Itch Is More Important than Clinical Disease Severity for Depression in Psoriasis: A Prospective Register Study

Psoriasis is associated with depression, but the reasons for this are not fully understood (Kleyn et al., 2020Kleyn CE, Talbot PS, Mehta NN, Sampogna F, Bundy C, Ashcroft DM, et al. Psoriasis and Mental Health Workshop Report: Exploring the Links between Psychosocial Factors, Psoriasis, Neuroinflammation and Cardiovascular Disease Risk. Acta Derm Venereol 2020;100(1):adv00020.Google Scholar). It is believed that the severity of the skin disease, itch intensity, and systemic inflammation are important factors (Farzanfar et al., 2018Farzanfar D. Dowlati Y. French L.E. Lowes M.A. Alavi A. Inflammation: a contributor to depressive comorbidity in inflammatory skin disease.Skin pharmacology and physiology. 2018; 31: 246-251Crossref PubMed Scopus (42) Google Scholar, Reich et al., 2016aReich A, Mędrek K, Szepietowski JC. Interplay of Itch and Psyche in Psoriasis: An Update. Acta Derm Venereol 2016a;96(217):55-57.Google Scholar, Tee et al., 2016Tee S.I. Lim Z.V. Theng C.T. Chan K.L. Giam Y.C. A prospective cross-sectional study of anxiety and depression in patients with psoriasis in Singapore.J Eur Acad Dermatol Venereol. 2016; 30: 1159-1164Crossref PubMed Scopus (25) Google Scholar). To better understand the extent to which skin disease severity and itch contribute to depression in psoriasis, we conducted a real-world registry study on the DermaReg register. DermaReg enrolls patients treated with systemics for psoriasis in Stockholm, Sweden, and follows them prospectively (Svedbom et al., 2020Svedbom A, Nikamo P, Ståhle M. Interaction between Smoking and HLA-C*06:02 on the Response to Ustekinumab in Psoriasis. J Invest Dermatol 2020;140(8):1653-6.e1.Google Scholar, Svedbom and Ståhle, 2020Svedbom A. Ståhle M. Real-world comparative effectiveness of adalimumab, etanercept and methotrexate: a Swedish register analysis.J Eur Acad Dermatol Venereol. 2020; 34: 525-532Crossref PubMed Scopus (8) Google Scholar). The study was approved by the Swedish Ethical Review Authority. All patients give written informed consent to participate in DermaReg. The main outcome in this study was depressive symptoms measured using the Montgomery-Åsberg Depression Rating Scale-Self Assessment (MADRS-s) (Svanborg and Asberg, 1994Svanborg P. Asberg M. A new self-rating scale for depression and anxiety states based on the Comprehensive Psychopathological Rating Scale.Acta Psychiatr Scand. 1994; 89: 21-28Crossref PubMed Scopus (490) Google Scholar). MADRS-s covers nine domains and the total score ranges from 0 to 54. We categorized the scores into: “No” (0–12), “mild” (13–19), “moderate” (20–34), and “severe depression” (≥35). In sensitivity analysis we used the Anxiety/Depression (AD) dimension from the EQ5D-3L instrument (EQ5D-3L-AD) (Oppe et al., 2007) as an outcome. Respondents to the EQ5D-3L-AD choose one of three levels for level of anxiety/depression and we categorized these as “No”, moderate”, and “severe depression” The exposures were psoriasis disease severity measured using the Psoriasis Area and Severity Index (PASI) (Fredriksson and Pettersson, 1978Fredriksson T. Pettersson U. Severe psoriasis--oral therapy with a new retinoid.Dermatologica. 1978; 157: 238-244Crossref PubMed Scopus (2347) Google Scholar) and average itch during the last three days measured using a visual analogue scale (VAS) ranging from 0-10 (Reich et al., 2016bReich A. Riepe C. Anastasiadou Z. Mędrek K. Augustin M. Szepietowski J.C. et al.Itch Assessment with Visual Analogue Scale and Numerical Rating Scale: Determination of Minimal Clinically Important Difference in Chronic Itch.Acta Derm Venereol. 2016; 96: 978-980Crossref PubMed Scopus (59) Google Scholar). We included the following confounders: Sex, age, current smoking, risk use of alcohol, and body mass index (BMI). Covariates were obtained as close as possible to the visit with exposure and outcome data. In terms of mediators, we investigated the mediating effect of the Dermatology Life Quality Index (DLQI) (Finlay and Khan, 1994Finlay A.Y. Khan G. Dermatology Life Quality Index (DLQI)—a simple practical measure for routine clinical use.Clinical and experimental dermatology. 