Sage Journals: Discover world-class research

Abstract

Objective

To analyse time trends in mortality due to cutaneous malignant melanoma (CMM) in Spain between 1980 and 2020, and to identify possible spatial clusters of provinces with an excess risk of CMM mortality during the period 2011–2020.

Methods

Joinpoint regression models were used to assess temporal trends in data from the Spanish National Institute of Statistics. Crude rates, standardized mortality ratio (SMR), smoothed relative risk (RR), and posterior probabilities (PP) of RR >1 during the period 2011–2020 were calculated. The Global Moran I index was used to assess global spatial autocorrelation.

Results

Two time-periods were detected in women: a significant increase during 1980–1994 (annual percent change [APC], 6.5% per year) and rate stabilization during 1994–2020 (nonsignificant APC, −0.17%). A similar pattern was observed in men, with three periods comprising a significant increase during 1980–1985 (APC, 16.59%), a slowing of the increase during 1985–1998 (APC, 4.40%), and stabilization during 1998–2020 (nonsignificant APC, 0.37%). Spatial analysis showed greater spatial heterogeneity with an east-north pattern in men compared with the pattern in women, which tended to be concentrated in north-western areas.

Conclusion

Mortality rates associated with CMM in Spain have remained stable in recent years. There were provincial clusters that exhibited an excess risk of mortality from CMM, with different patterns according to sex.

Keywords

Melanoma epidemiology mortality ecological study skin cancer

Introduction

In 2020, about 325 000 people worldwide (174 000 men and 151 000 women) were diagnosed with cutaneous malignant melanoma (CMM), and around 57 000 (32 000 men and 25 000 women) died from the disease.¹ If the rates are maintained, the burden of CMM is expected to increase to 510 000 new cases (approximately 50% increase) and up to 96 000 deaths (approximately 68% increase) by 2040.¹

Overall trends in CMM are increasing in terms of incidence and mortality,^2,3 but there are national and regional exceptions, with great variability between countries (incidence rates vary six-fold across EU-27 countries, while mortality rates vary three-fold),⁴ and also between sexes and age groups. In addition, the persistent gender mortality gap persists in most countries and has increased over time.⁵

Since 2013, the mortality rate of CMM has declined rapidly in the USA,⁶ Canada,⁷ and Germany,⁸ with the introduction of new systemic treatments together with immune checkpoint inhibitors and targeted therapies for metastatic melanoma.⁹

Previous studies have analysed the trends of CMM mortality in Spain, at a national level and in some autonomous communities.^10–13 These studies showed that mortality rates sharply increased from the early 1970s to the early 1990s, followed by a stabilisation observed in the age-standardized rates for all ages. Analysis of mortality trends by age group and birth cohort showed a slower increase (men) or stabilisation (women) in age-adjusted CMM mortality rates, due to the decrease in mortality registered in younger age groups and recent cohorts. If current rates persist, overall rates are expected to grow in the coming years, although rates among young people will progressively stabilise or decline.¹⁴ Despite this, information on the geographical distribution of CMM mortality is scarce in Spain.¹⁵

Visualisation and cartographic analysis of events during a time-period have been considered as an essential instrument for disease monitoring. The identification of clusters is also important in CMM epidemiological studies.¹⁶ When a statistically significant excess of cases is observed, subsequent epidemiological studies can investigate whether the cluster is related to the environment, health care contact, or other characteristics of the resident population, and may explain previously unknown aetiologies. In addition, knowledge of the geographical patterns of CMM in a population can be used to more effectively target screening and prevention efforts.¹⁷ Therefore, the aim of the present study was to analyse the temporal trends in CMM mortality in Spain between 1980 and 2020, and to identify possible spatial clustering of CMM mortality at the provincial level between 2011 and 2020.

