Study variables
The study questionnaire was developed by a multidisciplinary team including a dermatologist, a psychologist and two epidemiologists. It consisted of validated scales measuring subjective well-being and depression and additional questions on happiness in general and the evaluated impact of psoriasis. The survey was pre-tested by three psoriasis patients and, based on the resulting feedback, slightly modified. For the scales used in this study, internal consistency was considered “excellent” for Cronbach’s alpha values > 0.9 and “good” for Cronbach’s alpha values > 0.8 [
36].
Subjective well-being
Following Diener et al. [
25], subjective well-being was operationalized as
PA, NA and
SWL. PA and NA were measured using the validated German version of the Scale of Positive and Negative Experience (SPANE) [
37,
38]. We chose the SPANE over the more frequently used Positive and Negative Affect Schedule (PANAS) [
39], as it assesses a wider range of positive and negative emotions than the PANAS, which focusses on high arousal emotions, and as it is coherent with Diener and colleagues conceptualization of subjective well-being [
37]. The SPANE consists of two subscales measuring PA and NA, respectively. Each subscale consists of six adjectives, e.g., “pleasant” and “positive” for PA and “unpleasant” and “negative” for NA). The respondents were asked to indicate how often they had felt the respective feelings over the past two weeks on a 5-point scale from 1 (“very rarely or never”) to 5 (“very often or always”). The two distinct subscales, showed excellent and good reliability in this study with Cronbach’s alphas of 0.93 and 0.86 for PA and NA, respectively. For each subscale, the items were averaged to form an index.
SWL was measured using the German version of the Satisfaction With Life Scale (SWLS) [
40,
41]. The scale consists of five items (e.g., “in most ways my life is close to my ideal”), each rated on a 7-point scale from 1 (“strongly disagree”) to 7 (“strongly agree”). The scale showed excellent reliability in this study with a Cronbach’s alpha of 0.91 and the items were averaged to form an index. As some researchers argue that happiness is not in fact a multidimensional but a unidimensional construct [
29], we decided to also include a
heuristic measure of happiness, meaning a single question asking participants for their overall happiness [
20]. In order to achieve comparability with the general population, a single question from the European Social Survey was used: “Taking all things together, how happy would you say you are?” [
42]. Respondents could give their answer on a 11-point scale from 0 (“extremely unhappy”) to 10 (“extremely happy”).
As an additional question on happiness in the context of psoriasis, the participants were asked for a subjective evaluation of the impact of psoriasis on their own happiness (“Do you think that your psoriasis has a negative impact on how happy you are?”—“No”, “Yes, a little”, “Yes, moderately”, “Yes, very much”).
Depression
The WHO-5 Well-Being Index [WHO-5,
43] is a validated screening questionnaire for
depression. It consists of five statements (e.g., “My daily life has been filled with things that interest me”), which respondents rate on a 6-point scale from 0 (“at no time”) to 5 (“all of the time”). The items showed good reliability in this study with a Cronbach’s alpha of 0.87. Following the instructions, the values of the items were added, resulting in an overall score ranging from 0 to 25, with lower scores indicating a higher risk for depression. Using a cut-off of ≤ 7, the WHO-5 has shown good properties for the screening of major depression with a sensitivity of 94% and specificity of 78% [
44]. Consequently, a score of ≤ 7 was considered a positive screening result for depression in this study.
Participants’ characteristics
As possible parameters associated with happiness, data on age (in years), gender, years since first diagnosis, and current treatment status (currently receiving treatment vs. not receiving treatment) were collected. Subjective general disease severity was measured as “mild”, “moderate”, “severe”. In addition, data on current disease severity were collected using the same given options (“mild”, “moderate”, “severe”). Based on general and current disease severity, new dichotomous variables indicating current phases of relative improvement or relative deterioration compared to general disease severity were derived.