1994; 19: 210-216Crossref PubMed Scopus (3752) Google Scholar) on the associations between PASI and Itch-VAS and MADRS-s. Should the association between MADRS-s and the exposures be fully mediated by the DLQI, assessing the DLQI may be sufficient to measure the impact of signs and symptom of psoriasis on depression. We included all observations with full data on exposures and outcomes in the analysis. We described the crude associations between exposure and outcomes by estimating the prevalence of depression by deciles for each exposure. We also estimated the Pearson correlation between the exposures and outcome. To estimate and compare the impact of itch-VAS and PASI on depression we centered and scaled PASI and Itch-VAS by their respective means and standard deviations. We estimated the associations between mean standardized (MS) exposures and mild, moderate, and severe depression using three separate logistic regression models with confounders for each outcome: One for each exposure and one including both exposures. We analyzed moderate and severe depression defined using the EQ5D-3L-AD dimension in the same way. All regression models were fit with cluster robust standard errors (Cameron and Miller, 2015Cameron A.C. Miller D.L. A practitioner’s guide to cluster-robust inference.Journal of human resources. 2015; 50: 317-372Crossref Scopus (2040) Google Scholar). To explore the association between individual depressive symptoms and PASI and itch-VAS we estimated MS PASI and itch-VAS for each level for each of the nine MADRS-s dimensions. We estimated the mediating effect of DLQI on the associations between PASI and MADRS-s and Itch-VAS and MADRS-using a 4-way decomposition technique (VanderWeele, 2014VanderWeele T.J. A unification of mediation and interaction: a 4-way decomposition.Epidemiology. 2014; 25: 749-761Crossref PubMed Scopus (283) Google Scholar). All analyses were performed in STATA 17.2 and p-values below 0.05 were considered statistically significant. 460 patients had 3,053 observations with complete data. Table 1 presents patient characteristics, the number of visits per patient, and distribution of mild, moderate, and severe depression across those visits.Table 1Patient characteristics and depression categorized across all visitsVariableMeasurementPatients461Age, mean (SD)49.0 (15.9)Male sex, n(%)298 (64.6%)Current smoking99 (21.5%)Risk use of alcohol*81 (17.6%)BMI, mean (SD)28.7 (5.8)Age at onset, n (%)0-1061 (13.2%)11-20143 (31.0%)21-30103 (22.3%)30-50101 (21.9%)50+40 (8.7%)Missing13 (2.8%)Highest PASI during follow-up, median (IQR)11.2 (6.6, 17.0)Highest Itch-VAS during follow-up, median (IQR)3.5 (1.3. 6.9)Nail lesions, n (%)258 (56.0%)Psoriatic arthritis, n (%)149 (32.3%)Registered visits, median (IQR)6 (3,10)Visits with categorized depression according to MADRS-S, n (%)Mild327 (10.7%)Moderate282 (9.2%)Severe60 (2.0%)Any669 (21.9%)Visits with categorized anxiety or depression according to EQ-5D, n (%)Moderate990 (32.4%¤)Severe170 (5.6%¤)Any1160 (38.0%)Note: SD stands for standard deviation, PASI for Psoriasis Area and Severity Index. Risk use of alcohol was defined as more than 4 standard drinks on a single occasion or more than 14 standard drinks per week for men and more than 9 standard drinks per week for women. All patient characteristics were measured as close to the first observation as possible, except, PASI, nail lesions, itch, and psoriatic arthritis which were evaluated over the full follow-up. Categorized depression is measured across all 3,054 registered visits for the 461 patients. Open table in a new tab Note: SD stands for standard deviation, PASI for Psoriasis Area and Severity Index. Risk use of alcohol was defined as more than 4 standard drinks on a single occasion or more than 14 standard drinks per week for men and more than 9 standard drinks per week for women. All patient characteristics were measured as close to the first observation as possible, except, PASI, nail lesions, itch, and psoriatic arthritis which were evaluated over the full follow-up. Categorized depression is measured across all 3,054 registered visits for the 461 patients. The prevalence of any depression increased monotonically with itch-VAS (Figure 1A). The increase was approximately 4-fold from 13.8% in the 0-50th itch-VAS percentiles to 53.4% in the 90th-100th percentiles. For PASI, the corresponding increase was not monotonic, albeit the prevalence of any depression increased approximately two-fold from 18.3% in the 0-50th PASI percentiles to 37.0% in the 90th-100th percentiles (Figure 1B). The Pearson correlation coefficients was higher for Itch-VAS and MADRS-s (0.42; 95% CI: 0.36-0.47) than for PASI and MADRS-s (0.17; 95% CI: 0.11-0.24). The associations between MS itch-VAS and mild, moderate, and severe depression controlling for confounders ranged from Odds Ratio (OR) 1.94 (95% CI: 1.69-2.22) to OR 2.20 (95% CI: 1.76-2.76). These