Materials and methods

Data source

For this observational study, data on deaths due to CMM and the population required for the calculations were obtained from the Spanish National Institute of Statistics (INE).¹⁸ All deaths with CMM codes 172 and C43 of the International Classification of Disease (ICD-9th and ICD-10th revision) were analysed. The study was not deemed to require patient consent or ethics committee approval, as all data extracted from the National Institute of Statistics were anonymized, the study followed the principles of good clinical practice and the Declaration of Helsinki (as revised in 2013), no participants were identified, and no personal information was accessed. The study was performed according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Statistical analyses

To analyse time trends, age-standardized mortality rates (ASMR) were calculated in both sexes for the period 1980–2020 according to the direct method, taking the European Standard Population as a reference.¹⁹ Rates are presented per 100 000 inhabitants and were analysed by Joinpoint regression with a maximum of three possible points of change in the trend, using a segmented Poisson model with National Cancer Institute Joinpoint regression software, version 4.9.1.0.²⁰

To examine the spatial pattern of CMM mortality in both men and women at the provincial level in Peninsular Spain, the crude rates, standardized mortality ratio (SMR), smoothed relative risk (RR), and posterior probabilities (PP) of RR > 1 during the period 2011–2020 were calculated.

To calculate the SMR, the expected cases were estimated for each province, taking the rate of Spain as a reference. Next, the SMR with the number of observed cases was computed as the numerator and the number of expected cases for each province was computed as the denominator. The SMR measures the RR of a province regarding the reference region.

Regarding spatial data analysis, different statistical tests were used to assess overdispersion, spatial autocorrelation and overall spatial clustering. The standard χ²-test and Potthoff-Whittinghill test were used to estimate the spatial heterogeneity of the risk scores and to test for overall significant differences of observed and expected cases. The existence of global spatial autocorrelation was assessed with the Global Moran I index, determining whether the analysis units tended to cluster, scatter, or randomise. The local indicators of spatial association (LISA) statistics define the spatial patterns of hotspots/clusters of CMM mortality at the provincial level. The LISA reveal significant hot spots (high values next to high [HH]), cold spots (low values next to low [LL]) and spatial outliers (high among low [HL] or low among high [LH]) of mortality rates. Moran’s index, with a significance level of 0.05, was used to assess whether the rates showed distributions with a propensity to clustering, dispersion or randomisation.

The RRs were estimated for each province. Structured (clustered) and unstructured (heterogeneous) spatial dependence were modelled using the proposals of Besag, York and Mollié.²¹ Spatial dependence was assumed whereby risk estimates in a given area depended on neighbouring areas. Since there were no neighbouring areas, the island provinces and the cities of Ceuta and Melilla were excluded. First or second order ‘random walks’ (RW1, RW2) were considered to model the temporal effects. Models that included one of the possible spatiotemporal interactions proposed by Knorr-Held were also evaluated.²² Model fitting and inference were performed by means of the approximate Bayesian inference ‘Integrated Nested Laplace Approximation’ (INLA) technique. Model selection techniques, such as Deviance Information Criterion (DIC), were used to identify the most suitable model for capturing the temporal trend in mortality rates across Spanish provinces.²³ If the analysis selected a model with an RW1 term for the time effect, the changes in mortality rates were random fluctuations from the previous period. Conversely, if a model with an RW2 term was chosen, the changes depended on both the immediate past change and the change from two periods earlier, potentially indicating a more complex temporal trend. The model with the smaller DIC value was chosen. One strength of the INLA method is that it may be executed within free R software through the R package ‘R-INLA’,^24,25 which allows calculations of: (i) accurate estimates of the spatiotemporal pattern of risk (RRs), based on data from neighbouring areas and adjacent years; and (ii) the posterior probabilities that the RRs were >1 (PP).

To identify areas with a higher-than-expected mortality rate, decision rules based on the estimated probability that the specified RR is >1, and the following thresholds of 0.05, 0.20, 0.80 and 0.95,²⁶ were applied. Areas with a value >0.80 were considered at risk, and excess mortality was significant if the probability of death was >0.95. Probabilities between 0.2 and 0.8 were considered to represent only weak indications that the percentage ratio is >1, so that the specific mortality rate in these provinces is analogous to the reference mortality rate. Areas <0.20 were considered low-risk areas, and areas with a probability <0.05 were considered to be areas with a specific mortality rate significantly below the reference rate.

The maps were created with the open-source software tool for exploratory spatial data analysis GeoDa, version 1.20 for Windows.²⁷

Results

Age-standardized melanoma mortality rates and the time trend by sex for the general population in Spain are shown in Figure 1. Although a significant increase of 1.09% per year was observed in women during the overall period, Joinpoint analysis detected a trend change with two periods: 1980–1994, marked by a significant increase in melanoma mortality (APC, 6.5% per year; P < 0.05) and 1994–2020, in which rates stabilized (APC, −0.17%; P > 0.05). A similar pattern was observed for men. During the whole period, there was a significant increase in melanoma mortality (1.76% per year), but Joinpoint analysis detected two points of change that determined three time-periods: The first period showed a significant increase between 1980–1985 (APC, 16.59%; P < 0.05), followed by a slower statistically significant increase between 1985–1998 (APC, 4.40%; P < 0.05) and finally a stabilization of melanoma mortality rates between 1998–2020 (APC, 0.37%; P > 0. 05). Compared with the general population trend, the age standardized melanoma mortality rate was higher in males for the whole period.

