Impact of Melanoma on Patients' Lives Among 562 Survivors: A Dutch Population-Based Study

Objective  To assess the impact of melanoma on the health-related quality of life of patients from the general population up to 10 years after diagnosis and its determinants.

Design  A cross-sectional Dutch population-based postal survey among patients with melanoma for the years 1998 to 2008 using the Eindhoven Cancer Registry.

Main Outcome Measures  The 36-Item Short-Form Health Survey (SF-36), Impact of Cancer (IOC) questionnaire and specific melanoma-related questions. The SF-36 scores of the cases were compared with normative data. Multiple linear regression models were used to identify associated factors of SF-36 and IOC scores.

Results  The response rate was 80%. The mean age of the 562 respondents was 57.3 years; 62% were female, and 76% had a melanoma with a Breslow thickness of less than 2 mm. The SF-36 component scores of patients with melanoma were similar to those of the normative population. In a multiple linear regression model, stage at diagnosis, female sex, age, and comorbidity were significantly associated (P < .05) with the physical and mental component scores. Women were significantly more likely to report higher levels of both positive and negative IOC. Time since diagnosis, tumor stage, and comorbidity were significant predictors of negative IOC scores. Women seemed to adjust their sun behavior more often (54% vs 67%; P < .001) than men and were more worried about the deleterious effects of UV radiation (45% vs 66%; P < .001).

Conclusion  The impact of melanoma seems to be specific and more substantial in women, suggesting that they may need additional care to cope with their melanoma optimally.

Although the prognosis is relatively good for about 80% of patients with melanoma, they remain at risk for disease progression and have an increased risk of developing subsequent melanomas.^1-6 A systematic review on the impact of melanoma on patients' lives that included a variety of psychometric measures suggested that a third of patients with melanoma have reported clinically significant levels of distress.⁷ Therefore, melanoma can be considered a chronic life-threatening disease that may affect patients' lives considerably. Because most patients are aware that past sun exposure might have played a role in the development of their melanoma,⁸ they are likely to change their lifestyle to minimize UV exposure, which consequently may affect patients' health-related quality of life (HRQoL) as well.

The HRQoL impact of melanoma is not very well documented.⁷ Of the 13 eligible studies included in a systematic review, most included a relatively small number of patients who were almost always treated in specialized centers and assessed impact within 2 years after diagnosis. They used a variety of questionnaires assessing patient reported outcomes. Furthermore, a comparison of HRQoL impairment was not made between men and women and between patients with melanoma and the general population and/or other cancer groups.⁷ Although the Functional Assessment of Cancer Therapy–Melanoma (FACT-Melanoma) and the Skin Cancer Index have been introduced, neither of them is well suited to measure melanoma-specific HRQoL among the general population of melanoma survivors.^7,9-11 The FACT-Melanoma was developed for clinical trial purposes that include high-risk patients who have lower survival rates than most patients with melanoma in the general population and who receive additional surgical and/or systemic therapy. The Skin Cancer Index was designed for keratinocytic cancers and not for melanoma.

The objective of this cross-sectional survey was to investigate the impact of melanoma on patients' HRQoL up to 10 years after diagnosis and the association between HRQoL impairment and patient and tumor characteristics. We hypothesized that melanoma survivors would report an impaired HRQoL compared with that of the normative Dutch population and that this impact would diminish over time and differ across sex. A generic and cancer-specific instrument and additional melanoma-specific items were used to evaluate HRQoL and associated factors in a large cohort of (long-term) melanoma survivors from the general Dutch population.

A cross-sectional survey was conducted with the assistance of the Eindhoven Cancer Registry (ECR), Eindhoven, the Netherlands. The ECR records data of patients newly diagnosed as having cancer in the southeast part of the Netherlands,¹² which has a population of about 2.3 million inhabitants and is served by 18 hospital locations and 2 large radiotherapy institutions. Of the 18 hospitals, 3 large regional hospitals were asked to participate. From the ECR database, all patients diagnosed as having melanoma (International Classification of Diseases, Oncology, codes C44.0-C44.9 with morphology codes 8720-8790) from January 1, 1998, to August 1, 2007, in 1 of the 3 hospitals were identified. The time since diagnosis ranged from 6 months to a maximum of 10 years, owing to the fact that there is a 6-month delay before tumors are completely registered in the ECR.

