The Analytic System: Finding Patterns in the Data

© 2014 Health Catalystwww.healthcatalyst.comFollow Us on Twitter #TimeforAnalytics

© 2014 Health Catalystwww.healthcatalyst.comProprietary and ConfidentialFollow Us on Twitter #TimeforAnalytics

John L. Haughom, MDJune 2014

The Analytic System:Finding Patterns in the Data

© 2014 Health Catalystwww.healthcatalyst.comFollow Us on Twitter #TimeforAnalytics 2

Healthcare: The Way It Should BePart One – Forces Driving Transformation • Chapter One – Forces Defining and Shaping

the Current State of U.S. Healthcare • Chapter Two – Present and Future

Challenges Facing U.S. Healthcare

Part Two – Laying the Foundation for Improvement and Sustainable Change • What will it take to successfully ride the

transformational wave?

Part Three – Looking into the Future • What will it take to successfully ride the

transformational wave?

Available for FREE download at: http://www.healthcatalyst.com/ebooks/healthcare-transformation-healthcare-a-better-way/

http://www.healthcatalyst.com/ebooks/healthcare-transformation-healthcare-a-better-way/

http://www.healthcatalyst.com/ebooks/healthcare-transformation-healthcare-a-better-way/


Implementing an Effective System of Production in Healthcare

Analyticsystem

Contentsystem

Deploymentsystem

Scalableand

sustainableoutcomes


Analytic System Components


Metadata: EDW Atlas security and auditing

Common, linkable Vocabulary

FinancialSource Marts

AdministrativeSource Marts

DepartmentalSource Marts

PatientSource Marts

EMR Source Marts

HRSource Mart

Diabetes

Sepsis

Readmissions

Less transformationMore transformation

FINANCIAL SOURCES (e.g. EPSi, Peoplesoft,

Lawson)

ADMINISTRATIVE SOURCES

(e.g. API Time Tracking)

EMR SOURCE

DEPARTMENTAL SOURCES (e.g. Apollo)

PATIENT SATISFACTIONSOURCES

(e.g. NRC Picker, Press Ganey)

Human Resources(e.g. PeopleSoft)

Late-Binding™ Data Warehouse


Population Health ManagementClinical Integration hierarchy - care process families

Hyperlipidemia

Acute Myocardial Infarction

(AMI)

Percutaneous Intervention

(PCI)

Coronary Artery Bypass Graft (CABG)

Cardiac Rehab

Ischemic Heart Diseasecare process family

Home OutpatientClinic Care Inpatient SNF Home Health Hospice

CoronaryAtherosclerosis


Population Health ManagementClinical Integration hierarchy - clinical programs

Vascular Disorders care process family

Heart Rhythm Disorders care process family

Heart Failurecare process family

Ischemic Heart Diseasecare process family

Cardiovascular clinical program

Home OutpatientClinic Care Inpatient SNF Home Health Hospice


Clinical Integration hierarchyClinical programs – ordering of care

Primary Care

careprocessfamilies

e.g.,Diabetes

CV

careprocessfamilies

e.g.,Heart

Failure

W&C

careprocessfamilies

e.g., Pregnancy

GI

careprocessfamilies

e.g., Lower GIDisorders

Resp-iratory

careprocessfamilies

e.g., Obstructive Lung

Disorders

Neuro Sciences

careprocessfamilies

e.g.,Spine

Disorders

Musculo-skeletal

careprocessfamilies

e.g., Joint

Replace-ment

Surgery

careprocessfamilies

e.g.,Urologic

Disorders

GeneralMed

careprocessfamilies

e.g.,Infectious Disease

Oncology

careprocessfamilies

e.g., BreastCancer

Peds Spec

careprocessfamilies

e.g.,Peds

CV Surg

Mental Health

careprocessfamilies

e.g., Depressio

n


Linking the three systemsClinical Integration hierarchy


Top 10 care process families account for

over 40% of the opportunity

Top 32 care process families account for

80% of the opportunity

Care process families by resources consumed (high to low)