Figure 1.

Time trend of age-adjusted melanoma mortality rates in Spain between 1980 and 2020.

The spatial distribution of crude rates, the SMRs, the posterior median estimates (RRs), and the PPs in men and women, respectively (for mainland Spain 2011–2020) are shown in Figures 2 and 3. For both sexes, the lowest DIC value corresponded to the spatiotemporal model with a type IV interaction and an RW1 for the time effect. Type IV interaction considers a structure in both space and time and is appropriate when time trends vary from one province to another but are similar to that of the adjacent provinces.

Figure 2.

Spatial distribution of crude rates (per 100 000 inhabitants), standardized mortality rates (SMRs), posterior median estimates (relative risk [RR]), and posterior exceedance probabilities (PP) for melanoma mortality in women (mainland Spain 2011–2020).

Figure 3.

The maps obtained showed consistent results when applying the different spatial analysis methods. The pattern in men showed greater spatial heterogeneity with an east-north pattern, compared with the pattern in women, which tended to be concentrated in north-western areas. All spatial tests were statistically significant (P < 0.05) for both males and females.

In women, seven of the 47 studied provinces were identified to show an excess of melanoma mortality, with PP ≥ 80%, all of which were located in the north on the Cantabrian coast (A Coruña, Lugo, Asturias, Cantabria and Bizkaia) or close to it (Palencia and Navarra). In men, 16 of the 47 studied provinces displayed an excess melanoma mortality, with PP ≥ 80%, most of which were located on the Cantabrian coast (Gipuzkoa and Bizkaia) or in the vicinity (Araba and Navarra), on the Mediterranean coast (Cádiz, Málaga, Murcia, Alicante, Valencia and Tarragona) or in the vicinity (Zaragoza, Huesca, Teruel, Cuenca and Albacete). Additionally, Madrid showed elevated melanoma mortality.

The estimated smooth spatiotemporal RRs for the period 2011–2020 for all provinces in mainland Spain are shown in Figure 4, displaying distinct geographical patterns of mortality risk. For women, a persistent north-south gradient was observed, with the northern provinces consistently showing a significantly higher excess risk compared with the national average. For men, the pattern followed an east-west gradient, with the easternmost provinces showing significantly increased risks. In addition, the figure reflects changes in mortality trends over time, consistent with the results of the national Joinpoint analysis.

Figure 4.

Geographical representation of the spatiotemporally smoothed relative risks (RRs) and posterior exceedance probabilities (PP) of melanoma mortality in all provinces in Spain, for each year between 2011 and 2020, for women and men, respectively.

Moran’s I was found to be 0.58 for women and 0.57 for men, and was statistically significant in both cases (P < 0.05). When using the LISA method (Figure 5), spatial groups were clustered in the northern and eastern zones of the country for men (all P < 0.05), while the spatial groups were distributed to a greater extent in the western and northern/western provinces for women, showing significant groups with P < 0.05.

Figure 5.

Melanoma mortality hot spots/clusters at the provincial level in Spain (2011–2020), calculated by the Local Indicator of Spatial Association (LISA) method, showing: significant hot spots (high values next to high [HH]), cold spots (low values next to low [LL]), and spatial outliers (high among low [HL] or vice versa [LH]) of mortality rates.

The number of deaths that occurred in the peninsular provinces in Spain, distributed by sex in each year of study, are shown in Supplementary Table S1.

Discussion

The results of the present study showed a stabilization of melanoma mortality risk in men and women, which concurs with a previously published study that reported standardized rates in men showing a significant increase during the period 1992–2014, followed by a stabilization until 2018. In this same study, the melanoma mortality rates in women remained stable during the period 1994–2018.¹¹ Despite improvements in diagnosis, early detection and new treatments, the trend in mortality rates has not reversed as observed in other studies,^7–9 even for advanced disease. Therefore, further studies are needed to try to identify the determinants of this situation in Spain.

