The Bridging Vein to Vein with Big Data session included the following presentations:
1. Shuoyan Ning: Transfusion-related immune modulation and the use of big data analytics
2. Jingcheng Zhao: Big forward-looking data - time for Hemovigilance 2.0?
3. Angelo D'Alessandro: Chemical individuality of the blood donor as gleaned by high-throughput metabolomics of over 13,000 end of storage red blood cell samples and multi-omics analyses of 643 recalled donors
4. Abdirahaman Musa Jibrail: Bridging the Gap Geospatial Analysis to Estimate Demand and Unmet Need of Blood Products in Rural Kenya
MODERATORS: Gustaf Edgren and Katja van der Hurk
After the presentation, there was a questions and answers session, which is also included in the recording.
Transfusion-related immune modulation and the use of big data analytics
S Ning1,2, N Li3, Y Liu1, J Acker4,5, D Arnold1, C Hillis1, K Lucier1, B Rochwerg1, S Syed1, M Zeller1,2, N Heddle1
1McMaster University, Hamilton, 2Canadian Blood Services, Ancaster, 3University of Calgary, Calgary, 4University of Alberta, 5Canadian Blood Services, Edmonton, Canada
Background/aims: Transfusion-related immune modulation (TRIM) may lead to an increased risk of patient complications, none of which are currently monitored by post-transfusion surveillance systems. The impact of blood policy, collection and process changes on TRIM outcomes also remain unexplored.
Methods: All adult transfused and non-transfused hospitalized in-patients from 2002 to 2018 in Hamilton, Ontario Canada were included in this retrospective study. Data were analysed using IN-TRUST (an interactive interface to our large Transfusion Research Utilisation Surveillance and Tracking (TRUST) database); TRUST is a multihospital database with clinical, laboratory, and transfusion data. TRIM outcomes (sepsis, respiratory failure, venous thrombosis and organ dysfunction) were captured by International Statistical Classification of Diseases and Related Health Problems (ICD-10-CA) codes, Canadian Classification of Health Interventions (CCI) codes, and laboratory parameters where applicable with validation studies performed. Blood supplier provided data on changes made to blood policy, collection and processing, as well as their quality control impacts. Time series trend graphs using aggregate data were used to identify the CBS change(s) most correlated with changes in TRIM outcomes.
In the second phase of the study, we performed a traditional logistic regression analysis adjusting for key covariates to explore the association between consolidation of blood production in Brampton Ontario (consolidation) and TRIM outcomes. Admitted hospital in-patients who received 1 or more red blood cell (RBC) transfusions from Jan 2010 to Dec 2014 were included. Primary outcome was in-hospital mortality, and secondary outcomes included sepsis, respiratory failure and organ dysfunction.
Results: ICD-10 sepsis codes were validated using prospectively collected observational study from a published study (DYNAMICS), showing specificity of 94% and sensitivity of 42.2%. Respiratory failure CCI codes were validated using manual chart review with specificity of 89.4% and sensitivity of 71.6%. The blood supplier identified 10 key product policy, collection or production changes. A total of 32 time series trend graphs were generated comparing transfused with non-transfused patients—identifying consolidation as a key production change. In the second phase of the study, 9871 and 7871 patients with an index hospital admission receiving 1 or more RBC transfusion were identified pre- and post-consolidation, respectively. Multivariate analysis found no increase in in-hospital mortality when post-consolidation was compared to pre-consolidation (odds ratio [OR]1.003, 95% confidence interval [CI] 0.887–1.135, p = 0.954). Respiratory failure (OR 0.831, CI 0.650–1.062, p = 0.139) and organ dysfunction (OR 0.949, 95% CI 0.836–1.078, p = 0.421) similarly showed no harm following consolidation. There was a statistically significant reduction with sepsis following consolidation (OR 0.811, 95% CI 0.743–0.886, p < 0.001).
Conclusion: Using a hypothesis-generating analytics approach, consolidation of blood production was identified as a key change made by the blood supplier. Consolidation was not associated with changes in in-hospital mortality, respiratory failure and organ dysfunction but was associated with a reduced risk of sepsis.