TEMPLATE DESIGN © 2008 www.PosterPresentations.com Introduction Seasonal Patterns More information on this and related projects can be found at www.SocialMediaLab.ca Acknowledgements Method Demographic by Clinic Type Conrad Ng ([email protected]), Anatoliy Gruzd ([email protected]) – School of Information Management, Dalhousie University Calvino Cheng, Bryan Crocker, Don Doiron, Kent Stevens – Capital District Health Authority, Halifax, Nova Scotia, Canada Clinic to Clinic Network Conclusions Physician to Clinic Network This project is funded by MITACS and CDHA. We also thank the CDHA Pathology Informatics Group for assisting in the data extraction and verification process. This research uses data visualization techniques and social network analysis to determine the status and efficiency of laboratory ordering for the outpatient system in Nova Scotia, Canada. Currently, the Capital District Health Authority (CDHA) model demonstrates that approximately 60% of laboratory ordering originates in the outpatient setting and is costing the province approximately $3.3 million per month. The goal of this pilot project is to turn the vast amount of data in the CDHA’s laboratory information system into usable information and allow the CDHA to identify usage trends to better understand the future demands on lab testing and allow policymakers more insight into the Nova Scotia primary care landscape. 1. Extracted anonymized, outpatient lab test orders from CDHA’s Laboratory Information Systems for the period from May 2009 to May 2011 2. Re-indexed and cleaned records (e.g. assign unique identifiers and work addresses to physicians and clinics) 3. Descriptive analysis & visualization with Microsoft Excel 2010 4. Network analysis & visualization with ORA 2.3.2 (developed by CASOS at Carnegie Mellon University) based on the 3 networks: Clinic to Clinic (C2C), Physician to Clinic (P2C), Physician to Physician (P2P) Dataset Summary # of Records 925,680 # of Clinics 196 # of Physicians 426 # of Patients 278,689 0 2000 4000 6000 8000 10000 12000 May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Average # of Referrals Average Weekly Referrals May 2009 - April 2010 May 2010 - April 2011 This chart confirms seasonal patterns based on holidays and long weekends. There are consistently less tests ordered during major holidays (see the “valleys” in the chart), often followed by a spike of these orders. Connection = physician’s affiliation with a clinic(s) Node Size = # of patients Most physicians who work at the Family Focus and Walk-in clinic groups also work at other clinics. The nodes (dots) are clinics; the size of the nodes represents the total number of unique referrals from that clinic. Two nodes (clinics) are connected if they share 50 or more patients (“strong” connections). While the Family Focus and Walk-in clinics only account for about 10% of all lab testing referrals, they appear to be relatively “central” in this network. This network visualization can be used to identify “well connected” clinics, ideal for disseminating new information to physicians and patients. Even relatively simple visualizations can offer useful insights to managers and other health professionals while helping them build a predictive model of laboratory utilization. The network visualizations uncovered hidden connections between clinics and provided some additional insights into the migration practices of patients among clinics. These visualizations can also be applied to make more effective health spending and planning decisions in other similar healthcare systems. Walk-in, Family Focus, and Specialist type clinics are more likely to refer younger patients (18-30 years of age) to the outpatient laboratory testing facilities, while General-type clinics are more likely to refer older patients (48-66 years of age). 0 0.02 0.04 0.06 0.08 0.1 0.12 Density Patients' Age Group Network Density of Clinic-to-Clinic Networks for Different Age Groups Density = # of actual connections in the network divided by the number of possible connections. The densest networks corresponded to the age group between ~20 and 35. This suggests that young adults are less likely to stay with the same clinic. Funded by: