Outcomes and Costs - HIMSS365

1

Team-Based Decision Support in DiabetesOutcomes and Costs

Session 89, 8:30 a.m. February 13, 2019

Gary Ozanich, Ph.D. - College of Informatics, Northern Kentucky University

2

Gary Ozanich, Ph.D.

– Has no real or apparent conflicts of interest to report

Conflict of Interest

3

Acknowledgement

Research reported in this presentation was supported, in part, by the Kentucky Cabinet for Health and Family Services, Department of Medicaid Services under the Agreement titled “A Study on Poorly Controlled Diabetes Mellitus for Patients Among Medicaid Beneficiaries in Kentucky”

The content is solely the responsibility of the authors and does not necessarily represent the official views of the Cabinet for Health and Family Services, Department of Medicaid Services

4

• Learning Objectives

• Type 2 Diabetes care in Kentucky Medicaid

– Overview

– Why teams?

– How to share decision-making

– Building decision support software

– Lessons learned

• Questions

Agenda

5

• Describe how shared decision-making tools help patients and providers

recognize barriers and support solutions to adherence in diabetes treatment

Learning Objectives

• Evaluate customized regimens from a decision support tool within the

context of the current literature for clinical decision support and patient

engagement

• Explain the need for a broad interdisciplinary team and each team

member’s role in the shared decision-making processes and decision

support tools

• Appraise the unique problems in treating the diabetes mellitus population

that can be addressed through clinical and financial decision support at the

point of care

• Create strategies for avoiding clinical inertia through new treatment

algorithms

6

Project Overview

7

Bringing Stakeholders Together

Kentucky Department of Medicaid Services

– Logistics and data support

– Funding conceptualization, processing & management

St. Elizabeth Physicians/St. Elizabeth Healthcare

– Facilities, provider and staff engagement

– Generous in-kind contributions for the funding match

Northern Kentucky University, College of Informatics

– Facilities, faculty and staff engagement

– Contributions for the funding match

8

Project Funding

• Initial pilot funded internally by Saint Elizabeth Healthcare

• Based upon pilot success our stakeholder group extended the model to KY Medicaid patients

• Funding secured through the Public University Medicaid Partnership Program www.universitypartnerships.org/content/about

• Federal Medicaid rules sanction the formal participation of state universities in the administration of Medicaid

• Subject to very particular requirements, some state university Medicaid activities may be eligible for Federal Financial Participation (FFP) through matching funding

9

Chronic Condition1

Total

Spend

Member

Count

Claim

Count

Ranking

by Total

Spend

Ranking

by Claim

Count

Hypertension $ 392MM 260,419 1,382,703 1 1

Substance Use Disorder $ 285MM 221,716 830,004 2 4

Diabetes $ 284MM 117,706 1,041,257 3 2

Type 2 Diabetes in KY Medicaid

Sources: (1) 2017 Kentucky Diabetes Report, Table 22: Kentucky Medicaid Chronic Condition Summary, Medical Claims Only; Dates of Service in SFY 15

(2) 2017 Kentucky Diabetes Fact Sheet, Page 1

60% higher per-member spend on diabetes vs hypertension

Adult diagnoses rate doubled since 20002

37% of adults have pre-diabetes2

10

Type 2 Diabetes in KY Medicaid

Sources: (1) 2017 Kentucky Diabetes Report, Chart 2, page 1

23.6%

15.3%

13.4% 13.1%

9.7%

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

< $15K $15K to < $25K $25K to < $35K $35K to < $50K $50K or more

Diabetes Prevalence by Income1

11

Goals– Increase adherence through shared decision-making

– Reduce complexity and improve treatment via point-of-care

decision support tool

– Member and system medication cost transparency and

financial decision support

Measures of Impact– Member HbA1c

– System medication cost

– Member medication cost

– Claims for diabetes-related unplanned hospital treatments

A Quality Improvement Project

12

Quality Improvement Initiative– 200+ Kentucky Medicaid members

– Adult

– Currently under provider care at Saint Elizabeth Healthcare

– 8.0+ HbA1c

Offices in three Northern Kentucky counties– Grant, Campbell, Kenton counties

Project Term– July 2018 to June 2020

Project Details

13

Team-Based Approach

14

Diverse skills– Clinical investigation

– Endocrinology

– Pharmacy

– Biostatistics

– Informatics / Computer science

Faster for providers– Project already explained

– Member conversations about priorities take time

• Cost

• Daily routine

Minimize workflow disruptions

Why Teams?

