P427 Development of an index that predicts escalation of therapy at an outpatient appointment in patients with known ulcerative colitis (UC)

Abstract

Background: Conventional follow-up for UC places demands on health services. Demand might be better managed by targeting appointments to those patients who need therapeutic decisions. The aim was to develop a model that would predict the likelihood of escalation of therapy at an appointment. Method(s): TrueColours UC is a comprehensive real-time web-based programme for patients with UC. It monitors multiple parameters via electronic questionnaires. The TrueColours UC pilot collected data in real time for 6 months. Appointments and treatment decisions continued independently of the study and were documented in the patient's hospital record. Escalation of therapy was defined as any increase in therapy. Each appointment (date and need for escalation of therapy) was cross-matched with the corresponding TrueColours UC electronic record. All variables collected by TrueColours UC (simple clinical colitis activity index (SCCAI), IBD Control-8 quality of life, faecal calprotectin, haemoglobin, white cell count, platelets, C-reactive protein, albumin, transferrin saturation and ferritin) were retrieved. Logistic regression and backwards elimination were used to create a model. Performance and internal validation were assessed. An escalation of therapy calculator was developed. Result(s): Sixty-six patients had 208 appointments, of which 62 resulted in escalation of therapy. All 10 predictors were included in the initial model. Four significant predictors were identified: SCCAI, IBD Control-8, faecal calprotectin and platelets. Because blood results are rarely available before an appointment, platelets were excluded. Points were assigned to levels of SCCAI, IBD Control-8 and faecal calprotectin (Figure 1) to construct a model that predicted 99% probability of therapy escalation if the total point score was >100, with a calibration intercept of 0.01 (95% CI-0.47 to 0.48), slope 1.09 (95% CI 0.78-1.40), and apparent c-statistic of 0.95 (95% CI 0.91-0.97) (Figure 2). Conclusion(s): Patient reported outcomes collected in real-time through TrueColours UC can be incorporated into a prediction score that identifies patients with UC who are most likely to receive escalation of therapy at an outpatient appointment. If externally validated, this may help manage demand for outpatient appointments.