ORIGINAL ARTICLE
Tamara W. VAN HAL1,2, Michelle L. M. MULDER1,2, Mark H. WENINK1, Frank H. J. VAN DEN HOOGEN3, Jake S. F. MAURITS, Marcel C. PASCH5, Juul M. P. A. VAN DEN REEK2,5 and Elke M. G. J. DE JONG3,5
1Department of Rheumatology, Sint Maartenskliniek, 2Radboud Institute for Health Sciences, Radboud University Medical Center, 3Faculty of Medical Sciences, Radboud University, 4Department for Health Evidence and 5Department of Dermatology, Radboud University Medical Center, Nijmegen, The Netherlands
Patients with psoriasis are at risk of developing psoriatic arthritis, which can lead to joint damage. While screening questionnaires have been developed, their performance varies. The objective of this study was to develop a referral tool for dermatologists to identify psoriasis patients with concomitant psoriatic arthritis for rheumatological referral. This study used data from the DAPPER study, in which psoriasis patients were screened by a rheumatologist for the presence of concomitant psoriatic arthritis. Multivariable regression analysis was used to identify predictive variables for the presence of concomitant psoriatic arthritis: treatment history with conventional systemic drugs (odds ratio (OR) 2.97, 95% confidence interval (95% CI) 1.01–8.74, p = 0.04), treatment history with biologicals/small molecule inhibitors (OR 2.90, 95% CI 1.52–5.53, p = 0.01), patient-reported history of joint pain not caused by trauma (OR 4.23, 95% CI 1.21–14.79, p = 0.01), patient-reported history of swollen joints (OR 4.25, 95% CI 2.17–8.32, p < 0.001), and patient-reported history of sausage-like swollen digits (OR 2.38, 95% CI 1.25–4.55, p = 0.01). Based on these variables, a referral tool was created with an area under the curve of 0.82. This referral tool could be used to aid dermatologists to identify psoriasis patients with concomitant psoriatic arthritis, who may benefit from rheumatological referral.
Key words: psoriasis; psoriatic arthritis; screening.
Psoriasis patients are at risk of psoriatic arthritis. However, it is difficult for dermatologists to select patients who could benefit from rheumatological referral. In a dermatology psoriasis cohort, variables associated with concomitant psoriatic arthritis were identified: treatment history with: (i) conventional systemics, or (ii) biologicals/small molecule inhibitors; and patient-reported history of: (iii) swollen joints, (iv) sausage-like swollen fingers, or (v) joint pain not associated with trauma. Using these variables, a referral tool was developed for dermatologists to identify psoriasis patients with concomitant psoriatic arthritis. By using wide inclusion criteria and a large cohort, potential pitfalls, such as selection bias, were minimized.
Citation: Acta Derm Venereol 2023; 103: adv5269. DOI https://doi.org/10.2340/actadv.v103.5269.
Copyright: © Published by Medical Journals Sweden, on behalf of the Society for Publication of Acta Dermato-Venereologica. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (https://creativecommons.org/licenses/by-nc/4.0/)
Accepted: Mar 2, 2023; Published: Apr 27, 2023
Corr: Tamara W. van Hal, Department of Rheumatology, Sint Maartens-kliniek, Hengstdal 3, PO Box 9011, NL-6500 GM Nijmegen, The Netherlands. E-mail: t.vanhal@maartenskliniek.nl
Competing interests and funding: TvH received personal fees from Eli Lily and Novartis, non-financial support from UCB, outside the submitted work. MM received non-financial support from UCB, outside the submitted work. JvdR carried out clinical trials for AbbVie, Celgene and Janssen and has received speaking fees/attended advisory boards from AbbVie, Janssen, BMS, Almirall, LEO Pharma and Eli Lilly and reimbursement for attending a symposium from Janssen, Pfizer, Celgene and AbbVie. All funding is not personal, but goes to the independent research fund of the department of dermatology of Radboud University Medical Center Nijmegen, the Netherlands. EdJ has received research grants for the independent research fund of the department of dermatology of the Radboud University Medical Center Nijmegen, the Netherlands from AbbVie, BMS, Janssen Pharmaceutica, Leo Pharma, Lilly, Novartis, and UCB for research on psoriasis and has acted as consultant and/or paid speaker for and/or participated in research sponsored by companies that manufacture drugs used for the treatment of psoriasis or eczema including AbbVie, Amgen, Almirall, Celgene, Galapagos, Janssen Pharmaceutica, Lilly, Novartis, Leo Pharma, Sanofi and UCB. All funding is not personal, but goes to the independent research fund of the department of dermatology of Radboud University Medical Center Nijmegen, the Netherlands. All other authors have no conflicts of interest to declare.
One in three patients with psoriasis (PsO) attending the dermatology clinic will develop psoriatic arthritis (PsA), which can lead to disability, discomfort, and irreversible joint damage (1, 2). In the majority of patients, PsO precedes the development of PsA (3). Early treatment of arthritis is important to prevent joint damage, and to improve the physical functioning and quality of life of affected patients (4, 5). Therefore, early recognition by dermatologists and rheumatological referral of PsO patients with arthritis is crucial. Unfortunately, a considerable proportion of PsO patients with PsA are not diagnosed in clinical practice (6).
To aid dermatologists in selecting patients with a high risk of PsA, several screening questionnaires have been developed (7–15). Nevertheless, the diagnostic accuracy of these questionnaires varies widely between studies (16). For the most-studied questionnaires (Psoriatic Arthritis Screening and Evaluation (PASE) tool) (8), Psoriasis Epidemiology Screening Tool (PEST) (7), Toronto Psoriatic Arthritis Screen (ToPAS) (10)), sensitivities were in the range 24–100%, 28–92%, and 41–96%, while specificities were in the range 20–94%, 37–98%, and 30–97%, respectively (16).
Due to varying performance results, we developed a new cohort to overcome some of the problems encount-ered in the development of the above-mentioned tools (17). Specifically, by using an outpatient dermatology cohort with a sufficient number of PsO patients with concomitant PsA relative to the number of possible predictive parameters, we aimed to avoid overfitting (7, 14, 15) and the need to enrich the sample with PsA patients from other sources (e.g. the rheumatology department) (7, 10, 14).
The objective of this study was to develop a new referral tool to aid dermatologists in identifying PsO patients with concomitant PsA. We selected patients with concomitant PsA in a cohort of Pso patients at the dermatology outpatient clinic. Parameters that distinguished PsO patients with and without concomitant PsA were identified and used to build a new referral tool. In addition, we explored the possibility to build a referral tool to identify PsA patients with active PsA, because these are most likely to benefit from rheumatological referral.
We used data from the prospective observational DAPPER study, conducted at the department of dermatology of Radboud University Medical Center from June 2019 until April 2022. The study protocol and initial results have been published previously (17, 18). Briefly, 304 adult patients with PsO visiting the dermatology outpatient clinic were included. Patients were stratified 1:1:1 for current treatment modality (topicals only, conventional systemics, biologicals/small molecule inhibitors (SMI)). Patients with previously diagnosed, concomitant PsA were not excluded. Patients were screened by a rheumatologist at the dermatology outpatient clinic for signs and symptoms of PsA, using a structured interview and physical examination (Appendix S1). If PsA was suspected at study visit, and the patient was not currently treated by a rheumatologist, they were referred to a rheumatology centre for additional examinations and confirmation of PsA diagnosis. The study was approved by the medical ethics committee of the region Arnhem-Nijmegen, Radboudumc (NL68137.091.18), registered prospectively in the Dutch Trial Register (NTR 7604), and performed according to the Declaration of Helsinki and Good Clinical Practice. The current report was written according to Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD) guidelines (19).
The outcome of the prediction model was presence of concomitant PsA. A patient was classified as “PsO with concomitant PsA” if either they had been previously diagnosed by a rheumatologist, or if a (new) diagnosis of PsA was made after study referral to the rheumatology department. Patients without a previous diagnosis of PsA, and patients without signs/symptoms of concomitant PsA, or with rejection of PsA diagnosis after referral, were classified as “PsO only”. Patients with “PsO with concomitant active PsA” were PsO patients with concomitant PsA who, in addition, fulfilled the following criteria at study visit: ≥ 1 swollen joint and/or active enthesitis and/or active axial spondyloarthritis. In case of suspicion of active enthesitis and/or axial spondyloarthritis, affirmation by imaging was required. Patients with “PsO with concomitant inactive PsA” were PsO patients with a previous diagnosis of concomitant PsA who did not have swollen joints, active enthesitis, or active axial spondyloarthritis at study visit.
