Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Copyright © 2016 Health Catalyst 1
INTRODUCTION: WHAT IS POPULATION HEALTH MANAGEMENT?
Articles on population health management—and population health analytics—are showing up everywhere. It seems there is a lot of curiosity and concern about implementing a population health management strategy and getting solid population health analytics in place. But what do those two terms really mean?
A quick review of the literature shows that there are a lot of different definitions for population health management. And there are different ideas about how population health analytics can help. For example: Is population health management managing the health outcomes of a population of patients with a similar condition? Does it mean going at risk with payers for the outcomes of a population of patients (fee-for-value)? Is it using care management to improve outcomes for high-risk, high-cost patients? Or is population health management simply engaging patients and communities for better health outcomes?
Common Thread in the Definitions: Outcomes
There is a common thread running among these disparate definitions, though. And that thread is outcomes. Put simply, population health management’s objective is to provide the highest quality care (quality outcomes) with an
Executive Report
A Guide to Successful Outcomes Using Population Health Analytics
by Thomas D. Burton, Co-Founder and Executive Vice President
Copyright © 2016 Health Catalyst 2
optimal care experience for a population of patients (experience outcomes) at the lowest appropriate cost (cost outcomes).
The key question for population health management is this: How do we systematically improve outcomes for a population of patients, one patient at a time?
GETTING STARTED IN POPULATION HEALTH MANAGEMENT: THREE INGREDIENTS OF FIRE
To get started, an analogy is useful. Just as it takes three things to create fire (oxygen, fuel, and heat), it takes three things to ignite outcomes improvement in populations (these will be covered in depth later, but as a high-level summary: it takes best practices, analytics, and adoption).
So to reiterate and continue the analogy, fire and combustion take oxygen, heat, and fuel. If any one of these three elements is missing, the fire doesn’t work, and there is less efficient combustion. All three elements are needed for outcomes improvement too; not just technology; not just population health analytics.
And the Three Ingredients of Outcomes Improvement
The goal is to create a combustion-like outcomes improvement engine. This can only be accomplished by systematically asking and finding solutions for the following questions:
What should the organization and individual clinicians do to provide optimal care? What are the evidence-based guidelines? (This is best practices.)
How are the organization and individuals within the organization performing? What is the compliance rate with a given guideline? Is care being provided to patients in the most effective and safest way? (This is analytics.)
FUEL
HEAT
OXYGEN
Figure 1: Three Ingredients of Fire
Copyright © 2016 Health Catalyst 3
Finally, how does the organization transform? How are innovations diffused across the organization? How are improvements deployed at every facility and with every clinician? (This is adoption.)
In summary, there are three essential questions in this approach: What should be done? How well is it being done? And how to get better and transform? Again, if one of these is missing—if one of these questions is not being asked—the outcome improvement effort suffers, and there is no outcomes improvement engine.
Redesign for Improvement
Paul Batalden, M.D., Emeritus Professor at the Geisel School of Medicine at Dartmouth, said, “Every system is perfectly designed to get the results it gets.” So, why doesn’t an organization redesign its systems to get better results and better outcomes?
Returning briefly to the fire analogy, leaders should ask if the organization rubbing two sticks together to get a fire. If so, that is a labor-intensive, inefficient approach—if it works at all. Does the organization have a machine that systematically produces combustion like an engine? One where it can produce outcomes improvements time after time, efficiently and sustainably?
ANALYTICS
ADOPTION
BEST PRACTICES
OUTCOMESIMPROVEMENT
How do we transform?
How are we doing?What should webe doing?
Figure 2: The Three Ingredients of Improved Population Health Management
FUEL
HEAT
OXYGEN
Figure 3: A Systematic Approach to Outcomes Improvemet
Copyright © 2016 Health Catalyst 4
WHAT SHOULD BE DONE TO PROVIDE OPTIMAL CARE?: BEST PRACTICES
Best practices describe what the organization and its clinicians should be doing to provide optimal care to patients. They provide a repeatable method for integrating new medical knowledge into evidence-based guidelines. Best practices clearly articulate the knowledge that should be applied to improve outcomes and shortens the time from when new medical knowledge is discovered to when it becomes the everyday practice of frontline clinicians.
Map of the Process: Care Continuum Map
To start considering what should be happening to provide the best care, a health system needs to look carefully at the entire continuum of care using a care improvement map.
Figure 4 is a care improvement map for sepsis showing the major processes included in delivering safe care for sepsis patients. Also identified are potential problem areas (storm clouds) and the metrics that are important to understand for a particular step in the process. The care improvement map shows components across the continuum of care that will be important to understand for clinical improvement. It also shows places where a clinical team and a technical team can collaborate.
