Risk factors for the development of psoriatic arthritis: a population based nested

Objective. To identify factors influencing the development of psoriatic arthritis (PsA) in a population-based, inception cohort of psoriasis (PS) patients. Methods. Using the population-based data resources of the Rochester Epidemiology Project, which ensures virtually complete ascertainment of all clinically defined conditions, we previously identified all incident cases of PsA and prevalent cases with PS from 1/1/1982 to 12/21/1991. In this nested case-control study, we assessed potential factors influencing the development of PsA in this cohort using medical record and patient survey information. Each case of PsA was matched with 2 PS controls on age, gender and PS duration/date of onset. Factors influencing the development of PsA were identified, adjusting for the influence of other variables using conditional logistic regression for medical record data and logistic regression for survey data. Results. Sixty incident PsA cases were matched with 120 controls with PS. The median age at onset of PS was 31.7 (3.0–78.3) years, and 49% of subjects were male. There were 67% (n = 40) survey responders among cases and 48% (n = 58) among controls. Corticosteroids were used by 10 cases and 6 controls in the 2 years prior to onset of PS through to the development of PsA, and increased the risk of developing PsA (odds ratio 4.33, 95% CI = 1.34–14.02). Pregnancy occurred in 2 cases and 12 controls in the same period, and decreased the risk of developing PsA (odds ratio 0.19, 95% CI = 0.04–0.95). These associations remained significant after adjusting for the influence of gender, age, and duration of psoriasis. Conclusion. Corticosteroid use and pregnancy, both of which modulate the immune response, may influence the development of PsA in patients with PS. (J Rheumatol 2002;29:757–62) Key Indexing Terms: PSORIATIC ARTHRITIS RISK FACTORS GLUCOCORTICOIDS PREGNANCY CASE-CONTROL STUDIES Personal non-commercial use only. The Journal of Rheumatology Copyright © 2002. All rights reserved. Rheumatology The Journal of on July 9, 2020 Published by www.jrheum.org Downloaded from Medical record abstraction and questionnaire design. Data from an existing database of all patients with PS in Olmsted county, MN from January 1, 1982 to December 31, 1991 was utilized5. In assembling this database, 1056 cases of PS were identified from 1844 potential cases of PS, and the entire medical record was reviewed using a pretested data collection form to collect information on demographics, clinical manifestations, laboratory findings, treatment, and outcome. Additional data on possible risk factors for the development of PsA including surgical procedures, trauma, streptococcal infections, psychological stress, diabetes mellitus, hypertension and treatments for PS (topical or oral medications and phototherapy) were collected by a physician abstractor (JT) using a pretested data collection form and encompassing the period up to the diagnosis of PsA. A pre-tested, self-administered survey questionnaire with a single mailing was utilized to obtain information not usually recorded in the medical record. This included information on family members with PS or PsA, ethnicity, number of years of education, tobacco, alcohol and medication use, pregnancies, and menopause. To minimize recall bias, the questionnaire inquired about the patient’s PS and health in general, and did not specifically mention PsA as a focus of the study. None of the 66 cases declined research authorization at the time of medical record abstraction. Three controls declined research authorization and 3 alternative control patients with PS were therefore selected. This study was approved by the Mayo Clinic and Olmsted Medical Center Institutional Review Boards. Definitions and diagnostic criteria. The diagnosis of PS was based on documentation of typical skin changes of PS in the medical record by a dermatologist. All questionable cases of PS were reviewed with our dermatologist co-investigator (LEG). Skin severity of PS was defined as limited (≤ 2 sites) or generalized (> 2 sites) based on involvement of the scalp, elbows, knees, other flexural areas, palms/soles or other sites. The diagnosis of peripheral joint involvement due to PsA was based on documentation in the medical record of inflammatory synovitis with objective joint swelling noted by a physician in any upper or lower limb joint in a patient with PS. Patients with seropositive rheumatoid arthritis, Reiter’s syndrome, crystalline arthritis, arthritis of inflammatory bowel disease, and inflammatory osteoarthritis were excluded, as were patients with uncharacterized inflammatory arthritis who were persistently rheumatoid factor positive. Axial disease was defined as involvement of the cervical, thoracic, or lumbosacral spine with documented inflammatory spinal pain, sacroiliac tenderness, reduced chest expansion or radiographic changes consistent with PsA (sacroiliitis or