15

Workflow

16

Shared Decision-Making

17

Clinical Inertia in Diabetes Treatment

Source: Clinical Inertia in Individualizing Care for Diabetes, Strain, et al. www.ncbi.nlm.nih.gov/pmc/articles/PMC4269638/

Failure to establish appropriate targets and

escalate treatment to achieve treatment goals.

Failure to establish appropriate targets and

escalate treatment to achieve treatment goals.

Recent studies show that clinical inertia may result up to 80

percent of heart attack and strokes related to

management of chronic conditions like

hypertension, diabetes, and lipid

disorders.

Recent studies show that clinical inertia may result up to 80

percent of heart attack and strokes related to

management of chronic conditions like

hypertension, diabetes, and lipid

disorders.

18

Factors Affecting Clinical Inertia

Source: Nonadherence, Clinical Inertia, or Therapeutic Inertia? Allen, et al, 2009. www.jmcp.org/doi/pdf/10.18553/jmcp.2009.15.8.690

19

Decision-Making Today

Source: “Giving Voice - Developing a medication decision aid for patients with type 2 diabetes”, Mayo Clinic Center for Innovation

Patient and clinician begin consultation

Patient and clinician discuss medication

Patient leaves with a prescription

Patient makes decision about medication

20

Implementing Shared Decisions

Ask patient about things that affect adherence

Cost

– Monthly budget for medication

– Cost sensitivity (“What if you had to spend entire amount?”)

Adherence

– Intentional and unintentional non-adherence

– Work-related hypoglycemia risk

Review regimen pros and cons

Prompt patient to express values and preferences

21

Decision Support Tool

22

Complexity is a Problem

Regimen Selection– 60 treatments for Type 2 Diabetes

– 1 to 5 typically prescribed

– ~6 million possible regimens

Insurance coverage and prices– 6 plans for KY Medicaid

– Unique formularies

– Different member, system costs

This is too much for providers

5,461,635

487,635

34,220

1,770

60

Unique Regimens

Possible

1 to 5 Treatments

23

How Complexity Is Handled Now

Simplify by omission– Only consider 5 to 7 medicines

Pros– Familiar with typical behavior on patients

– Fast decisions

Cons– Patient data may eliminate options

• Comorbidities

• Preferences e.g., no injections

– Pleiotropic benefits considered?

– Fit with patient budget and insurance?

– Clinical inertia

24

How Complexity Is Handled Now

Use professional treatment algorithms– American Association of Clinical Endocrinologists (AACE)

– American Diabetes Association (ADA)

Pros– Consistent logic

– Considers all medication classes

– Addresses many concerns• Avoid hypoglycemia, encourage weight loss

Cons– Many pages for an office setting

– Classes, not medicines

– Hard to address cost, patient values and preferences

– Clinical inertia

25

26

Decision Support Tool Goals

Evaluate all 6 million regimens

Standard data and rules

Web- and cloud-based for access anywhere + scale

Known (or directional) system cost

Directed by patient:– Formulary and out-of-pocket cost

– Budget and lifestyle goals

– Discuss options

– Use patient feedback in new recommendations

27

Decision Support - Scoring SystemAll 6 million regimens get composite 0 to 100 score

28

Decision Support Inputs

* For display purposes, this example shows only a subset of available decision support inputs

EMR Data

Hasn’t / Won’t Work

Insurance Coverage

Patient Budget

29

Decision Support Outputs

* For display purposes, this example shows only a subset of available decision support outputs

Patient Out-of-Pocket Cost

Regimen

Estimated A1c

Weight Change

Side Effects & Reminders

30

Shared Decision-Makingwith Support Tool

31

Typical Office Visit

32

Example

33

Lessons Learned

34

Pilot Results

35

Lessons Learned – Team Approach

Software recommendations challenge providers to:– Think outside their traditional prescribing habits

– Explore full spectrum of medications available

Recruit providers willing to:– Look critically at their own prescribing patterns

– Trade a measure of time for potentially better outcomes

36

Lessons Learned – Shared Decisions

Discussions take time– Prepare before office visit

– Read body language in office

Some conversations need prompting

– “Let’s see what happens if…” is a good ice-breaker

Incremental change is better than no change

37

Lessons Learned – Decision Tool

Usability and UI Design– Workflow, workflow, workflow

– Be easy for providers to navigate

• Minimize clicks, mouse use, data entry

– Analysis must be fast (results in seconds)

– Be clear about what prompts mean

• “Already Taking” means taking not prescribed

38

Next Steps

EHR integration

Expand geography

Telehealth

Population health & drug price modeling

Predictive analytics and machine learning

39

Gary Ozanich, Ph.D.e-mail: [email protected]

Please complete online session evaluation

Questions