Patients older than 85 years at the time of the survey were excluded because it was expected that they would have difficulty completing a self-administered questionnaire. Deceased patients were excluded by linking the database to the Central Bureau for Genealogy, which collects data on all deceased Dutch citizens or permanent residents via civil municipal registries.

The treating physicians of the 3 hospitals informed their eligible patients by sending a standardized invitation letter and the questionnaire. Patients were reassured that nonparticipation would not have consequences for their follow-up or treatment. By returning a completed questionnaire, patients consented to linkage of the HRQoL data with those of their disease and treatment history as registered in the ECR. Questionnaires were coded for anonymous data collection tracking and linkage with the ECR database. Data were collected from February through May 2008. The medical ethics committee of Catharina Hospital in Eindhoven gave their approval to this study.

The ECR provided data on melanoma and demographic characteristics of the patients. The questionnaire included a generic and cancer-specific HRQoL instrument (36-Item Short-Form Health Survey [SF-36] and Impact of Cancer [IOC], respectively) and additional questions regarding demographic variables, disease progression, and current comorbidity (adapted from the Charlson comorbidity index).^13-15 For the SF-36, standard scoring procedures of the 8 scales) (ie, Physical Functioning, Role Physical, Bodily Pain, General Health, Vitality, Social Functioning, Role Emotional subscale, Mental Health) were followed where higher scores indicate better functioning (range, 0-100).¹³ Two higher-order component scores for physical (PCS) and mental health (MCS) were also calculated. The IOC was developed by Zebrack et al¹⁵ in 2006 to measure the well-being of long-term cancer survivors and their adjustment to changes. It consists of 41 questions (5-point response scale), 6 domains (ie, physical, psychological, social, existential, meaning of cancer, health worry) with 10 dimensional subscales.^14,15 These 10 subscales are Physical: Health Awareness (4 items); Physical: Body Changes (5 items); Psychological: Positive Self-Evaluation (8 items); Psychological: Negative Self-Evaluation (4 items); Existential: Positive Outlook (3 items); Existential: Negative Outlook (4 items); Social: Life Interferences (3 items); Social: Value of Relationships (2 items); Meaning of Cancer (5 items); and Health Worry (3 items). These subscales were combined to form a higher-order positive or higher-order negative scale (range, 0-5) with higher scores indicating a more positive or negative impact of cancer, respectively (Table 1). In addition to these preexisting validated instruments, several melanoma-specific items that assess issues related to treatment, impact on daily life, and follow-up were included (based on expert opinion) because these issues were not covered by the SF-36 or IOC questionnaires (see the subsections titled “Melanoma-Specific Items” and Practical Issues Related to Melanoma” in the “Results” section). Responses were categorized as “less,” “the same,” “a little bit more,” “more,” or “a lot more.” In the analyses, “more” and “a lot more” were combined. To evaluate the physical symptoms experienced due to melanoma or its treatment, questions regarding pain, itch, swelling, and numbness with a 3-point scale were included.⁹ Several items pertained to the extent to which the diagnosis of melanoma affected patients' sun (exposure) behavior (ie, activity in the sun, worry about the effects of the sun on the skin, and sun-protective measures) were included. Furthermore, patients were asked whether applications for health, life, and disability insurance or home mortgage had been hampered because of their melanoma.