syndesmophytes). Potential risk factors for PsA. We studied risk factors associated with the development of either PsA or PS, as evidence suggests that shared genetic and environmental factors may influence the development of both the articular6-9 and cutaneous10-12 manifestations of PS. As these factors may influence the development of PsA before or after the onset of PS, their influence was studied in both time intervals. The presence of potential risk factors prior to the onset of PsA was determined from medical record and/or patient survey data. The time period during which individual risk factors might influence the development of PsA (summarized in Table 2) was determined by the plausibility of cause and effect over a period of time, anticipated accuracy of patient recall, and the literature on trigger factors13–18. Risk factors whose influence on the development of PsA was unlikely to vary with time (e.g., menopause, diabetes mellitus) were included in the analysis if they were present at any time before the development of PsA. Risk factors whose influence was likely to be time dependent (e.g. smoking or use of medications) were included if they were present in the 2 years prior to the onset of PS through to the onset of PsA. Medical record ascertained risk factors were defined as follows: trauma as a documented motor vehicle accident, fracture, sprain/ contusion, surgical procedure (including minor surgical procedures) or burn; psychological stress as a documented physician/psychiatrist diagnosis of a depressive illness (major depressive disorder, dysthymic disorder, depressive disorder not otherwise specified), adjustment disorder, anxiety (panic attacks, agoraphobia, acute stress disorder, post-traumatic stress disorder, general anxiety disorder, anxiety related to a general medical condition or substance abuse); hypertension as a blood pressure of > 140/90 on 3 separate occasions19; diabetes mellitus as 2 consecutive outpatient fasting plasma glucose values ≥ 140 mg/dl (7.8 mmol/l) or 1 oral glucose tolerance test for which the 2 hour and one other value were both > 200 mg/dl (11.1 mmol/l); streptococcal infection as positive bacteriological culture of Streptococcus from a The Journal of Rheumatology 2002; 29:4 758 Table 1. Characteristics of cases and controls. Cases, n = 60 Controls, n = 120 n % n % Entire Cohort Male 29 (48.3) 59 (49.2) White 59 (98.3) 118 (99.2) Median age (range) at onset of PS, yrs* 31.6 (3.0–76.5) 31.7 (4.5–78.3) Median age (range) at onset of PsA, yrs* 41.2 (19.8–82.3) 40.6 (16.2–82.4) Median duration (range) of PS at onset of PsA, yrs* 5.8 (0–52.2) 6.2 (–2.4–49.0) Responders Non-responders Responders Non-responders n (%) n (%) n (%) n (%) Survey responders and non-responders Number 40 (66.7) 20 (33.3) 58 (48.3) 62 (51.7) Male 21 (52.5) 8 (40.0) 27 (46.6) 32 (51.6) White 39 (97.5) 20 (100.0) 58 (100.0) 60 (98.4) Median age (range) at 39.7 53.1 42.4 40.4 onset of PA, yrs* (24.0–65.3) (19.8–82.3) (16.2–77.3) (22.9–82.4) Median duration (range) 4.6 9.5 6.0 7.0 of PS at onset of PsA, yrs* (0–52.2) (0–37.9) (–2.4–49.0) (–2.2–47.3) * For controls, the date corresponding to the matched case’s date of onset was used. Personal non-commercial use only. The Journal of Rheumatology Copyright © 2002. All rights reserved. Rheumatology The Journal of on July 9, 2020 Published by www.jrheum.org Downloaded from clinically infected site; HIV infection as Western blot confirmed HIV infection. The patient survey was used to obtain data on ethnicity, use of tobacco (smoked at least 100 cigarettes), alcohol, or specified medications (for which generic and commercial names were stated), number and dates of pregnancies, date of menopause, and the presence of parents or siblings with PsA (defined as “arthritis caused by psoriasis”) or PS. Data on topical corticosteroid use was not collected. Statistical analyses. Dates of onset of PS and PsA were used to define time intervals for analysis. Demographic characteristics and possible risk factors influencing the development of PsA of interest in cases and controls were described using appropriate summary statistics. Conditional logistic regression comparing cases and matched controls was used to assess possible risk factors obtained from the medical record (Table 2). Each factor was assessed individually in a model. The stepwise process was used to build a multivariable model to identify factors independently important as risk factors for PsA among patients with PS. A p value < 0.05 was considered statistically significant. Possible risk factors obtained from the patient survey were analyzed using logistic regression models, with data on predictor variables incorporating both medical record and patient survey data (in contrast to the medical record analysis, which used data on predictor variables from the medical record only). Each factor was assessed univariately. A multivariate model was developed using the stepwise process and adjusting for significant medical record factors. Matched analyses of p