Background/Purpose: Patients with acute gout are frequently treated in emergency departments (ED) and represent a typically underserved and understudied population. A key limitation of conducting gout research in the ED is the timely ability to identify potential gout patients during an ED encounter. The objective of our study was to develop a multi-criteria electronic medical record (EMR)-based gout flare alert for acute gout flares and determine its diagnostic properties for gout flare identification among ED patients.

Methods: The alert used EMR data entered by ED nursing staff and the medical problem list. The alert was triggered by: the term ‘gout’ preceding the past medical history in the chief complaint field, the term ‘gout’ and a musculoskeletal problem in the chief complaint, or the term ‘gout’ in the problem list and a musculoskeletal chief complaint (Table 1). We validated the presence/absence of gout through manual EMR review using adjudicated expert consensus (assessed with kappa coefficient) as the gold standard. We abstracted ED cases of possible gout patients in whom gout flare symptoms were not documented but who might have had gout based on other clinical parameters (e.g., those with an ICD-10 code for gout at the ED visit (M10.X), those who had an arthrocentesis performed in the ED and had an order placed for uric acid crystal analysis in the synovial fluid, and any patient who had an order for colchicine during their ED visit) to identify possible false negative cases.

Results: From January 1 to February 1, 2020, there were 38 patients with acute gout flare identified by the components of the gout flare alert and an additional 93 were identified as potentially having an acute gout flare and were included in the analysis of the alert characteristics. The positive predictive value (PPV) of the alert was 42% (95% CI [26 – 58%]) and the negative predictive value (NPV) was 93% (95% CI [87 – 98%]). The gout flare alert had a sensitivity of 70% (95% CI [51 – 88%]) and specificity of 80% (95% CI [72 – 87%]) (Table 2). Prevalence of a true gout flare was 16/38 (42%) among patients in whom the gout flare alert fired and 7/93 (8%) among patients in whom the gout flare alert did not fire. The kappa coefficient for agreement between the consensus expert determination of acute flare and the documented ED diagnosis of flare was 0.9. A second alert validation procedure occurred between October 24, 2020 to November 24, 2020. During this period, the gout flare alert had a lower overall PPV of 24% likely due to fewer gout flare cases seen in the ED due to the COVID-19 surge in our region.

Conclusion: We developed, refined, and validated a multi-component EMR gout flare alert with reasonable predictive properties to detect patients with an acute gout flare during the ED visit. This screening tool achieves good sensitivity and NPV, which preferentially increases the number of detected cases and allows the ED team to rule out false positives with secondary data. An automated EMR alert may help identify patients with acute gout flare in real time upon presentation to the ED. These patients may be approached for study recruitment or quality improvement strategies beyond the standard of care.

« Back to ACR Convergence 2021

ACR Meeting Abstracts - https://acrabstracts.org/abstract/development-refinement-and-validation-of-an-emergency-department-gout-flare-electronic-medical-record-alert/