Perc

ent o

f tot

al re

sour

ces

cons

umed

Inpatient per case KPA


Frequency distribution with control limits

0.5% 0.5%

99%

2.33 std. devs. 2.33 std. devs.Num

ber o

f tim

es o

bser

ved

(Num

ber,

rate

, per

cent

age,

pro

porti

on)

Value observed

Defect DefectWithin

specifications

Upper control limit

Lower control limit

Centerline

Spread


Variation in a process is due to

Random causes (common causes)

Assignable causes (special causes)

The Causes of Variation


Condition Acceptable INR range

DVT/Pulmonary Embolus 2.0-3.0

Atrial Fibrillation 2.0-3.0

Anterior Myocardial Infarction (AMI) 2.5-3.5

Valve Replacement 2.5-3.5

A Clinical Example

Frequency distributions and control limits are common in healthcare


Statistical process control chart(How a process behaves over time)

Valu

es O

bser

ved

Time

The further a point moves off the center line the higher theprobability it is not random variation and the greater the probability you can identify an assignable cause.

Centerline

Clinical process XYZ

Lower control limit

Assignable (special cause) variation

Random (common cause) variation

Upper control limit

Title


Process improvement

Uses of control chartsQ

ualit

y

Time

Unstable process

wor

sebe

tter

Control limits

Assignable variation

suggesting an unstable process

Stable process

Process capability

Random variation suggesting a stable

process

Impr

ovem

ent


Approach to improvement

# of cases

# of Cases

Option 1: Punish the outliers

Mean

Focus on minimumstandard

metric

Excellent outcomesPoor outcomes Excellent outcomesPoor outcomes

1 box = 100 cases in a year

Current condition

• Significant volume• Significant variation

Punish the outliers


Approach to improvement

Excellent outcomesPoor outcomes

# of Cases

Current condition

• Significant volume• Significant variation

Excellent outcomes

# of Cases

Option 2: Identify best practice “Narrow the curve and shift it to the right”

Mean

Focus on best practice care

process model

Poor outcomes

1 box = 100 cases in a year

Focus on better care



# of Cases


# of Cases

Excellent outcomes

# of Cases

Poor outcomesExcellent outcomes

# of Cases

Poor outcomes

1

2

3

4Varia

bilit

y

High

Low

Resource consumptionLow High

Improvement Approach - Prioritization

18


A Demonstration

Demonstrating the power of modern analytics…

…Finding Meaningful Patterns in your data


What Does Health Catalyst Do?● Enterprise Data Warehouse

“single source of truth”● Library of data acquisition

adapters● Metadata repository● Auditing and access control● Supports a variety of

analytic applications‒ Health Catalyst‒ Client developed

Platform


What Does Health Catalyst Do?● Reports & Dashboards● Ad-hoc query● Registries● Quality measures● Population health● Data mining● Clinical improvement● Workflow analysis● Modeling and predictive

analytics

Applications

Platform


What Does Health Catalyst Do?

Installation● Configuration● Data Architecture

Improvement● Project Management● Clinical Improvement● “Lean” Process Improvement

Applications

Services

Platform


Application Families

Foundational Applications

Discovery Applications Advanced Applications

Provide deep insights into evidence-based metrics that drive improvement in quality and cost reduction through managing populations, workflows, and patient injury prevention.

Encourage broad use of the data warehouse by presenting dashboards, reports, and basic registries across clinical and departmental areas.

Allow users to discover patterns and trends within the data that inform prioritization, inspire new hypotheses, and define populations for management.


DemosDiscovery ApplicationsFoundational Applications Advanced Applications`

Population Suitese.g., Ischemic Heart Disease

Workflow / Operational Suitese.g., Acute Medical

Patient Injury Prevention Suites e.g., Infection Prevention

Patient Injury Prevention Modules

e.g., CAUTI, CLABSI, SSI

Workflow/Operational Modulese.g., ICU, MedSurg, Emergency

Population Modulese.g., CABG, Stent, AMI

Labor Management Explorer

Rev Cycle Explorer

Patient Satisfaction Explorer

General Ledger Explorer

Readmission Explorer

Population Explorer

Patient Flow Explorer

Practice Management Explorer Suite

Financial Management Explorer

CAFE—Comparative Analytics Framework and Exchange—across Healthcare Systems and National Benchmarks

EDIT—Executive Dashboard Integration Tool (Key Performance Indicator editable collage from all app categories)

Key Process Analysis (KPA)

Cohort Builder

Comorbidity Analyzer

Payment Model Analyzer

Readmission Predictor

Patient Flight Plan Predictor

ACO Explorer Suite

Metric Correlation Analyzer

Regulatory Explorer

Attribution Modeler


Demo 1: Key Process Analysis (KPA). Identify areas of greatest opportunity for quality improvement and savings

Demo 2: Population Explorer. Identify potential risk by understanding relative size of disease populations and risk profiles

Demo 3: Heart Failure. Achieving quality improvement and cost reductions by directing targeted interventions to high-risk patients

Demo 4: Community Care. Monitoring high-risk patients in primary care to prevent expensive acute treatment

Demos: How Analytics Drive Improvement & Savings


Dr. J.15 Cases$15,000 Avg. Cost Per Case

Mean Cost per Case = $10,000

$5,000 x 15 cases = $75,000 opportunity

Total Opportunity = $75,000Total Opportunity = $175,000

$4,000 x 25 cases = $100,000 opportunity

Total Opportunity = $500,000Total Opportunity = $1,200,000

Cost Per Case, Vascular Procedures

KPA: Measuring OpportunityUsing provider variation to calculate the potential financial impact of improving and standardizing care processes


Poll Questions2. Does your organization effectively engage front line

clinicians in improvement projects where they routinely analyze care processes to eliminate inappropriate variation and improve processes over time?

91 Respondentsa. 5 – Definitely – 19%

b. 4 – 22%

c. 3 – 26%

d. 2 – 25%

e. 1 – Not at all – 8%


Demo 1: Key Process Analysis (KPA). Identify areas of greatest opportunity for quality improvement and savings

Demo 2: Population Explorer. Identify potential risk by understanding relative size of disease populations and risk profiles

Demo 3: Heart Failure. Achieving quality improvement and cost reductions by directing targeted interventions to high-risk patients

Demo 4: Community Care. Monitoring high-risk patients in primary care to prevent expensive acute treatment

Demos: How Analytics Drive Shared Savings


In Summary… • A good Analytic System that unlocks your data, automates its distribution

and makes it easy to see important patterns in the data is necessary to support meaningful and sustainable improvement.

• The data model on which your EDW is based matters.

• A Clinical Integration Hierarchy can help you organize how you think about and manage health care delivery.

• Differentiating random variation from assignable or “special cause” variation is important in healthcare and in improvement.

• Good use of your data can help guide you in an effort to maximize improvement and value for the investment.


Poll Question3. Using our discussion of an Analytic System as a

guide, on a Scale of 1-5, how effective is your organization’s analytical strategy and capability?

78 Respondentsa. 5 – Very Effective – 9%b. 4 – 15%c. 3 – 35%d. 2 – 25%e. 1 – Very Limited – 15%


Thank YouUpcoming Educational OpportunitiesLate-Binding Data Warehousing: An Update on the Fastest Growing Trend in Healthcare AnalyticsDate: July 10th Presenter: Dale Sanders, Senior Vice President, Health CatalystRegister at http://healthcatalyst.com/

Healthcare Analytics SummitJoin top healthcare professionals for a high-powered analytics summit using analytics to drive an engaging experience with renowned leaders who are on the cutting edge of healthcare using data-driven methods to improve care and reduce costs.Date: September 24th-25th Location: Salt Lake City, UtahSave the Date: http://www.healthcatalyst.com/news/healthcare-analytics-summit-2014

For Information Contact:[email protected]

http://healthcatalyst.com/

http://healthcatalyst.com/

http://www.healthcatalyst.com/news/healthcare-analytics-summit-2014

http://www.healthcatalyst.com/news/healthcare-analytics-summit-2014

The Analytic System: Finding Patterns in the Data

Healthcare

Transcript of The Analytic System: Finding Patterns in the Data