In terms of the present geographical analysis, some areas presented a higher CMM mortality than that expected for the entire country. Although with certain nuances, the geographical patterns observed in men and women were similar to those observed in the ‘Atlas of mortality from cancer and other causes in Spain, 1978–1992’.¹⁷ Herein, it was found that, whilst no statistically significant geographical aggregations were detected, the provinces with the highest rates were concentrated in the eastern and northern coastal areas in men and women, respectively. Furthermore, provincial differences and the geographical pattern were considerably more marked among men.

The ‘Atlas de mortalidad en Castilla-La Mancha 2003–2014’ showed that the spatial patterns of mortality for men and women were different, similar to that observed in the present study, with excess mortality in the province of Albacete in men but not in women.²⁸

The present geographical patterns may reflect an increased vulnerability to UV radiation due to unhealthy recreational sun exposure patterns, population ageing and/or ozone depletion.²⁹ Some risk factors for the development of CMM are not modifiable by the individual (e.g., Fitzpatrick phototype I-III skin), but others, such as lifestyle factors, could have an influence. As prevention measures in this regard, it is essential to avoid periods of intense but intermittent sun exposure, such as during holidays in sunny locations, and to limit prolonged exposure, as experienced when working outdoors.³⁰

Notably, previous reports in Canada and the USA,^31,32 comparing coastal and inland areas, documented a greater incidence of CMM cases along the coast, after adjusting for socioeconomic status, UV index and latitude. The present graphical results highlight these findings visually, as well as providing specific details on the provinces at risk. Inhabitants of these provinces are likely to enjoy the beach or outdoors more frequently than in other provinces, exposing them to a higher risk of developing CMM. In addition, water and sand have a higher surface reflectance than grass, possibly further increasing the risk of CMM.

Data on increased risk of CMM and latitude are contradictory. In Europe, there is evidence of a relationship between latitude and CMM risk,³³ and the present results in women may corroborate this, whilst in men this relationship is weaker. The different patterns by sex may be attributed to differential exposure to risk factors by sex, mediated by lifestyles, and could be due to gender differences in exposure to potential risk factors, partly attributable to gender differences in lifestyles and behavioural habits among men and women.³⁴

Several studies have observed a relationship between socioeconomic status (educational level, income, and employment status) and the incidence, morbidity, and mortality of CMM. High socioeconomic status was associated with higher incidence and low socioeconomic status was associated with delayed diagnosis and poor prognosis in cases of CMM.³⁵ In general, in countries with universal health systems (as is the case in Spain), there is a relationship between a high socioeconomic level and a higher incidence of CMM, while a low socioeconomic level is correlated with advanced stages of the disease.³⁶ In addition, areas with a larger proportion of rural residents are significantly associated with higher incidence and higher mortality rates of CMM.^37,38

Unlike incidence rates, which depend primarily on exposure to risk factors, mortality rates may also reflect differences in medical care, as well as the effect of prevention campaigns. Beyond discrepancies in prevention, early detection, and registration, the significant differences in CMM outcomes in Spain may be rooted in disparities throughout the continuum of CMM care. Total health expenditure per capita correlates closely with estimated CMM survival in Europe,³⁹ and this parameter remains highly heterogeneous in Spain.⁴⁰

Improvements in access to health care and early detection of CMM, with the subsequent improvement in survival, may explain part of the geographical differences observed in Spain. However, despite the characteristics of the Spanish National Health System, with universal coverage, there are suspicions that there may be differences in access to healthcare and diagnosis between regions.^41,42

In the USA, immunotherapy contributes to improving survival times in metastatic CMM, but geographic access to clinical trials of such therapies is difficult for a significant part of the population, which may also be happening in Spain.⁴³ Although new treatments have played a significant role, other factors, such as disparities in access to health care, regional differences in health resources, and the effectiveness of public health campaigns, might also influence the observed outcomes. These factors may affect both early detection and management of advanced melanoma, contributing to the geographic variations identified in the present study.

When interpreting the current results, several factors should be considered. Mortality is not the best indicator to study CMM distribution; however, it remains the only comprehensive source of information on CMM in Spain. Geographical differences might have resulted in variations in the quality of certification of causes of death in Spain.⁴⁴ Although two ICDs were used during the present study period (9th and 10th ICD revisions), it is unlikely to have impacted the mortality rates for CMM, as this category is similar in the two editions used.

Ecological studies have several shortcomings. The statistical techniques employed in the present study allowed the spatial pattern of CMM mortality to be described, but cannot explain the differences between areas. Since the data are aggregated, the level of exposure to any risk factor of the deceased and non-deceased is unknown. Therefore, a main limitation of the study was that the importance of factors including genetics, sunburns, and tanning bed use in CMM aetiology could not be assessed. Moreover, since deaths in Spain are registered according to the last known address without specifying how long the person had lived there, it is impossible to know whether individuals currently residing in a province have lived there most of their lives and have been exposed to the associated environmental risk factors. Considering that, any hypothesis that suggests an association between the excess mortality observed in any province and social inequalities, the use of healthcare services, or the individual’s environmental exposures could fall into the well-known ecological fallacy.⁴⁵ Further research combining ecological data with individual-level information is needed to understand the complete picture of CMM mortality in Spain.

The proposed models control for variability of the SMR, which shows instability when calculated for provinces with low population or low number of deaths per CMM. Mapping the SMR using the Besag, York and Mollié method tends to eliminate some of this random variability. One of the limitations of the method is that in areas where risk is concentrated, the excess could be due to high values in neighbouring provinces and vice versa. This could lead to false positives (increasing the risk in some provinces) or false negatives (weakening the risk in others). Another limitation is related with the possible presence of missing or incomplete data that may have influenced the results of the present study. However, official databases from the National Institute of Statistics were used, widely recognized for their quality and coverage. Although the data were considered complete for the variables analysed, small inconsistencies in the certification of causes of death or variations in provincial records cannot be ruled out, and may have slightly impacted the estimates. This aspect underlines the need for cautious interpretations and future independent validations. Despite these limitations, this type of study serves to identify areas of increased risk and generate hypotheses based on current knowledge regarding the epidemiology of CMM and its risk factors.

The maps obtained in the present study provide a basis for further hypotheses, such as whether the distribution of provincial demographic and socioeconomic characteristics influences the observed geographical patterns. The region of residence and educational level have been noted to influence the stage of CMM and therefore excess mortality. These observations suggest that both geographical and socio-economic data should be considered in the prevention of CMM.⁴⁶

Increasing the number of trained dermatologists and the future widespread implementation of technologies, such as teledermatology or mobile skin self-exam applications, with the help of artificial intelligence for diagnosis, might contribute to filling the gaps in CMM care.⁴⁷

The present results strengthen the relevance of this type of analysis in dealing with public health problems, by identifying areas in which to target programmes or strategies to reduce or eliminate the impact of the disease analysed. Future research is needed to characterise specific risk factors within each province, considering the context of the universal health system to implement specific strategies to reduce CMM incidence, morbidity, and mortality in Spain.

Supplemental Material

sj-pdf-1-imr-10.1177_03000605251319609 - Supplemental material for Spatial clusters and temporal trends of cutaneous malignant melanoma mortality in Spain

Supplemental material, sj-pdf-1-imr-10.1177_03000605251319609 for Spatial clusters and temporal trends of cutaneous malignant melanoma mortality in Spain by Lucía Cayuela, José-Juan Pereyra-Rodriguez, Juan-Carlos Hernández-Rodríguez and Aurelio Cayuela in Journal of International Medical Research

Footnotes

Author contributions

All authors contributed to the conception and design of the work; the acquisition, analysis, and interpretation of data; drafting the work and revising it critically for important intellectual content; approved the version to be published; and were responsible for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work were properly investigated and resolved.

Data availability

The data used in this study are freely provided by the National Institute of Statistics ().

Declaration of conflicting interest

The Authors declare that there are no conflicts of interest.

Funding

This research received no specific grant from any funding agency in the public, commercial or nonprofit sectors.

Supplemental material

Supplemental material for this study is available online.

ORCID iD

José-Juan Pereyra-Rodriguez

References

Arnold

Singh

Laversanne

, et al. Global burden of cutaneous melanoma in 2020 and projections to 2040. JAMA Dermatol 2022; 158: 495–503.

Briatico

Mancuso

Argenziano

, et al. Trends in cutaneous melanoma mortality in Italy from 1982 to 2016. Int J Dermatol 2022; 61: 1237–1244.

