with proven benefit.
Overall completeness and applicability of evidence
Despite extensive searches, we found no data from randomised trials on the primary outcomes of this review and very limited data on two secondary outcomes (from one trial). This review did not investigate the effects of screening people with a history of malignant melanoma or in people with a genetic disposition for malignant melanoma (e.g. familial atypical mole and melanoma syndrome).
Quality of the evidence
Aitken 2002 was at high risk of bias for blinding and study design (cluster RCT) and unclear risk for the remaining domains. Weinstock 2007 was at high risk of performance bias and unclear risk for detection bias; the remaining domains were at low risk. We did not use GRADE to assess the quality of evidence because neither included study measured our primary outcomes or key secondary outcomes.
Potential biases in the review process
There were no data to assess.
We did not include non‐randomised studies but we did report and discuss the result of some non‐randomised studies that have had influence on screening policy in the Agreements and disagreements with other studies or reviews section. Since we did not perform a systematic search for non‐randomised studies, this should not be viewed as a systematic or exhaustive summary of the evidence from non‐randomised studies of the effects of screening for malignant melanoma.
We chose to include both studies of screening performed by a healthcare professional and skin self‐examination because the mechanism of effect is similar (earlier detection). However, there may be differences regarding the magnitude of both benefit and harm depending on whether the screening is done by oneself or performed by a healthcare professional.
There are several methods to estimate overdiagnosis in cancer screening (Biesheuvel 2007). A more detailed description of these methods is outside the scope of this review, but randomised trials with long follow‐up would be highly desirable to quantify this outcome.
Agreements and disagreements with other studies or reviews
The conclusions of this review are in accordance with recommendations from the US Preventive Services Task Force (USPSTF 2016), as well as official bodies in Canada (CTFPHC 2013), Australia and New Zealand (ACNMGRWP 2008), of which none recommend screening for malignant melanoma in the general population based on a lack of evidence for a beneficial effect of the intervention. The US Preventive Services Task Force concluded that the current evidence is insufficient to assess the balance of benefits and harms of visual skin examination by a clinician to screen for skin cancer in adults. Like the findings of our review, the US Preventive Services recommendation does not apply to people with familial atypical mole and melanoma syndrome, neither to people with previous melanoma.
Professional organisations, such as EUROMELANOMA (EADO 2016), the American Cancer Society (American Cancer Society 2017), and the American Academy of Dermatology (American Academy of Dermatology 2017), recommend skin screening, and some have argued for screening elderly, white men due to their increased risk of malignant melanoma (Coups 2010).
Evidence from non‐randomised studies
An apparent effect of screening on melanoma mortality and on the rate of thick melanomas observed in non‐randomised studies has been used in the argument for melanoma screening (Breitbart 2014; Curiel‐Lewandrowski 2012; Geller 2015; McCleskey 2015; McFarland 2015; Robinson 2016a; Shellenberger 2016; Wainstein 2015). We did not search systematically for non‐randomised studies and did not include them in our review. However, considering their potential and current impact, we decided to describe and comment on some of the most influential non‐randomised studies.
The SCREEN study
The SCREEN study was the direct reason for implementation of melanoma screening in Germany, the world's only national, organised screening programme for malignant melanoma (Breitbart 2012; Katalinic 2012). The study, conducted in Schleswig‐Holstein in Germany, compared trends for melanoma mortality and melanoma incidence before, during, and after the screening programme to trends in adjacent German and Danish regions with no screening programmes. The screening programme consisted of 1. an advertisement campaign (described in Breitbart 2012) aimed at the public to consult their doctor about primary and secondary preventive activities for skin cancer, and 2. eight‐hour training courses where physicians were trained to actively inform and recruit people for skin cancer screening. The screening programme was gradually developed and implemented:
-
2000 to 2001: a pilot project with courses for 200 physicians; 6000 people were screened;
-
2001 to 2003: skin cancer awareness campaigns;
-
2003: courses for 1673 (out of 2614) physicians with outpatient activities and 116 (out of 118) dermatologists in the region; and
-
2003 to 2004: population‐based once‐only skin screening of 360,288 people.
The screening region had 2.8 million inhabitants, of whom 1.9 million met the eligibility criteria for screening within the programme (aged 20 years or more and being a policy holder of statutory health insurance (which applies to approximately 85% of the German population)). Physicians were paid approximately EUR 20 for every performed screen. Nineteen per cent of the eligible population were screened within the programme during the one‐year main screening period (2003 to 2004). The male:female ratio for those screened was 1:3 (Breitbart 2012). The lowest participation rate was in people aged 70 years or more (12%). Fifty‐two per cent of all melanomas diagnosed during the screening period were detected as part of the project. Data on the incidence of melanoma was extracted from the State Cancer Registry and data on mortality from melanoma was extracted from official mortality statistics. Analyses of incidence and mortality were based on the whole population (i.e. not only on the people screened).
The study showed that the age‐standardised mortality rate of melanoma in the screening region decreased from 1.7 per 100,000 (95% CI 1.4 to 2.0) to 0.9 per 100,000 (95% CI 0.7 to 1.1), that is, a 48% reduction from the prescreening period (1998 to 1999) to five years postscreening (2008 to 2009), and almost equally in both sexes. In the adjacent regions and in the rest of Germany, the mortality rates from melanoma were stable.
We evaluated the risk of bias in the SCREEN study using the ROBINS‐I tool (Risk Of Bias in Non‐randomized Studies – of Interventions) (Sterne 2016). We found a low risk of bias with unpredictable direction in the following domains; 'bias in selection of participants into the study', 'bias in classification of interventions', and 'bias due to missing data'. We found a low risk of bias favouring controls in the domain 'bias due to deviations from intended interventions'. We found a moderate risk of bias with unpredictable direction for the domain 'bias due to confounding'. We found a serious risk of bias favouring the screened group for the domains 'bias in measurement of outcomes' and 'bias in selection of the reported results'. We judged the overall risk of bias to be serious and favouring screening. For further details, see the ROBINS‐I form (Appendix 6).
In addition to the overall serious risk of bias, further data suggested that the validity of the results from the SCREEN study was questionable.
First, the five‐year follow‐up showed a decrease in melanoma mortality of almost 50% (Katalinic 2012), but after those five years, the mortality from melanoma increased rapidly in the screening region, and in 2012 to 2013 mortality rates were close to the rates observed before the SCREEN project and the same as those seen in the rest of Germany (Boniol 2015). Additionally, five years after the implementation of a nationwide skin screening programme in Germany, there was no beneficial effect (Katalinic 2015), as would be expected at this time based on the very large reduction seen in the SCREEN study. It has been argued that this disparity in outcomes was due to screening activities within the nationwide screening programme being less intensive than they were in the SCREEN study (Katalinic 2015). However, increases in melanoma incidence at the time of screening implementation were similar in the national screening programme (29% increase; Boniol 2015) and the SCREEN study (34% increase; Breitbart 2012). In addition, participation rates in the nationwide programme was reported as 31% in 2009 and 2010 (i.e. two years) (Boniol 2015), compared to 19% during one year in the SCREEN study. While the SCREEN study was a one‐year project, the nationwide programme offers screening continuously which would also increase the chance of seeing a benefit. In conclusion, this indicates that the disparity in the mortality reduction seen in the SCREEN study and in the nationwide screening programme is unlikely to be explained by different intensity of screening and is more likely due to bias in the SCREEN study.
Second, in the SCREEN study, the participation rate was only 19% while the mortality reduction was almost 50%. This discrepancy could be explained by screening activities outside the programme, or a higher risk of melanoma death among people who attended screening compared to non‐attendees. However, in the SCREEN project, 74% of people attending screening were women and participation rates were lowest in older age groups. Elderly men have the highest risk for mortality from melanoma, thus implying that the screening programme actually did not attract those people with the highest risk. Indeed, reductions in melanoma mortality were practically identical in men and women, despite only 10% of men participating compared to 27% of women. This raised a suspicion of systematic error rather than a true screening effect, especially as a 50% reduction based on 10% participation in men in itself seems highly unlikely, even accounting for strong self‐selection bias.
