Visualizing Data to Inform Decision Making, Strengthen Advocacy, and Assess Performance Mason Forando Data Informatics, HANYS

Transcription

1 Visualizing Data to Inform Decision Making, Strengthen Advocacy, and Assess Performance Mason Forando Data Informatics, HANYS HANYS 2016 Healthcare Communications Conference Albany, NY Nov. 2, 2016 Photo Credit: content/uploads/lightbulb idea.jpg

2 Agenda Data Visualization Advantages Data Visualization Disadvantages Consumer Data Visualization Media Data Visualization HANYS Use of Data Visualization Building Data Visualization Capabilities

3 Computer Programming Math Art

4 Sight Touch 1250 MB/second* 125 MB/second* Hearing/Smell 12.5 MB/second* Same bandwidth as a computer network USB Drive Source: Hard Drive Taste *Figures published in 2010

5 Similarity Closure Continuation Proximity Figure & Ground

6 Data Visualization Advantages Response Time Message Retainment Pattern Recognition Information Context Simplify the Message Show the Big Picture Consumer Engagement

7 Response Time Rapidly communicate the structure of data set: Relationship between data points Comparison Groups and Benchmarks Distribution and Outliers Patterns and Trends Enhanced through visual cues and text: Color Size Shape Annotations Data to Pixel (Ink) Ratio: Spreadsheet/Table vs. Data Visualization Text vs. Data Visualization

8 Tabular Reporting Useful for quick mash-ups of data Requires more time for interpretation/understanding Tough to discover new questions & patterns

9 Impact of DSH Policy Alternatives by State ACA Formula Change Shift from Proxy to S 10 Shift from Proxy to S 10 (Incl. Medicaid Losses & GME Costs) This data visualization shows: Consistency of ACA s reduction across states Redistributive Effect of Policy Effect of Potential Policy to Include Medicaid Losses and GME Need to retain the proxy until S-10 is ready and transition over time *Table shows change in Medicare DSH payments as a percent of total Medicare Inpatient payments

10 Message Retainment: Texas DSH Policy Impact ACA Formula Change Shift from Proxy to S 10 Shift from Proxy to S 10 (Incl. Medicaid Losses & GME Costs) This data visualization shows: Consistency of ACA s reduction across states Redistributive Effect of Policy Effect of Potential Policy to Include Medicaid Losses and GME Need to retain the proxy until S-10 is ready and transition over time *Table shows change in Medicare DSH payments as a percent of total Medicare Inpatient payments

11 Information Context Put data points in perspective against: Regional, Statewide, or National Benchmarks Similar Hospitals Previous Time Periods National Benchmarks Critical Values CMS Hospital Star Preview Reports: Includes summary score and star rating Lacked Statewide comparison No information on Star Rating cutoff points Statistical Significance from other Star Ratings

12 Pattern Recognition Data Set I Data Set II Anscombes Quartet Data Set III Data Set IV Four different data sets Same Mean, Variance, and Correlation Summary statistics do not tell the complete story Source:

13 Pattern Recognition Financial Strength Index Days Cash On Hand Surgery Percent Commercial Percent Medicaid Percent Occupancy Rate Hospital Size Good Lower % Largest Hospitals Bad Higher % Smallest Hospitals Upstate Distressed Downstate Distressed

14 Pattern Recognition Upstate Distressed Downstate Distressed Financial Strength Index Days Cash On Hand Surgery Percent Commercial Percent Medicaid Percent Occupancy Rate Size Financial Strength Index Days Cash On Hand Surgery Percent Commercial Percent Medicaid Percent Occupancy Rate Size Lower Occupancy Rate Small & Mid-Sized Hospitals Less Than Average Surgery % Unfavorable Payor Mix Higher Occupancy Rate Includes Larger Hospitals

15 Simplify the Message The CMS Star Ratings Oversimplify 64 Quality Measures and 7 Domains of Performance into a single metric.

16 Show the Big Picture Flow of funds by goal reported in Percentages (of Provider Category Spend) Difficult to compare project investment between Provider Categories

17 Show the Big Picture Size of Investment DSRIP Takeaways: R Goals are separated into Domains/Subdomains Investment varies between Goals and Provider Category Funding strategy varies significantly between PPS s Data visualization can provide context and communicate the larger picture

18 Consumer Engagement Too much information, Not enough time Limited Attention Span Grab attention with visual Guided Learning Consumer Expectations Provide Information Rapidly

19 Data Visualization Disadvantages Data Precision Misinterpretation Encoding Errors Limitations on Media 1 1,800, ,100, ,900, ,800, ,900, ,800, ,100, ,900, ,800, ,900,000 Expensive/Time- Consuming to Develop

20 Consumer Data Visualization

21 Data Visualization in the Media New York Times Mike Bostock D3 (Data Driven Documents) Census.gov Albany Tableau Users Group (ANYTUG) Arcadia Solutions

22 Kaiser Health

23 Kaiser Health (Hospital Stars)

24 Kaiser Health (cont.)

25 Kaiser Health (cont.)

26 HANYS Use of Data Visualization New Meeting Format Internal Databook created to make key datasets consumable Databook Mission: Self-Service Consumable Collaborative Aligned with HANYS Agenda

27 HANYS Use of Data Visualization Data Visualization For Advocacy: Disproportionate Share Payment Advocacy Vulnerable Hospitals Hospital Stars Staffing Ratios Managed Care Member Facing Analytics: Hospital stars Potential Future Analyses: - Hospital Financial Analysis - Managed Care Plan Analysis - DSRIP Reporting -TBD

28 Data Visualization Tools/Software Time & Cost Involved Visualization Flexibility Choose data visualization software based on current expertise, cost considerations, and desired flexibility in reporting

29 Excel 2016/Power View Cost-Effective Minimal Training/Support Required Mapping Capabilities Excel Web Apps New Visualizations (Excel 2016/add-ins) help/6 new awesome chart types in excel 2016.html

30 Plot.ly More Cost-Effective than Tableau Desktop/Server ($0-$33/Month) Cloud Based similar to Tableau Public/Server Integrated with R/D3/Python Embeddable on Websites More Flexibility than Excel Data Security Concerns

31 Tableau Reader/Desktop Can be expensive depending on distribution model Tableau Reader (Free) Tableau Desktop Tableau Server Create modern visualization within weeks of training Tableau Support Community Default Visualizations look good without much work Can be publicly posted on Tableau Public PHI Considerations Competitors: Qlikview Spotfire Domo Plot.ly

32 JavaScript based Programming Language Developed by Mike Bostock (former New York Times employee) Highly flexible for visualizing data Requires sophisticated developers & lengthy development time Embeddable on Webpages Client side data limitation

33

34 Growing Your Data Visualization Capabilities Leverage New Technology/Tools Excel 2016/Powerview ~ Tableau-Lite Tableau: VizQL ~ SQL Raw.DensityDesign ~ D3 Qlikview, Spotfire, Plot.ly, Domo Always visualize your data Build data visualization vocabulary Parallel Coordinates Sunburst Diagram Bubble Chart Pareto Chart Treemap/Tilechart Sankey Diagram Learn from others

35 Questions