During the study visit, the following variables were collected via structured interviews and chart reviews: demographics, intoxications, family history of PsO and PsA, treatment history, comorbidity, (previous) disease activity of skin and nails (Psoriasis Area and Severity Index (PASI), range 0–72; (20), Nail Psoriasis Severity Index (NAPSI), range 0–160 (21); Nijmegen Nail Psoriasis Activity Index Tool (N-NAIL), range 0–150 (22)); (previous) signs and symptoms of joint disease, and questions from several screening questionnaires for PsA in PsO (Early Psoriatic Arthritis Screening Questionnaire (EARP, PEST, and ToPAS)) (7, 9, 10).
This study included patients from the DAPPER-study (n = 304). For the prediction model, we aimed to use a maximum of 10 parameters with a restriction of 1 parameter per 10 events. Therefore, assuming a prevalence of PsA in PsO of 30% (23), the current study included 300 PsO patients.
Data were described with mean (standard deviation; SD), median (interquartile range; IQR), or absolute frequencies (percentages), where appropriate.
Possible associations between disease or patient characteristics and presence of PsA were explored using logistic regression. Missing data were not imputed. All models presented are based on complete cases.
For possible predictors, dichotomous questions (yes/no, presence/absence) were included to ease use in clinical practice. Because the study included patients with a known PsA diagnosis in the development cohort, questions referring to previous diagnosis of arthritis were not included (e.g. “Did a doctor ever tell you that you have arthritis?”).
Possible predictive variables were preselected in 2 steps for entry in the multivariable model. First, univariable logistic regression was used to select variables with a p < 0.20. Secondly, variables with overlapping concepts (based on biological plausibility and/or collinearity) were removed. Both forward and backward selection multivariable logistic regression models were employed. p < 0.05 was considered significant in the multivariable regression models. The area under the receiver operating characteristics (ROC) curve (AUC) was used to assess the performance of the models.
Internal validity was assessed by estimating the optimism of the models using repeated K-fold cross-validation, with 10 splits and 20 repeats. A sensitivity analysis was performed, in which a scenario was created in which patients were reclassified with an uncertain diagnosis (n = 4). These were classified as PsO with concomitant PsA in the original scenario, and in the sensitivity analysis they were classified as PsO only.
Based on the variables associated with concomitant PsA, a referral tool for dermatologists was developed. The goal of the referral tool was to alert the dermatologist when PsO patients have a high chance of concomitant PsA. If these patients are not under current rheumatological care, a referral to a rheumatologist could be considered. Test characteristics of the referral tool were tested using 2-by-2 tables to assess sensitivity and specificity.
Using the same methodology (i.e. logistic regression analysis followed by the construction of a referral tool), we explored the possibility of developing a referral tool for active PsA only. For this analysis, we compared the patient groups “PsO only” and “PsO with concomitant inactive PsA” vs “PsO with concomitant active PsA”.
All analyses were performed in SPSS Statistics software version 25.0 (IBM, Armonk, NY, USA) and R studio version 3.6.2 (RStudio Inc, Boston, MA, USA) using the caret package.
A total of 303 PsO patients of the DAPPER study were included in this study (drop-out n = 1). Mean age was 54 ± 16 years, 109/303 patients (36%) were female. Seventy-four percent of patients (225/303) were classified as PsO only; 17% as having concomitant inactive PsA (50/303); and 9% as having active PsA (28/303). Clinical characteristics of the cohort are shown in Table I.