To test for statistical differences, the t test and χ² test for continuous and categorical variables, respectively, were used. The scores of the SF-36 scales were compared with an age- and sex-matched random normative sample of 1742 Dutch adults from the general population by means of analysis of variance.¹³ The age- and sex-matched sample was drawn from a nationwide sample of the Dutch adults to whom the SF-36 was mailed (2800 households drawn at random from the national telephone registry). Nonresponders were sent reminder letters after 2 months and 3 months following initial mailing. In total, 1771 were returned, representing a 63% response rate, of whom 56% was male.¹³ We used Norman's “rule of thumb,”¹⁶ which says that the threshold of discrimination for changes in HRQoL for chronic diseases seems to be approximately half a standard deviation.

Multivariate linear regression analyses were performed to estimate the association between SF-36 and IOC scores and independent variables (expressed as standardized β-coefficients and 95% confidence intervals). Variables were included in the multivariate model if they were significantly associated with the outcome in the univariate analysis and/or were considered a priori to be of clinical relevance. In these models, age and time since diagnosis were entered as continuous variables, tumor stage ranged from I to IV, and presence of comorbidity was binary (yes/no). Four levels of education were used, and marital status was categorized as having a partner vs no partner.

A 2-sided P <.05 was considered statistically significant. All statistical analyses were performed using SAS statistical software (version 9.1 for Windows; SAS Institute Inc, Cary, North Carolina).

Of the 992 patients diagnosed as having cutaneous melanoma in 1 of 3 hospitals between January 1998 and August 2007, 802 patients were alive on August 1, 2007 (Figure 1). Of these patients, 103 could not be contacted (Figure 1). The remaining 699 survivors received a questionnaire, and 562 responded (a 80.4% response rate). The distribution of the demographic and melanoma characteristics of the respondents, nonrespondents, and patients with unverifiable addresses were similar with respect to sex, Breslow thickness, stage at diagnosis, and initial treatment (Table 2). Nonrespondents and patients with unverifiable addresses were generally younger (P < .05). On average, nonresponders also had longer survival times.

About 70% of patients had been diagnosed as having stage I melanoma, and 49% had a lesion with a Breslow thickness of less than 1 mm (Table 2). Almost all patients underwent local surgical excision, and 20% underwent a sentinel node procedure (SNP). Chemotherapy or radiotherapy was administered in less than 1% of patients. No statistically significant differences were found for Breslow thickness (P = .64) and stage distribution (P = .71) between responders, nonresponders, and patients with unverifiable addresses.

The mean ages at diagnosis and at time of survey of the respondents were 52.6 years and 57.2 years, respectively. Sixty-two percent of respondents were female. Most respondents had at least a high school education, were married, had children, and were employed at time of survey (Table 3). Of all respondents, 35% reported at least 1 other medical condition: hypertension, joint complaints, and/or a history of a malignant disease.

Compared with an age- and sex-matched sample from the general Dutch population, melanoma survivors did not report impaired HRQoL as measured by the SF-36 (Figure 2). Interestingly, patients with melanoma scored statistically significantly higher on PCS and several subscales (Physical Functioning, Role Limitations: Emotional Problems, Mental Health, Bodily Pain, and General Health; data not shown) than the general population, but the mean score differences were less than half a standard deviation suggesting no clinical significant differences, according to Norman's rule of thumb (Figure 2).¹⁶

After adjusting for age at time of the survey, years since diagnosis, disease stage, SNP, comorbidity, marital status, and educational level in a multivariate analysis, female sex, older age, tumor stage, and comorbidity were significantly negatively associated with PCS, whereas an SNP was positively associated (Table 4). Female sex and comorbidity were also negatively associated with MCS, whereas having a partner was positively associated with MCS. Women scored significantly lower (indicating higher HRQoL impairment) than men: 1.8 points (P < .05) on the PCS and 2.3 points (P < .05) on the MCS; however, this was not clinically relevant. In addition, female melanoma survivors reported higher impairment on 5 of 8 subscales (Physical Functioning, Bodily Pain, General Health, Vitality, and Mental Health; data not shown).