associations were higher than the corresponding associations for PASI and MADRS-s which ranged from OR 1.37 (95% CI: 1.20-1.57) to OR 1.45 (95% CI: 1.22-1.73) (Figure 1C). Furthermore, the associations between Itch-VAS and mild, moderate, and severe depression remained stable whereas the associations between PASI and depression decreased and were not statistically significant when both PASI and Itch-VAS were included in the same regression models (Figure 1C). The associations were both qualitatively and quantitatively similar when moderate and severe depression were defined using EQ5D-3L-AD (Figure 1D). Figure 1E and 1F present the MS Itch-VAS and PASI scores per level in each of the nine MADRS-S domains. Evident from the figures, both PASI and Itch-VAS were associated with each of the domains, but the associations were stronger for itch-VAS than PASI across all domains. In the four-way decomposition analysis, DLQI fully mediated PASI, whereas 24% of the full effect of itch-VAS on MADRS-s was independent of DLQI (Supplementary Table S1 and S2). This study found that symptoms of psoriasis, measured using itch-VAS, was more strongly associated with depression than signs of psoriasis, measured using PASI. In fact, PASI had no independent effect on depression above and beyond itch-VAS. The importance of itch for depression in psoriasis was underscored by the findings that the itch-VAS affected all nine domains in MADRS-s and influenced depression independent of its effect on DLQI. This study has several limitations including limited generalizability, possible residual confounding, and measurement error. In addition, treatment and markers for systemic inflammation are two potentially important factors that were not accounted for. Furthermore, we only measured itch using one instrument, limiting the aspects of itch captured in the study. In conclusion, it is important to evaluate itch in patients with psoriasis and discuss its impact on mental well-being, even beyond its impact on quality of life. Oppe et al., 2007Oppe M, Devlin NJ, Szende A. EQ-5D value sets: inventory, comparative review and user guide: Springer, 2007.Google Scholar. We thank Helena Grieshel for data collection. Table S1Key results from 4-way decomposition: delta methodEffect and decompositionCoefficientStandard errorP-value95% confidence intervalTotal effect1.290.13<0.0011.04 ; 1.54Proportion controlled direct effect-0.220.120.064-0.46 ; 0.01Proportion reference interaction-0.030.010.004-0.04 ; -0.01Proportion mediated interaction-0.050.020.002-0.08 ; -0.02Proportion pure indirect effect1.300.11<0.0011.07 ; 1.52Note: The table shows the total effect on MADRS-S from PASI and divides the total effects into four components: Controlled direct effect, two interaction effects, and the pure (not attributable to interaction) indirect effect mediated via DLQI. In the analysis, the controlled direct effect of PASI on MADRS-s was not statistically significant. Open table in a new tab Table S2Key results from 4-way decomposition: delta methodEffect and decompositionCoefficientStandard errorP-value95% confidence intervalTotal effect3.080.15<0.0012.79 ; 3.38Proportion controlled direct effect0.240.04<0.0010.16 ; 0.33Proportion reference interaction-0.010.000.002-0.01 ; 0.00Proportion mediated interaction-0.060.01<0.001-0.09 ; -0.03Proportion pure indirect effect0.820.04<0.0010.73 ; 0.91Note: The table shows the total effect on MADRS-S from Itch-VAS and divides the total effects into four components: Controlled direct effect, two interaction effects, and the pure (not attributable to interaction) indirect effect mediated via DLQI. In the analysis, the controlled direct effect of Itch-VAS on MADRS-s was statistically significant. 24% of the effect of Itch-VAS on MADRS-s was direct (p<0.001). Open table in a new tab Note: The table shows the total effect on MADRS-S from PASI and divides the total effects into four components: Controlled direct effect, two interaction effects, and the pure (not attributable to interaction) indirect effect mediated via DLQI. In the analysis, the controlled direct effect of PASI on MADRS-s was not statistically significant. Note: The table shows the total effect on MADRS-S from Itch-VAS and divides the total effects into four components: Controlled direct effect, two interaction effects, and the pure (not attributable to interaction) indirect effect mediated via DLQI. In the analysis, the controlled direct effect of Itch-VAS on MADRS-s was statistically significant. 24% of the effect of Itch-VAS on MADRS-s was direct (p<0.001).