Analysis
Prior to the analysis, data were checked for completeness (at least 80% of questions on happiness answered) and plausibility (e.g., data were considered implausible if the participants indicated a higher number of years since first diagnosis than age). As only very few participants did not meet our criteria for completeness and plausibility (
n = 8), these cases were excluded from all further analysis rather than using imputation in order to keep the analysis as simple as possible. The remaining data were analyzed descriptively. Pearson’s correlations were calculated in order to explore associations between the examined variables. Correlations were considered moderate for
r > 0.5 and strong for
r > 0.7 [
45]. To further differentiate the four constructs PA, NA, SWL, and depression, exploratory factor analysis using Promin Rotation, an oblique rotation method appropriate for correlated variables, was conducted after checking the respective requirements were fulfilled. As all variables were measured on at least 5-step scales (with equal distances between the answer options and numbers suggesting equal steps between the categories), variables were treated as quasi-metric and Pearson’s correlations were used for factor analysis. Following Guadagnoli and Velicer [
46], factor loadings of 0.4 and higher were considered stable, which is why all smaller coefficients were suppressed. Factor retention was determined using parallel analysis [
47]. However, as the retention of three factors, which was the number of factors suggested by parallel analysis, did not result in stable factor loadings for all examined items, we additionally conducted the analysis retaining one more factor, which was in line with the theoretical assumption of four differing constructs of PA, NA, SWL, and depression.
Means of PA, NA, heuristic happiness, SWL and WHO-5 were compared to norm data [
41,
48,
49] or, in case of PA and NA, to data of a validation study [
38] as norm data were not available. For both SWL and WHO-5, norm data had been collected in representative samples of the German general population (with the assistance of a demographic consulting company (SWL: mean age 48.9 ± 18.3 years, 52.2% women; WHO-5: mean age 48.3 years, SD not indicated for the overall sample, 52.7% women) [
41,
48]. For heuristic happiness, norm data for the German general population were collected within the European Social Survey (mean age 48.2 ± 18.1 years, 49.9% women) [
49]. For PA and NA, the data of the German validation study among a total of 498 participants recruited in university lectures (
n = 264) and via mailing lists/social media (
n = 234) were used as reference data (mean age not indicated, 18.5% < 20 years, 47.8% 20–29 years, 16.3% 30–39 years, 7.7% 40–49 years, 5.3% 50–59 years, 4.5% > 59 years, 75.1% women) [
38]. As for PA, NA, and heuristic happiness the datasets of the reference samples were accessible, the group comparisons for these variables were conducted using ANCOVA and planned contrasts while controlling for sex and age. Adjusted means (m
a) are reported. For SWL and WHO-5, the datasets of the reference populations could not be retrieved, which is why analyses were conducted stratified for age and sex, using Student’s
t-tests. The age groups for these analyses were chosen to match the age groups reported in the respective reference samples. As Student’s
t-tests and ANCOVA have been shown to provide robust results even when normality and equal variance assumptions are violated [
50‐
52], the analyses were conducted without prior verification of these assumptions.
Parameters associated with subjective well-being and a positive screening for depression were identified using multiple linear regression models and binary logistic regression, respectively. In all regression models, age, gender, years since first diagnosis, general subjective disease severity, improvement or deterioration of skin condition and current treatment status were entered as independent variables. As a result, adjusted raw (B) and standardized regression coefficients (
β) and Odds Ratios (OR) as well as corresponding 95%-confidence intervals (CIs) and the percentage of variance explained by each model (adjusted R
2) are reported. As depression is characterized by the inability to feel happy [
35], we conducted additional sub-group analyses in participants who did not receive a positive screening result for depression in order to explore differential findings for affective well-being which are not explained by the presence of depression.
The level of significance was set at
α = 0.05 for all analyses. All statistical analyses except exploratory factor analysis were conducted using IBM SPSS Statistics Version 24 (IBM Corporation, Armonk, NY, USA). Exploratory factor analysis was conducted using FACTOR software [
53].