To estimate the impact of melanoma on patients' psychological well-being in more detail, the IOC was included¹⁴ (Table 1). The mean scores of patients with melanoma for each subscale of the IOC differed from 2.2 to 3.1. The lowest scores were seen on the subscale Social: Life Interferences, which focuses on cancer- or treatment-related symptoms of the cancer that interferes with a patients' socializing, traveling, or time with family. The highest score of 3.1 was seen on the Existential: Positive Outlook subscale, which covers increased wisdom and spirituality due to the cancer experience.

In multivariate analyses, female sex was statistically significantly associated with higher-order positive and negative scales, suggesting that women experience both a more negative and more positive impact of melanoma on their psychological well-being than men (Table 4): compared with men, women scored 0.21 points (P < .05) higher on the higher-order positive scale and 0.24 points (P < .001) on the higher-order negative scale. Analyses showed that the trends of responses were similar, but that women responded more extremely compared with men (data not shown). On the higher-order negative scale, time since diagnosis was significantly associated with a negative β-coefficient, which implies that for every year further from diagnosis the patient may be less negatively influenced by their cancer. In contrast, more comorbidities and advanced melanoma stage (III vs I) were associated with a higher-order negative IOC score (Table 4).

Compared with men, a significant larger proportion of women reported going on vacation to sunny destinations less frequently (67% vs 56%, respectively; P = .03). Women were also more worried about the effects of sunlight on their skin (66% vs 45%; P < .001) and that of their spouses and children (49% vs 32%; P < .05).

Furthermore, female patients reported seeking shade and/or using sunscreen more often than men (67% vs 48%; P < .001) (Figure 3). Most men and women reported using sunscreen only in the summer or when the sun shines (83% vs 90%, respectively), with a sun protection factor of 20 to 30 or more (69% vs 72%, respectively). Women were more likely to use sunscreen more times a day than men (25% vs 64%; P < .001).

Compared with men, women experienced more pain (26% vs 11%; P < .001), itchiness (26% vs 21%; P = .26), and numbness (29% vs 21%; P = .10) of their scars. Swelling of the scar or one of the extremities was reported in approximately 10% of men and women.

Most survivors (94%) stated that their professional situation had not changed after being diagnosed as having melanoma. Five percent of survivors indicated that they had changed jobs, reduced the number of hours worked, had been retrained, or stopped working entirely (including work disability) as a result of their melanoma. Of the 258 patients who attempted to obtain health insurance in the survey period, 5% reported having experienced cancer-related problems in obtaining health insurance. Thirty-five percent (n = 84) reported difficulties obtaining life insurance, 15% (n = 98) reported problems with obtaining a mortgage, and 18% (n = 38) experienced trouble obtaining disability insurance when attempting to obtain one of these.

To our knowledge, this is the first study to investigate SF-36 and IOC scores in a population-based sample of melanoma survivors, showing relatively little impairment of the generic health status and HRQoL. In contrast to our hypothesis, the SF-36 scores were not significantly different compared with the normative SF-36 scores of the general Dutch population. This observation suggests that generic and even cancer-specific HRQoL questionnaires might not be sensitive enough for patients with predominantly low stages of melanoma. More specific, but individual, items showed that several UV-related aspects of patients' lives were affected and that having been diagnosed as having malignant melanoma had practical implications for their daily life (eg, getting a mortgage, health insurance, or life insurance).

Of the studied determinants, female sex was most strongly associated with HRQoL. Also, in the multivariate analyses of SF-36 and IOC scores, older age, comorbidity, and higher tumor stage were significant predictors for lower HRQoL.

