Non-specific arthritis, fatigue, the presence of pain at different body sites, and swelling represent several musculoskeletal symptoms which often precede the diagnosis of psoriatic arthritis (PsA) and may be indicative of PsA risk in certain patients, according to an exploration of Bayesian networks (BNs) outlined in Rheumatology.
The investigators in this study used the UK Clinical Research Practice Datalink (CPRD) to identify incident psoriasis cases recorded from 1998 to 2015. Medcodes were also used to identify musculoskeletal symptoms. Other variables extracted from the database included baseline demographic information for sex, age, body mass index (BMI), psoriasis severity, as well as alcohol and tobacco use.
A combination of data-oriented modeling and expert knowledge was used to compose several BN structures using primary musculoskeletal symptom groups, musculoskeletal symptom subgroups, and demographic data. The receiver operating characteristic curve (area under the curve [AUC]) was calculated to identify the predictive ability of the networks.
A total of 1409 of the 90,189 patients with incident psoriasis in the study population developed PsA. In the overall study cohort, more than 1,000,000 musculoskeletal symptoms were extracted, and approximately 800 unique symptoms were identified from patients’ CPRD records. There were direct associations between PsA and sex, BMI, finger pain, fatigue, hip pain, swelling, back pain, and myalgia.
A BN constructed using site specific musculoskeletal symptom subgroups featured a 76% accuracy in the prediction of PsA development in the test set. In addition, this BN featured an AUC of 0.73 (95% CI, 0.70-0.75).
A limitation of this study was the researchers’ ability to obtain confirmation of a rheumatologist diagnosis of PsA.
The investigators concluded that patients who experience “these symptoms may have PsA earlier than their PsA diagnosis and thus should be followed up more closely since they are more likely to develop the disease in subsequent years.”
Green A, Tillett W, McHugh N, Smith T; PROMPT study group. Using Bayesian Networks to identify musculoskeletal symptoms influencing the risk of developing Psoriatic Arthritis in people with psoriasis. Rheumatology (Oxford). Published online March 26, 2021. doi:10.1093/rheumatology/keab310