- Authors: Daniela Carvalho, Pedro Aguiar, Paulo Ferrinho
- Publication Year: 2021
- Journal: International Journal of Dermatology, 60(2), pp e45-e47
- Link: https://doi.org/10.1111/ijd.15237
CORRESPONDENCE
Dear Editor,
The Skindex-29 instrument has been widely studied and can be used comfortably in clinical and research settings that need to measure changes in quality of life (QoL) impact. In addition, Skindex-29 has a broad emotional component and can be used to further characterize the emotional impact of skin disease. It is one of the most used questionnaires and the first choice in dermatological practice in several countries.
As a high receiver operating characteristic (ROC) curve level (per dimension and total) was found in a validation study, it was possible to validate a high discriminatory power for the Skindex-29 index. Thus, with a larger Portuguese community sample, optimal cutoffs were determined based on ROC curve analysis, maximizing sensitivity and specificity in relation to the disease. The sample was mainly patients with atopic dermatitis (AD), other eczemas, and urticaria, with a median (range) age of 21.5 (0.5–74), 43.0 (9–59), and 42.0 (7–90), respectively, and most patients were female. Further characteristics as well as the mean of Skindex scores are presented in an already published article.
Cutoff values were determined from ROC curves with area under the curve (AUC) >0.6 – that is, with a good discriminative power – using Skindex-29 scores by disease and disease severity in order to find a cutoff to distinguish between more negatively impaired QoL vs. less impaired QoL. For selection of this cutoff, the criteria were: (i) sensitivity and specificity values above 50% and (ii) sensitivity greater than specificity. The value that maximized the sum of sensitivity and specificity was selected as the optimal cutoff.
(…) Only results relevant to overall AD and severe AD were obtained. The present cutoffs categorize the sample into patients with negatively impaired QoL vs. less impaired QoL. The cutoff for our overall AD cohort was 31.35 while that for the severe AD cases was 26.46. Patients’ perceptions of QoL are not highly correlated with clinical assessments of the AD severity, varying from no, positive, or negative correlation. This may be the reason for the observed lower cutoffs among the severe AD patients. Having an inferior specificity, there might be some false positives, that is, some patients with score >26 where the QoL is less impaired.
Skindex revealed the ability to discriminate severe disease in a certain cutoff. We highlight the fact that this sample is limited. Only three diseases, individually, were included, and only one country is represented. Despite not being a robust sample, the found cutoffs are aligned with the previously, and a few, published cutoffs. Additionally, analyses with different statistical tests and methodology provide information on Skindex cutoffs in order to identify patients with more impaired QoL. It is crucial to know how to interpret the data resulting from QoL scales. In this way, these findings came to increase knowledge and to underline prior calculations.