PsO only n = 225 | PsO + PsA n = 78 | |
Age, years, mean (SD) | 53 (17) | 54 (15) |
Female sex, n (%) | 80/225 (36) | 29/78 (37) |
Body mass index, kg/m2, mean (SD) | 28.6 (5.7) | 29.1 (5.8) |
Smoking ever, n (%) | 159/225 (71) | 50/78 (64) |
Physically taxing job, n (%) | 41/225 (18) | 17/78 (22) |
Trauma in the past year, n (%) | 74/225 (33) | 28/78 (36) |
Family history, n (%) | ||
PsO | 128/225 (57) | 47/78 (60) |
PsAa | 34/224 (15) | 14/78 (18) |
Comorbidity, n (%) | ||
Major adverse cardiovascular event | 24/225 (11) | 9/78 (12) |
Depression | 25/225 (11) | 11/78 (14) |
Current therapy: No systemic | 85/225 (38) | 15/78 (19) |
Conventional systemic drugs, n (%) | ||
All | 88/225 (39) | 26/778 (33) |
Methotrexate | 63/225 (28) | 17/78 (22) |
Acitretin | 8/225 (4) | 3/78 (4) |
Fumaric acid | 14/225 (6) | 3/78 (4) |
Cyclosporine | 2/225 (1) | 0/78 (0) |
Biologicals/small molecule inhibitors, n (%) | ||
All | 57/225 (25) | 44/78 (56) |
Tumour necrosis factor-inhibitor | 28/225 (12) | 19/78 (24) |
Interleukin 17-inhibitor | 9/225 (4) | 12/225 (15) |
Interleukin 23-inhibitor | 1/225 (1) | 1/78 (1) |
Interleukin 12/Interleukin 23 p40 inh. | 19/225 (8) | 10/78 (13) |
Phosphodiesterase 4-inhibitor | 0/225 (0) | 2/78 (3) |
Skin disease, current, median (IQR) | ||
Age at startb | 27 (16, 44) | 23 (15, 32) |
Disease durationb | 21 (10, 35) | 2.7 (1.7, 3.9) |
Psoriasis Area and Severity Indexa | 2.8 (1.6, 4.5) | 2.4 (1.1, 4.0) |
Nail Psoriasis Severity Indexc | 15 (6, 26) | 12 (5, 20) |
Nijmegen Nail Psoriasis Activity Indexc | 4 (1, 10) | 4 (1, 9) |
Joint complaints, current, n (%) | ||
Joint pain | 159/225 (71) | 67/78 (86) |
Back pain | 95/225 (42) | 41/78 (53) |
Morning stiffness ≥ 30 min | 26/225 (12) | 19/78 (24) |
aMissing in 1 patient with PsO only. bMissing in 9 patients with PsO only, and 4 patients with PsO + PsA. cMissing in 44 patients with PsO only, and 7 patients with PsO + PsA. | ||
SD: standard deviation, IQR: interquartile range. |
Univariable logistic regression was used to compare the clinical characteristics of patients with PsO only and patients with PsO with concomitant PsA (Appendix S1; Tables SII and SIII). Using a cut-off of p < 0.2, 25 variables were deemed statistically relevant. By eliminating overlapping variables, 11 variables remained for input in the multivariable model.
Table II shows the results of multivariable logistic regression models using forward and backward selection. Both forward and backward selection showed independent association of presence of concomitant PsA with: treatment history with conventional systemics (odds ratio (OR) 2.97, 95% confidence interval (95% CI) 1.01–8.74, p = 0.04), treatment history with biologicals/SMI (OR 2.90, 95% CI 1.52–5.53, p = 0.01), patient-reported history of joint pain not caused by trauma (OR 4.23, 95% CI 1.21–14.79, p = 0.02), patient-reported history of swollen joints (OR 4.25, 95% CI 2.17–8.32, p < 0.001), and patient-reported history of sausage-like swollen digits (OR 2.38, 95% CI 1.25–4.55, p = 0.01). Overall fit of this multivariable logistic regression model as determined by AUC was 0.83.
Univariable OR (95% CI) | Multivariable OR (95% CI) | |
Treatment history | ||
All conventional systemic | 4.72 (1.82–12.28) | 2.97 (1.01–8.74) |
All biological/small molecule inhibitor | 3.80 (2.21–6.52) | 2.90 (1.52–5.53) |
Skin disease ever: Erythroderma | 1.68 (0.77–3.69) | |
Nail disease ever: Holes/pits | 2.32 (1.35–3.99) | |
Joint complaints ever | ||
Non-trauma joint pain | 9.30 (2.83–30.59) | 4.23 (1.21–14.79) |
Swollen joints | 6.62 (3.65–12.01) | 4.25 (2.17–8.32) |
Swollen digits | 4.53 (2.62–7.84) | 2.38 (1.25–4.55) |
Heel pain | 1.54 (0.88–2.69) | |
Joint complaints current | ||
Joint pain | 2.53 (1.26–5.09) | |
Back pain | 1.52 (0.90–2.54) | |
Morning stiffness | 2.47 (1.28–4.77) | |
Intercept | –4.89 | |
Area under curve | 0.83 | |
Possible predictors for PsA in PsO patients were tested using multivariable logistic regression. After elimination of overlapping variables, predictors with a p-value ≤ 0.20 were inserted in the multivariable model. Odds ratios (ORs) (PsO only vs PsO with concomitant PsA) are depicted with 95% confidence interval (CI). Complete regression formulas are shown in Appendix S1; Table SVII. |
The optimism of the model was estimated using repeated K-fold validation. The AUC of the model was 0.83, and the AUC of the internal validation model was 0.82, giving an optimism of 0.01.
