Evaluation of Bleeding Self-Assessments By Patients with Immune Thrombocytopenia (ITP): An Agreement Study

Introduction. Immune thrombocytopenia (ITP) is an autoimmune condition that causes an increase in the risk of bleeding. Bleeding is a patient-important outcome; however, timely and complete assessments of bleeding are time- and labour-intensive. ITP bleeding measurements may be simplified with patient self-assessments. We designed this study to compare the agreement of bleeding assessments done by ITP patients and by trained research staff. Methods. All patients were identified from the McMaster ITP Registry, a longitudinal registry study of consecutive adult patients with thrombocytopenia followed at the McMaster University Medical Centre, a tertiary referral clinic. A modified version of the ITP Bleeding Scale was used for all bleeding assessments, which captured the patients' worst bleeding event at each of 9 anatomical sites - skin, mouth, epistaxis, gastrointestinal, genitourinary, gynecological, pulmonary, ocular or intracranial - graded from 0 (no bleeding) to 2 (severe bleeding) from the time of the last assessment (typically 6 months prior). Patients were provided with instructions on how to use the scale and asked to complete bleeding self-assessments using an online tool. Once the patients completed their self-assessments, a trained research staff member contacted the patient to repeat the bleeding assessment by telephone. Chance-corrected interrater agreement was determined using the kappa statistic for 2-way agreement (Grade 2 vs. Grade 0 or 1 bleeding) and for 3-way agreement (Grade 0 vs. Grade 1 vs. Grade 2). Chance-independent 2-way agreement was also measured using the phi statistic. The primary analysis was the 2-way kappa, since the detection of Grade 2 bleeds is clinically important. Results. We enrolled 108 consecutive patients with ITP from the McMaster ITP Registry who had duplicate bleeding assessments done. The median time between assessments was 3 days (IQR, 2-5). Median age of patients in the study was 53 years (IQR, 38-64), 64% were female. The worst bleeding events, as determined by research staff, were Grade 0 (n=22, 20.4%), Grade 1 (n=34, 31.5%) or Grade 2 (n=52, 48.1%). There was perfect agreement for bleeding assessments at all anatomical sites for 44 patients (40.7%). There were no intracranial hemorrhages and no Grade 2 genitourinary or pulmonary bleeds were reported. Chance-corrected 2-way agreement was excellent for gynecological (k=0.86, 95% CI 0.71-1.02), gastrointestinal (k=1), genitourinary (k=1), pulmonary (k=1) an