Sex differences in gout characteristics: tailoring care for women and men

CORRONA creates prospective observational cohorts of arthritis patients enrolled by participating rheumatologists in both academic and private practice sites. There were 1273 patients with rheumatologist-diagnosed gout based on the 1977 diagnostic criteria [12] (the registry predated the 2015 American College of Rheumatology Gout Classification criteria) and enrolled in the CORRONA registry database between 11/1/2012 and 3/31/2014 from 34 rheumatology practices across 20 states with 76 participating rheumatologists. Approvals for data collection and analyses were obtained for academic and private practice sites from local and central institutional review boards, respectively.

Data were collected from patients and their treating rheumatologists using standard clinical research forms based on the American College of Rheumatology [13] treatment recommendations. Information collected included demographics, comorbid conditions, gout presentation, disease severity and activity, family history, body mass index (BMI), dietary intake over the past week, use of medications that can raise serum urate level (e.g., diuretics), use of medications for acute gouty inflammation (nonsteroid anti-inflammatory drugs [NSAIDs], colchicine, steroids) and urate-lowering therapy (ULT) including uricosurics, xanthine oxidase inhibitors, and recombinant uricase [pegloticase]. Documentation included physician exam findings of tophi and inflamed joints, physician and patient global assessments of disease activity, patient assessment of pain, the Health Assessment Questionnaire (HAQ) assessing physical function, and serum urate levels (SUA) from laboratory tests obtained within 10 days of the clinical encounter (this data is not mandated by the study protocol). Patients reported the number of days in the past 3 months there were unable to perform their usual activity. Health care utilization data, including gout hospitalizations within the last 3 years as well as Emergency Room and/or outpatients visits in the past 12 months for gout flares, were gathered. Additionally patients reported how many flares they managed themselves without seeing a health care professional.

We compared patients’ baseline characteristics at the time of enrollment, including demographics, gout-related comorbid conditions and medications, gout characteristics, contributing medical conditions and medications. We compared dietary factors associated with gout flares in men and women in terms of any intake over the past week of beef, pork, seafood and alcohol, as well as the number of weekly servings consumed. We also evaluated rate of use of ULT in the subset of subjects who met ACR criteria for urate lowering (e.g., 2 or more gout flares/year, presence of tophi, history of kidney stones or presence of chronic kidney disease (CKD; stage 3 or greater).

We use a cross-sectional study design, evaluating patients at the time of enrollment. Descriptive statistics were performed using t tests and chi square tests or Fisher exact tests as appropriate. As a sensitivity analysis, when examining the burden of comorbidity between the 2 sexes, we used 3 propensity score (PS) approaches to balance the differences in the ages and disease duration of the women and men with gout. We conducted these sensitivity analyses to address a potential concern that differences in comorbidity burden could have been related to sex differences in age (women were on average 10 years older than men) or disease duration (men had on average 5 more years of gout) The PS approaches allow us to compare women and men with similar demographic profiles. We used 3 approaches to assess the robustness of our findings. For approach #1, we derived a PS using age and disease duration and used PS trimming to excluded patients with a PS score in the <5 percentile and >95 percentile range. For approach #2, we included only those patients who fell inside the area of common support using age and disease duration separately and then used PS trimming. Approach #3 included only those patients with a conservative common support with PS trimming so that the resulting age and disease duration characteristics of the women and men with gout were within 3 years. All 3 approaches gave similar results to the unadjusted analyses. Given the reduction in the sample size with each of these 3 approaches, we show the unadjusted results in the body of the manuscript but provide the results of the PS approaches in Additional file 1: Table S1. To explore if there were dietary differences between men and women, we conducted adjusted logistic (yes/no dietary intake) and ordinal logistic regression models (0, 1, 2+ servings per week of dietary intake) controlling for age, BMI, duration of gout, comorbidity burden [hypertension, diabetes, renal disease, hyperlipidemia], HCTZ use, other diuretic use and current use of a urate-lowering drug. Sex based comparison for functional status was adjusted for age and presented herein. Lastly ULT use was explored in those who met criteria for treatment based on 2 or more attacks in 12 months or presence of tophi (dosages provided in Additional file 2: Table S2).