In the sensitivity analyses, patients who were not referred to the rheumatologist but did have a suspicion of PsA at study visit (n = 4) were re-classified as PsO only instead of PsO with concomitant PsA. This analysis denoted the same 5 variables as independent predictors, as shown in Appendix S1; Table SIV.
Based on the results of the above-mentioned analyses, we developed a referral tool for dermatologists to help them identify Pso patients with concomitant PsA. The following variables were included: treatment history with conventional systemics, treatment history with biologicals/SMI, patient-reported history of joint pain not caused by trauma, patient-reported history of swollen joints, and patient-reported history of sausage-like swollen digits. Every variable was scored 1 point if present, and 0 points if absent. ROC curve of this 5-variable model showed an AUC of 0.82.
To increase ease of use, and to anticipate the increased use of biologicals/SMI without earlier treatment of conventional systemics (as is recommended in treatment guidelines for PsA (24)), we also made a version in which we combined the variables “treatment history with conventional systemics” and “treatment history with biologicals/SMI “ into a single variable “treatment history with systemic medication”. The ROC curve of this 4-variable model showed an AUC of 0.80. Table III shows the sensitivity and specificity of both versions of the referral tool at different cut-off points.
Using the same methodology, we also explored the possibility to develop a referral tool to identify only PsO patients with concomitant active PsA. Appendix S1; Table SV shows the results of logistic regression analysis comparing the patient groups “PsO only” plus “PsO with concomitant inactive PsA” vs “PsO with concomitant active PsA”. Backward selection multivariable logistic regression analysis showed independent associations of active PsA with: a treatment history with biologicals/SMI (OR 3.33, 95% CI 1.44–7.71, p = 0.01) and current joint pain (OR 9.60, 95% CI 1.27–72.38, p = 0.03). Overall fit of the backward selection model as determined by AUC was 0.73. Forward selection multivariable logistics regression analysis also showed independent associations with a patient-reported presence of prolonged morning stiffness (OR 2.34, 95% CI 0.96–5.70, p = 0.06), in addition to a treatment history with biologicals/SMI (OR 2.92, 95% CI 1.24–6.88, p = 0.01), and current joint pain (OR 7.80, 95% CI 1.02–59.72, p = 0.05). Overall fit of the forward selected model, as determined by AUC, was 0.75. Translation of these variables into a referral tool is shown in Appendix S1; Table SVI.
In the DAPPER study, patients with PsO at the dermatology outpatient clinic were investigated for the presence of PsA (17). In this population, the current study identified 5 variables that were independent predictors for the presence of PsA: treatment history with conventional systemics, treatment history with biologicals/SMI, patient-reported history of swollen joints, patient-reported history of sausage-like swollen digits, and patient-reported history of joint pain not caused by trauma. Using these variables, a referral tool was developed to aid dermatologists in identifying PsO patients with concomitant PsA.
The referral tool included items about treatment history and musculoskeletal signs and symptoms, i.e. pain and swelling. Joint swelling is considered to discriminate between inflammatory and non-inflammatory joint diseases, while sausage-like swelling of the digits (dactylitis) is considered a hallmark of PsA (25). The item “history of joint pain not caused by trauma” is derived from the ToPAS questionnaire (10). While several other questionnaires include items enquiring about joint pain in general (8, 9, 26) or joint pain combined with redness and/or swelling (11, 13), a history of joint pain not caused by trauma is unique to ToPAS. Interestingly, in the current study cohort, a history of joint pain not caused by trauma was independently associated with concomitant PsA, while current joint pain was not. Presumably, the partial overlap of patients answering yes to both variables is the reason only one was selected using the backward/forward selection procedures.
The item “treatment history with systemic medication” has, to our knowledge, not been used previously to identify PsO patients with concomitant PsA. The relationship between the use of systemic medication and the risk of PsA is still unclear. Since the biologicals/SMI used for PsO are also effective for PsA, a protective effect is biologically plausible (27). However, PsO patients who use biologicals/SMI can still develop PsA (28). A higher burden of skin involvement is associated with a higher prevalence of PsA, and patients with more severe skin involvement are more likely to receive systemic medication (29). Moreover, patients with joint complaints are at a higher risk of PsA, and physicians might be more inclined to intensify treatment if joint complaints are present (protopathic bias) (30, 31).
