Pharmacoepidemiology
Assessing Reliability of an Electronic Gout Flare Assessment Protocol Applying GAFFO Criteria: The TRUST Trial Shravani Chitineni* Shravani Chitineni Daniel H. Solomon Hyon K. Choi Ana Fernandes Shreya Billa Kiara Tan Chio Yokose Michael J. Berry Misti L. Paudel
Background: The TRUST Trial is a multi-center randomized clinical trial comparing two gout management strategies: treat-to-target serum urate versus treat-to-avoid symptoms. This pilot study assessed whether electronic surveys could provide comparable accuracy to telephone surveys in reporting gout flares, while accommodating patient preference and reducing staff burden.
Objectives: To assess reliability of a two-step electronic self-reported gout flare assessment compared to a staff-conducted telephone questionnaire.
Methods: A convenience sample of TRUST trial participants completed biweekly electronic surveys via StudyTrax (ST), reporting gout flares, primary outcome, in the past two weeks (yes, maybe, or no). Upon completion, blinded staff conducted the gout flare survey at standard biweekly telephone visits. AT gout flare prevalence of 0.20, estimated 82 flare surveys and ≥ 17 flares are needed to detect Kappa lower confidence level of 0.75 at 0.15 precision. In R, agreement was assessed by extended Kappa statistic and intraclass correlation coefficient (ICC) with 95% CI. Then, GEE with family=binomial, link=logit and exchangeable correlation structure were used to model log odds of reporting a flare electronically using telephone responses as independent variable.
Results: Among 28 participants (mean age 60.5 years, SD 13; 93% male), 213 flare surveys were completed (108 electronic, 105 phone) with mean of 3.5 surveys per enrollee (Table 1). Total reported flares were 23 and 22 for electronic and phone surveys, respectively, with overall flare rate of 21%. Agreement between survey types was 96%, with extended Kappa statistic of 0.475 (95% CI: -0.16 to 0.55) and high expected agreement 91%. ICC = 0.88 (CI: 0.83–0.92) demonstrates strong concordance between survey types.
Conclusions: The pilot study shows strong agreement between electronic and telephone gout flare assessments, ICC = 0.88, supporting electronic surveys as an accurate, efficient alternative.