We compared the SF-36 scores of patients with melanoma with normative data for the SF-36 obtained from of a nation-wide random sample of Dutch adults (see the “Methods” section), which have been used by several previous studies to compare the SF-36 scores of patients with those of the general population.^17,18 Though not clinically relevant (statistical differences, <0.5 SD),¹⁶ the scores of several SF-36 domains and components were significantly higher among patients with melanoma than those from the general population.^19,20 The finding that patients show far more variability in the response distribution of the individual melanoma-specific items (eg, sun behavior and worry) than in the generic SF-36 may confirm the need to combine generic with disease-specific questionnaires.¹⁹ Although it seems counterintuitive at first and in contrast with our hypothesis, this observation is in accordance with studies investigating other cancers showing that patients who had “survived” a cancer rate their generic HRQoL as similar or even better than those from the general population.¹² Individuals can use various cognitive strategies such as belief, posttraumatic growth, or benefit finding to counteract the negative effect of illness (eg, melanoma) on their well-being.¹¹ Finding benefit and growth could be viewed as part of the so-called response shift. When diagnosed as having a life-threatening disease, patients may change their internal standard of what constitutes health and HRQoL (recalibration), adjust their priorities (reprioritization), and/or redefine what is important to them (reconceptualization) in face of their condition.^11,21

Of the variables studied in multivariate models, female sex was most strongly associated with lower SF-36 scores, higher-order negative and positive IOC scores, and more extreme responses to melanoma-specific items, which is confirmed by previous studies in melanoma and other cancers that show a sex difference in the assessment of HRQoL.^7,11,22 In clinical practice, this observation may imply that women need additional care, including follow-up and possibly counseling to optimally cope with their melanoma.⁸ However, men might be less aware of general measures of sun protection and need education about these measures after treatment. Besides, other studies that used specific tools to measure anxiety and/or depression suggest that around a third of patients with melanoma suffer substantially.⁷ This could be reduced by psychological counseling, which might even lead to better survival.²³ In addition to sex, comorbidity was strongly associated with HRQoL impairment due to melanoma. This association has been observed previously in patients with melanoma,⁷ nonmelanoma skin cancers,²² and solid cancers, suggesting that it remains difficult to differentiate the impact of different diseases. It seems that the impact of skin cancer is lowered by the presence of multiple or severe comorbidities. This observation emphasizes the need to test HRQoL instruments for item bias across the presence of comorbidity because an optimal HRQoL tool should not be influenced by external factors.¹⁹ The finding that older age at time of survey predicted lower scores on the physical scales is most likely explained by the presence of other disabilities and/or an aging effect of the cohort. In the long term, the effect of SNPs was reported as positive on the PCS, which is in agreement with other studies suggesting that additional therapy eventually increased HRQoL in patients with melanoma.²⁴ The relatively small group of respondents who had an SNP reported a higher PCS score than those who did not, which is surprising, as surgical complications (eg, lymphoedema and extensive scarring) are likely to impair a patient's well-being. This may suggest that patients who choose to undergo additional or even controversial investigations and/or therapies may differ from those who received local surgical excision only. Although it was initially hypothesized, time since diagnosis did not affect SF-36 scores among melanoma survivors, and the effect of additional years after diagnosis was modest for the negative IOC scale. In other diseases, such as diabetes mellitus, rheumatoid arthritis, non-Hodgkin‘s lymphoma, and breast cancer, the effect of time since diagnosis on finding benefit or growth is inconsistent.^11,17 The low adjusted R² scores of the multivariate models that assessed predictors of SF-36 and IOC scores suggest that this analysis suffers from residual confounding (ie, multiple HRQoL predictors, such as anxiety, personality type, and coping mechanisms were not included in the analysis).^7,13,14,25

Although the effect of melanoma on SF-36 and IOC scores seems to be limited, this cancer may have a profound impact on practical issues of patients' lives and thus affects HRQoL in different domains. A small proportion of individuals experienced difficulties in getting health insurance as a result of their melanoma, but up to a third of the patients experienced difficulty getting life insurance, disability insurance, and/or a mortgage. This type of information is not often assessed in patient populations but seems to be highly relevant in patients who have survived cancer.

This is the largest cross-sectional, population-based study using cancer registry data to investigate the impact of melanoma on patients' HRQoL and other aspects of their lives. Trask and Griffith²² investigated HRQoL among a large group of patients with melanoma ascertained from a multidisciplinary melanoma clinic. However, compared with our study, the design of their study²² was very different (eg, selected HRQoL measures, domain-specific tools, including anxiety and depression scales and categorization of patients), making a comparison between the study findings challenging.