Third, the decrease in melanoma mortality in the screening region started as early as 2002, that is, one to two years before the implementation of the main screening programme and only one to two years after the pilot project. Even if some of the mortality reduction would be attributable to the pilot project, it is still a remarkably prompt effect on population statistics given that only 6000 people participated in the pilot project out of an eligible population size of 1.9 million. Nothing similar has been seen in previous cancer screening programmes (Stang 2016a), and, considering the likely lead‐time and time from a clinical diagnosis of melanoma to death from the disease, such a prompt effect on mortality seems biologically implausible.
Last, the reduction in mortality from malignant melanoma in the screening region was accompanied by a simultaneous substantial increase in deaths from malignant neoplasms of ill‐defined, secondary and unspecified sites (Stang 2016b). Such trends were not observed in any of the adjacent regions. An incorrect counting of approximately 37 melanoma deaths per year in the screening region between 2007 and 2010 could explain the entire decline in melanoma mortality seen in the SCREEN study. The greatest reduction in melanoma mortality was seen in outpatient deaths, which are more prone to misclassification (Stang 2016b). It has been hypothesised that physicians practising in the screening region under‐reported melanoma as a cause of death, as cause of death assessment was not blinded to screening status (Boniol 2015).
In conclusion, the transient decline in melanoma mortality observed in the SCREEN study was most likely not due to screening but differential misclassification of cause of death. Indeed, the available data from Germany indicated that organised screening did not affect mortality from malignant melanoma in a 5‐ to 10‐year time frame, whereas it led to substantial increases in incidence, suggesting that overdiagnosis and overtreatment occurred. A longer follow‐up without increased melanoma mortality would further support this conclusion.
Effect on rate of thick melanomas
Several non‐randomised studies suggested a beneficial effect of screening through a decrease in the incidence of thick melanomas. For example, in one case‐control study from Australia including over 3762 cases and 3824 controls, whole‐body clinical skin examination in the three years before diagnosis was associated with a 14% lower risk of being diagnosed with a thick melanoma (greater than 0.75 mm) (odds ratio (OR) 0.86, 95% CI 0.75 to 0.98) (Aitken 2010). The reduction in risk of a diagnosis was greater for thicker melanomas: by 7% for melanomas 0.76 mm to 1.49 mm thick (OR 0.93, 95% CI 0.79 to 1.10; not statistically significant), by 17% for melanomas 1.50 mm to 2.99 mm thick (OR 0.83, 95% CI 0.65 to 1.05; not statistically significant), and by 40% for melanomas greater than 3 mm thick (OR 0.60, 95% CI 0.43 to 0.83). Screening was associated with a 38% higher risk of being diagnosed with a thin invasive melanoma (less than 0.75 mm) (OR 1.38, 95% CI 1.22 to 1.56) (Aitken 2010). In case‐control studies using self‐reported exposure, there is a risk of recall bias. In this case, recall bias would favour screening. However, it is less likely that recall bias would result in a gradient such as that seen for melanoma thickness. People who choose to participate in screening are often healthier and lead healthier lives (Raffle 2007). They are more likely to seek medical care and so their cancer is more likely to be detected earlier, even in the absence of screening (Raffle 2007). Therefore, screening attendees also often have a better prognosis when diagnosed than other people due to differences in risk factors, socioeconomic status, and disease awareness. This phenomenon is called 'the healthy screenee effect' (Raffle 2007). In this case‐control study, bias due to the healthy screenee effect may have explained the correlation between screening and melanoma thickness.
Although melanoma tumour thickness is the most important prognostic factor (Shaikh 2016), the rate of thick melanomas should be considered a surrogate outcome. This is because melanoma thickness is not a benefit in its own right but only relevant if translated into an effect on patient‐relevant outcomes, such as less aggressive treatment over time or reduced morbidity or mortality from the disease (Hudis 2015). While melanoma tumour thickness is correlated with prognosis, we cannot be certain that earlier detection through screening will change prognosis. The correlation may be due to the biology of the individual tumour rather than causal.
Study flow diagram.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
| Screening compared with no screening for malignant melanoma | |
| Patient or population: asymptomatic people Settings: any setting Intervention: screening Comparison: no screening | |
| Outcomes | Comments |
| Total mortality | Not measured |
| Overdiagnosis of malignant melanoma | Not measured |
| Quality of life/psychosocial consequences | Not measured |
| Mortality specific to malignant melanoma | Not measured |
| False positive rates | Not measured |
| False negative rates | Not measured |
