Predictive Analytics:

Transcription

1 Predictive Analytics: Real-world experiences of HIEs Transforming Themselves Mark J. Jacobs, MHA, CPHIMSS CIO, Delaware Health Information Network Becker's Hospital Review 3rd Annual Health IT + Revenue Cycle Conference

2 About DHIN All 10,000 physicians who order tests use DHIN > 2.2 million unique patient records from all 50 states > 14 million clinical results delivered annually 100% of Delaware hospitals and three from Maryland participate Nearly 100 % of clinical results from Delaware Labs and > 95% of results from Delaware commercial radiology facilities delivered to DHIN Exchange ADT data from Maryland, New Jersey, Philadelphia, West Virginia > Practices send Ambulatory CCDs to DHIN from 23 different EHRs.

3

4 Innovative Solutions Community Health Record Public Health Reporting Event Notification System (ENS) Clinical Image Viewing Medication History HISP Direct Secure Personal Health Record Analytics Clinical Gateway Services DMOST Other Business Solutions

5 Attractiveness of HIE Use by Stakeholders HIE Value Stream Decision Support Population Health Analytics - - Distribution Secure Messaging CCD Delivery Results Delivery Logical Consolidated Patient Record Community Health Record (Query) Alerting (ENS) Encounter Notification Service (ENS) Analytic Reporting Community Gateway Strength of Services used by Stakeholders

6 HIEs Vary Services and functionality Costs Liability Data

7

8

9

10 Meaningful Comparisons

11 Outcomes from Use of Analytics Promotes measure evolution from utilization of CMS quality measures (e.g., MIPS, MSSP) Increases awareness for adoption of value-based payment models Exposes need for improvement of data collection for provider performance Moves healthcare towards transparency for consumers

12

13

14

15

16 Federated - HIE Value & ROI Patient Matching MPI/RLS Public Health Reporting Query/Response Repatriation ED and EMS Image Exchange POLST Next : patient registries, Trauma, ESRD and reporting to the California Cancer Registry Total potential community annual shared savings because of HIE participation: $19,737,950+

17 ED and EMS - SAFR 808 paramedics and EMTs trained and utilizing SD City EMS to UCSD's two emergency departments-s 4,082 in the last 90 days Currently evaluating length of stay (LOS) measures Morbidity and Mortality outcomes for both Trauma and ED patients Ancillary test ordering patterns Patterns of frequent utilizers of the 911 Estimated a cost-saving of not having to prep & scan pre-hospital EMS reports and revenue reversals due to late/missing EMS documentation of nearly $230,000 annually

18 Strategic Focus Centralized HIE Longitudinal Health Record Improving the operations of health care delivery by having access to datan to deliver high quality and cost effective care to support payment reform and new payment models. Assisting stakeholders with better understanding and delivering informed care to patients. Assisting in the management of the risk on the populations stakeholders serve by focusing on the following: Data Access (& Acquisition) Data Use and Ability to take Action Maximizing Stakeholder ROI Payment Reform & Coordination

19 Focused is on filling analytics void that any one organization couldn t do on its own. Use Case : Analytics reporting in Flint Michigan region on how many people have had abnormal Lead blood levels. real-time basis as new results from the various labs are being sent to GLHC. Leverages historical lab data to in HIE to how things progressed. A proactive view to the data so care delivery can be addressed right away. Use Case : ED Super-utilizers Analytic reporting of people who have been to one of the 3 emergency departments in a community 5 times or more in the previous month. ~ 50% of people in this category are going to all 3 EDs with another 25% going to 2. 25% go back to the same ED which validates the value of having this cross-community view. spawned the conversations in the community about who should be following up health plan care manager, PCMH care manager, hospital care manager, community care manager? Today, some of all of those are reaching out to the same people. With this data we can now have a better conversation about who is on point so we can spread the resources out and not over-bother the people we are trying to serve. Big push - change the way things are done in a community through actional analytics and alerting

20 Readmission Prediction Humans Don t Predict Readmissions Well (Journal of General Internal Medicine, 2011) Tested five cohorts: Attending Physician Resident Physician Intern Physician Case Manager Bedside Nurse None were good at making readmission predictions The inability of providers to predict readmissions goes against our initial hypothesis that the providers experience and interpretation of innumerable factors would make them ideal candidates for this task.

21 ENS Value Proposition for Hospitals JHCP has seen a lower hospital readmission rate in 30 days post discharge for their self-insured patients [CUSTOMER]t are seen in 7 days by their PCP, compared to those [CUSTOMER]t are not seen in 7 days. Because PCPs are notified in a timely manner about hospitalizations and thus get the scheduling process started sooner, patients are able to be seen within the 7 day window after discharge. Additionally, in 2013, JHCP s Medicare readmission rate was 15.3%, which is lower [CUSTOMER]n the Maryland and national average. While ENS is not by any means the only factor, JHCP has had a number of successes [CUSTOMER]t can be attributed to ENS and its ability to immediately notify caregivers when, where, and why hospitalizations are occurring.

22

23

24 Readmissions for 2016 compared to CMS benchmark

25 Reduced ED visits leading to inpatient admissions

26 TOC Revenue for Employed and Affiliated Providers Ambulatory providers can capture Transitional Care Management (TCM) revenue for employed and affiliated providers. CPT code (99495 and 99496) can be billed when a patient is called within 48 of discharge and a office visit occurs within 7 to 14 days, depending on complexity. Encounter notifications are sent in real-time, or if preferred, a daily basis and include the most recent patient demographics, including contact numbers, allowing rapid follow-up scheduling. Billing TCM CPT codes represent a significant revenue opportunity but few providers bill for effectively, in large part because they don t know about a hospitalization quickly enough

27 Focus on the High Risk 1 3 High risk patients (>20%) Readmission Distribution Interventions that reduce the rate of readmission will have largest impact among high risk patients 1 3 Moderate risk (5-20%) 70-80% of readmissions 1 3 Low risk patients (0-5%) 15-25% <5%

28 SHIEC - PCDH

29 Valley Hospital Medical Center, Predict/Envision Target

30 Estimated Annual Savings for Valley Hospital Med. Ctr.

31 Estimated Annual Savings for Valley Hospital Med. Ctr.

32 Thank You