A response rate of 80% is very high and in accordance with those of previous studies performed in the southeast part of the Netherlands.^17,18 This high response rate and the fact that respondents and nonrespondents were comparable suggest that selection bias had a minimal effect on the results of this study. Because no specialized melanoma centers were included, the generalizability of our findings is likely to be good. The ECR is a well established cancer registry¹² but does not record recurrence or metastasis from/after melanoma. Consequently, the effect of these events could not be studied separately. No information was available regarding the social economic status (SES) of patients. Higher levels of SES are associated with increased melanoma risk and higher HRQoL.^26-28 As a proxy for SES, educational level was used; three-quarters of the patients had at most a high school degree (Table 2), suggesting a possible overestimation of melanoma on HRQoL.

The SF-36 was used because it is the most commonly used HRQoL instrument and has the advantage of available Dutch norm data, but it is has not been formerly tested in patients with melanoma.¹⁹ To expand the focus of this study, the IOC, which is a relatively new instrument (and has not been validated in patients with melanoma) and several single items were added. At the time of the survey, the FACT-M, which was intended for the use in clinical trials with a focus on physical limitations, had not yet been validated.^9,10 An indirect comparison of IOC scores suggests that patients' physical and psychological well-being is likely to be less affected by melanoma when compared with breast, prostate, colorectal, or prostate cancer. Patients with melanoma scored lower on all of the positive IOC subscales and higher on only 2 of the negative subscales (body changes and negative self- evaluation)¹⁵ (Table 1).

In conclusion, in a population-based sample of patients who have had malignant melanoma, the impact of melanoma seems to be fairly specific and is not driven by (long-term) treatment effects that affect generic or cancer-specific HRQoL. Therefore, there is need for a melanoma-specific HRQoL instrument that gauges the impact of thin and thick melanomas on patients' lives. Female sex and comorbidity were the main predictors of HRQoL impairment in patients with melanoma, but other (psychological) factors are likely to play a role as well and need to be studied in more detail in future studies.

Correspondence: Tamar Nijsten, MD, PhD, Department of Dermatology, Erasmus MC, Burg s’Jacobplein 51, Gk-316, 3000 CA Rotterdam, the Netherlands (t.nijsten@erasmusmc.nl).

Accepted for Publication: June 11, 2010.

Author Contributions: Drs Holterhues, Cornish, van de Poll-Franse, and Nijsten had full access to all of the data and take responsibility for the integrity of the data and accuracy of the data analysis. Study concept and design: Holterhues, Cornish, van de Poll-Franse, and Nijsten. Acquisition of data: Holterhues, Cornish, van de Poll-Franse, Krekels, Koedijk, and Kuijpers. Analysis and interpretation of data: Holterhues, Cornish, van de Poll-Franse, and Nijsten. Drafting of the manuscript: Holterhues, Cornish, van de Poll-Franse, and Nijsten. Critical revision of the manuscript for important intellectual content: Cornish, van de Poll-Franse, Krekels, Koedijk, Kuijpers, Coebergh, and Nijsten. Statistical analysis: Holterhues, Cornish, van de Poll-Franse, and Nijsten. Obtained funding: Nijsten. Administrative, technical, and material support: van de Poll-Franse, Krekels, Koedijk, Kuijpers, Nijsten, and Coebergh. Study supervision: van de Poll-Franse, Nijsten, and Coebergh.

Financial Disclosure: None reported.

Previous Presentation: An earlier version of these data was presented at the Fifth International Dermato-Epidemiology Association Congress; September 7, 2008; Nottingham, England (Abstract: Holterhues et al. Impact of melanoma on patients' lives: cross-sectional population-based study in the South Netherlands. J Invest Derm. 2008;128:2558).

Additional Contributions: The registry team of the Eindhoven Cancer Registry provided melanoma-specific data.