Dulskas

Cerkauskaite

Vincerževskiene

, et al. Trends in incidence and mortality of skin melanoma in Lithuania 1991–2015. Int J Environ Res Public Health 2021; 18: 4165.

European Commission. Skin melanoma burden in EU-27, https://ecis.jrc.ec.europa.eu/sites/default/files/2023-12/Melanoma_cancer_2021_en.pdf. (2021, accessed 2 November 2022).

Yang

Salciccioli

Marshall

, et al. Trends in malignant melanoma mortality in 31 countries from 1985 to 2015. Br J Dermatol 2020; 183: 1056–1064.

Berk-Krauss

Stein

Weber

, et al. New systematic therapies and trends in cutaneous melanoma deaths among US Whites, 1986–2016. Am J Public Health 2020; 110: 731–733.

Conte

Ghazawi

, et al. Population-based study detailing cutaneous melanoma incidence and mortality trends in Canada. Front Med (Lausanne) 2022; 9: 830254.

Hübner

Terheyden

, et al. Turning point in skin cancer mortality: an analysis of the trend in Germany from 1998 to 2017. Hautarzt 2019; 70: 989–992.

Kahlon

Doddi

Yousif

, et al. Melanoma treatments and mortality rate trends in the US, 1975 to 2019. JAMA Netw Open 2022; 5: e2245269.

10.

Cayuela

Rodríguez-Domínguez

Lapetra-Peralta

, et al. Has mortality from malignant melanoma stopped rising in Spain? Analysis of trends between 1975 and 2001. Br J Dermatol 2005; 152: 997–1000.

11.

Cayuela

Rodríguez-Domínguez

, et al. Is melanoma mortality declining in Spain? Analysis of trends from 1975 to 2016. Br J Dermatol 2018; 179: 991–992.

12.

Durán-Romero

Sendín-Martin

Conejo-Mir

, et al. Cutaneous malignant melanoma mortality in Spain from 1979 to 2018. Trends and new perspectives in the immunotherapy era. J Eur Acad Dermatol Venereol 2021; 35: 884–891.

13.

Ortiz-Álvarez

Durán-Romero

Hernández-Rodríguez

, et al. Cutaneous malignant melanoma mortality in Andalusia from 1979 to 2018. Toward new perspectives? Dermatol Ther 2021; 34: e14715.

14.

García-Souto

Durán-Romero

Pereyra-Rodríguez

JJ.

Melanoma mortality in Spain: predictions up to 2043. Int J Dermatol 2021; 60: 844–850.

15.

López-Abente Ortega

Pollán Santamaría

Escolar Pujolar

, et al. Atlas of cancer mortality and other causes of death in Spain 1978–1992. 2nd ed. Madrid: Instituto de Salud Carlos III, 2001, https://repisalud.isciii.es/bitstream/handle/20.500.12105/4953/Atlasdemortalidadpor_2001.pdf?sequence=1&isAllowed=y (accessed 10 January 2023).

16.

Cayuela

Pilo Uceda

Asuero Llanes

, et al. Spatial pattern of COPD mortality in the municipalities of Seville 2013–2017. Rev Esp Patol Torac 2020; 32: 188–194.

17.

Aragonés

Goicoa

Pollán

, et al. Spatio-temporal trends in gastric cancer mortality in Spain: 1975–2008. Cancer Epidemiol 2013; 37: 360–369.

18.

National Institute of Statistics. https://www.ine.es/index.htm (last accessed 10 January 2023).

19.

Eurostat. Revision of the European standard population. Report of Eurostat’s task force, https://ec.europa.eu/eurostat/en/web/products-manuals-and-guidelines/-/ks-ra-13-028 (2013, last accessed 10 January 2023).

20.

National Cancer Institute. Joinpoint Regression Program, version 4.9.1.0 – April 2022, https://surveillance.cancer.gov/joinpoint/download (accessed 1 November 2022].

21.

Besag

York

Mollié

Bayesian image restoration, with two applications in spatial statistics. Ann Inst Stat Math 1991; 43: 1–20.

22.

Knorr-Held

Bayesian modelling of inseparable space-time variation in disease risk. Statist Med 2000; 19: 2555–2567.

23.

Spiegelhalter

Best

Carlin

, et al. Bayesian measures of model complexity and fit. J R Stat Soc Series B Stat Methodol 2002; 64: 583–639.

24.

Rue

Martino

Chopin

Approximate Bayesian inference for latent Gaussian models using integrated nested Laplace approximations. J R Stat Soc Series B Stat Methodol 2009; 71: 319–392.

25.

R Development Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing, http://www.R-project.org/.

26.

Richardson

Thomson

Best

, et al. Interpreting posterior relative risk estimates in disease-mapping studies. Environ Health Perspect 2004; 112: 1016–1025.

27.

GeoDa 1.20 for 64-bit Windows, http://geodacenter.github.io/download.html (last accessed 10 January 2023).

28.

Gómez Rubio

Palmí Perales

Atlas de mortalidad en Castilla-La Mancha 2003–2014, http://atlasmortalidad.uclm.es/ (Last accessed 10 January 2023).

29.

Parker

ER.

The influence of climate change on skin cancer incidence – A review of the evidence. Int J Womens Dermatol 2020; 7: 17–27.

30.

Rodrigues

Sniehotta

Birch-Machin

, et al. Aware, motivated and striving for a ‘safe tan’: an exploratory mixed-method study of sun-protection during holidays. Health Psychol Behav Med 2017; 5: 276–298.

31.

Ghazawi

Cyr

Darwich

, et al. Cutaneous malignant melanoma incidence and mortality trends in Canada: a comprehensive population-based study. J Am Acad Dermatol 2019; 80: 448–459.

32.

Korgavkar

Lee

Weinstock

MA.

Higher melanoma incidence in coastal versus inland counties in California. Melanoma Res 2014; 24: 280–285.

33.

Chang

Barrett

Bishop

, et al. Sun exposure and melanoma risk at different latitudes: a pooled analysis of 5700 cases and 7216 controls. Int J Epidemiol 2009; 38: 814–830.

34.

D’Ecclesiis

Caini

Martinoli

, et al. Gender-dependent specificities in cutaneous melanoma predisposition, risk factors, somatic mutations, prognostic and predictive factors: a systematic review. Int J Environ Res Public Health 2021; 18: 7945.

35.

Jiang

Rambhatla

Eide

MJ.

Socioeconomic and lifestyle factors and melanoma: a systematic review. Br J Dermatol 2015; 172: 885–915.

36.

Bailey

Grose

, et al. Socioeconomic status and melanoma in Canada: a systematic review. J Cutan Med Surg 2021; 25: 87–94.

37.

Shellenberger

Johnson

Fayyaz

, et al. Disparities in melanoma incidence and mortality in rural versus urban Michigan. Cancer Rep (Hoboken) 2023; 6: e1713.

38.

Schumacher

Yang

, et al. Disparities in cancer incidence by rurality in California. J Natl Cancer Inst 2023; 115: 385–393.

39.

Forsea

Del Marmol

Stratigos

, et al. Melanoma prognosis in Europe: far from equal. Br J Dermatol 2014; 171: 179–182.

40.

Clemente

Lázaro-Alquézar

Montañés

Convergence in Spanish public health expenditure: has the decentralization process generated disparities?

Health Policy 2019; 123: 503–507.

41.

Guevara

Rodríguez-Barranco

Puigdemont

, et al. Disparities in the management of cutaneous malignant melanoma. A population-based high-resolution study. Eur J Cancer Care (Engl) 2019; 28: e13043.

42.

Salvador

Aparicio

Baron

, et al. Equity, barriers and cancer disparities: study of the Spanish Society of Medical Oncology on the access to oncologic drugs in the Spanish regions. Clin Transl Oncol 2017; 19: 341–356.

43.

Mulligan

Zheng

, et al. Geographic disparities in access to immunotherapy clinical trials for metastatic melanoma. Arch Dermatol Res 2023; 315: 1033–1036.

44.

Pérez-Gómez

Aragonés

Pollán

, et al. Accuracy of cancer death certificates in Spain: a summary of available information. Gac Sanit 2006; 20: 42–51.

45.

Sedgwick

Understanding the ecological fallacy. BMJ 2015; 351: h4773.

46.

Strömberg

Peterson

Holmberg

, et al. Cutaneous malignant melanoma show geographic and socioeconomic disparities in stage at diagnosis and excess mortality. Acta Oncol 2016; 55: 993–1000.

47.

Porter

Kimball

AB.

Predictions, surprises, and the future of the dermatology workforce. JAMA Dermatol 2018; 154: 1253–1255.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.07 MB