P669 Validation of prior infliximab exposure and total colectomy outcome in patients with ulcerative colitis in Sweden

Abstract

Abstract Background Clinical procedures and treatments may be recorded in Swedish national registers with different quality levels. The study objective was to assess the prevalence of any prior exposure to infliximab (IFX) treatment and validate register data of prior IFX exposure and first occurrence of total colectomy (TC) in ulcerative colitis (UC) patients using electronic medical records (EMR) as the reference. Methods Prior IFX exposure (from Swedish Prescribed Drug Register (SPDR) and National Patient Register (NPR)) and first occurrence of TC (from NPR) were validated against EMR data from 6 hospitals in 3 Swedish counties (Stockholm, Uppsala and Dalarna). UC patients (UC cohort) were identified from EMR between 19 Sep 2013 and 31 Dec 2018, from which 3 drug initiation sub-cohorts were formed: 1. Golimumab (GLM) as recorded in SPDR; 2. Adalimumab (ADA) as recorded in SPDR; 3. GLM, IFX, ADA or thiopurines as recorded in SPDR or IFX recorded in EMR or NPR (Super cohort). For prior IFX exposure, the look-back period was any time before cohort entry (at drug initiation) for as long as historical data were available. For the outcome of TC, the follow-up period was from cohort entry until 31 Dec 2018. EMR data were linked to SPDR/NPR data using unique person identification numbers. Validity was assessed based on 4 parameters: sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). In addition, prevalence of prior IFX exposure was assessed. Results Mean age of UC cohort was 47.5 years and males comprised 51.6%. Using EMR data alone; and combined data sources (EMR and SPDR/NPR), prevalence of prior IFX exposure was higher in the GLM cohort (33.0%; 38.6%) vs. ADA cohort (26.7%; 31.4%), respectively. For validation of prior IFX exposure in SPDR/NPR, sensitivity was low in both GLM (3.4%) and ADA (5.9%) cohorts, as was PPV (GLM: 16.7%; ADA: 25.0%). The specificity (GLM: 91.5%; ADA: 93.6%) and NPV (GLM: 65.9%; ADA: 73.2%) were relatively high. Most validation parameters for TC were very high in UC and Super cohorts (UC: sensitivity 100.0%, specificity 99.6%, NPV 100.0%; Super cohort: sensitivity 100.0%, specificity 98.8%, NPV 100.0%), with slightly lower PPV (UC: 80.6 %; Super cohort: 75.6 %). Conclusion Prevalence of prior IFX exposure was higher in UC patients initiating GLM vs. initiating ADA therapies. As expected, SPDR/NPR had an incomplete capture of prior IFX exposure, since IFX was intravenously administered in hospitals and not registered in SPDR/NPR. EMRs were most valid data sources for hospital administrated drugs. In contrast, all validation parameters for TC were high, indicating that the TC outcome identified from the NPR was highly valid.