Remarkably, prevalence of nail disease ever and heel complaints, 2 items which are present in many other screening questionnaires, did not reach significance in our multivariable model (16). Recently, Cui et al. tested 4 different questionnaires in a Japanese PsO population, and extracted key questions which were discriminative between PsO only and PsO with concomitant PsA. Previous nail disease and heel complaints were also not found to contribute significantly to the distinction between both patient groups (32). In contrast, in 2014, Coates et al. (12) found nail disease and heel complaints to be contributory. We hypothesize that, while the prevalence of previous nail disease and heel complaints are indeed higher in PsO with concomitant PsA (as shown by the univariable models), this effect is overshadowed by the discriminative capabilities of the other items in the referral tool.
Ideally, any referral tool should have a balance between sensitivity and specificity. We believe that, based on the current data, the 4-variable-test (ever use of systemic medication, non-traumatic joint pain, swollen joints, and swollen fingers) with a cut-off of 3 or higher has the best characteristics for this goal. With a sensitivity of 79%, a specificity of 69% and a prevalence of 26%, this would mean that, out of 100 patients with PsO, half of the patients would be referred, of which again half would have PsA. However, 1 in 5 patients with PsA would be missed.
Comparison of the performance of our referral tool with previously designed screening questionnaires is difficult, because of the large variation in the reported performances in different studies and the different populations used to develop and evaluate these questionnaires (16). In the DAPPER cohort, psoriasis patients with previously diagnosed PsA were not excluded. Because of inclusion of these patients with known PsA, we were unable to include predictors directly related to the PsA diagnosis, such as a question enquiring about a previous arthritis diagnosis by a physician. Inclusion of predictors related to a previous diagnosis would bias the performance results of the tool, leading to an inaccurately high estimation of specificity and sensitivity. However, several previously developed screening questionnaires do contain a question enquiring about a previous diagnosis of arthritis (e.g. PEST, ToPAS, PASQ) (16). In the DAPPER cohort, the sensitivity/specificity of PEST and ToPAS were 71/81% and 75/78%, respectively (17). This is in the same range as the performance of our referral tool. However, due to the use of the “previous diagnosis” question, the performance of PEST and ToPAS in this cohort might be inaccurately high.
Because patients with currently active PsA are most likely to benefit from referral to, and thus co-treatment by, a rheumatologist, the current study also explored the option of a referral tool to identify patients with active PsA. However, the analysis was hampered by a low number of events (n = 28 with active PsA), therefore the results must be interpreted with caution. Moreover, the performance of the model identifying active PsA only was low (AUC 0.75). Therefore, we conclude that the data gathered in our cohort were insufficient to develop a useful tool to identify patients with active concomitant PsA.
Limitations of the current study are the setting in an academic psoriasis expertise centre, and the inclusion of patients with known PsA in the study cohort. However, inclusion of these patients also made it possible to only use patients from the dermatology outpatient clinic, with-out the need to “supplement” cases from a rheumatology clinic. Moreover, the use of an “unfiltered” PsO population at the dermatology clinic (e.g. including patients with and without medication, in contrast to the EARP questionnaire (9)) improved the generalizability of the current study results. Another strength is the study size, with enough events relative to the amount of possible predictive parameters, minimalizing the risk of overfitting. In the future, validation of the DAPPER referral tool in a second validation cohort should be performed, preferably in a multicentre setting involving both academic and non-academic centres.
With this prospective observational study a referral tool was developed to aid dermatologists in identifying PsO patients with concomitant PsA. The study showed that a patient-reported history of swollen joints, sausage-like swollen digits, joint pain not caused by trauma, and a treatment history with systemic medication are independent risk factors for the presence of concomitant PsA in patients with PsO. To improve the detection of PsO patients with concomitant PsA, future research could benefit from collaborations forming large, combined cohorts of screened PsO patients, such as the Hippocrates consortium (33). In addition, the use of clinical parameters only may not be sufficient to adequately distinguish PsO patients with and without concomitant PsA. The combination of clinical parameters with laboratory and genetic markers could also be further explored as a means of screening (34). In the meantime, use of screening questionnaires is considered a cost-effective approach to improve the care of PsO patients with (undiscovered) PsA (35).
The authors thank Mr Hans Groenewoud for help with the statistical plan, and Mrs Sophie Wanten for help with the data management.
This investigator-initiated study was funded via a PhD grant from Radboud University Medical Center/Sint Maartenskliniek and local structural research funding of the Sint Maartenskliniek.