CARE PROCESS IMPROVEMENT MAP
Sepsis (Severe Sepsis and Septic Shock)
Monitoring/Maintenance
• Identify and control infection sources within 12 hours if feasible
• Reassess antibiotic regimen daily for potential deescalation (1B)
• Manage blood glucose per protocol; target upper blood glucose <180 rather than <110 mg/dL (1A)
• Provide daily VTE pharmacoprophylaxis (1B)
• Provide stress ulcer prophylaxis with H2 blocker, or proton pump inhibitor if bleeding risk (1B)
• Discuss goals and prognosis with patients and families (1B); incorporate goals into treatment and end-of-life care (1B)
• % of patients for whom infection source is discovered and control implemented
• % of patients for whom glucose control, VTE prophylaxis, etc. protocols were followed
• Days on broad-spectrum antibiotics
Supportive Therapy of Severe Sepsis: ICU• Continue goal-directed therapy from
6-hour bundle (1C)• For sepsis-induced ARDS, target tidal
volume of 6 mL/kg predicted body weight (1A); measure plateau pressures (1B); apply PEEP (1B)
• For mechanically ventilated patients, minimize continuous or intermittent sedation (1B); target specific titration endpoints; elevate head of bed 30-45 degrees (1B); follow weaning protocol (1A)
• For hemoglobin <7 g/dL, after tissue hypoperfusion has resolved, transfuse RBC to target a concentration of 7-9 g/dL in adults (1B)
• For persistent hypotension, consider low-dose steroids with standardized ICU policy within 24 hours
• % compliance with ventilator weaning protocol (if data available)
• Ventilator days• Time on vasopressors
Early Therapy for Septic Shock (6-hour Bundle)• 6-hour bundle:
- Give vasopressors for hypotension that does not respond to initial fluid resuscitation; use norepinephrine before other vasopressors (1B) - Insert central line and measure CVP and ScvO2 for persistent arterial hypotension despite volume resuscitation or initial lactate >4 mmol/L - If initial lactate was elevated, repeat lactate measure
• Monitor therapy goals: - CVP 8-12 mmHg - MAP >65 mmHg - Urine output >0.5 mL/kg/hr - ScVO2 70% or SVO2 65%
• Vasopressors: % of patients who received vasopressors within 6 hours; % of patients who receive norepinephrine before other vasopressors
• Central line: % of eligible patients with central line placed and CVP and ScvO2 measured within 6 hours
• Repeat lactate: % of eligible patients who had repeat lactate within 6 hours
• 6-hour bundle: % of eligible patients for whom all components of 6-hour bundle were met; average time to completion of 6-hour bundle
• Goal-based therapy: % of eligible patients with documented goal-based therapy results: CVP, MAP, urine output, ScVO2 or SVO2
Early Intervention (3-hour Bundle) - ED, IP, ICU• 3-hour bundle (1-hour if IP or ICU):
- Measure lactate level within first hour - Obtain 2 blood cultures prior to antibiotic administration (if antibiotics not delayed >1 hour); consider cultures from other potential infection sites (e.g., urine, CSF, sputum) - Give broad-spectrum antibiotics (1B) within 1 hour of recognition (1B-septic shock; 1C-severe sepsis without septic shock) - If hypotensive or lactate >4 mmol/L, provide fluid resuscitation with 30 mL/kg crystalloid (1B); consider albumin if large volume of crystalloids needed
• Obtain labs for organ dysfunction or inflammation: e.g., creatinine, bilirubin, INR, platelets, CRP, procalcitonin
• Target resuscitation to achieve normal lactate level
• Do timely transfer to ICU as indicated
• Lactate: % of patients with lactate measured within 1 hour of triage/recognition
• Blood culture: % of patients with blood cultures drawn before antibiotics; % of patients with cultures drawn from other sites
• Antibiotics: % of patients who received broad spectrum within 1 hour of recognition (IP, ICU) or within 3 hours of triage (ED)
• Fluid resuscitation: % of patients who receive appropriate fluid resuscitation (30 mL/kg crystalloid); % of patients with fluid resuscitation within 1 hour of recognition (IP, ICU) or 3 hours of triage (ED)
• 3-hour bundle: % of eligible patients for whom all components of 3-hour bundle were met; time from triage or recognition to completion of 3-hour bundle
Early Recognition in the ED
• As part of triage, routinely screen for sepsis; components may include: - Fever - Hypothermia - Tachycardia - Tachypnea - WBC - Altered mental status - Hypotension - Hypoxia - Immunocompromised status
• Notify MD of potential sepsis• Initiate standard sepsis order set
(bundles)
• % compliance to routine sepsis screening protocol
• Time from triage to sepsis recognition or alert
• Time from sepsis recognition to MD notification
• Time from sepsis recognition to initiation of 3-hour bundle /standard sepsis order set
• % of ED patients eventually coded for sepsis, who screened positive in the ED
Home
Potential severe sepsis
Persistent sepsis-induced
hypotension / septic shock
Early Recognition Inpatient (IP) or ICU• Standardize nursing practice and
training in inpatient units to recognize and respond to emerging symptoms
• Routinely screen in the ICU to recognize and respond to emerging symptoms
• Notify MD of potential sepsis• Initiate bundle/standard sepsis
order set
• % compliance to sepsis screening protocol
• Time from sepsis recognition to MD notification
• Time from sepsis recognition to initiation of 1-hour bundle /standard sepsis order set
Responsive to early
resuscitation or test negative
for severe sepsis (i.e., no organ
dysfunction, normal lactate,
etc.)
See the Sepsis Outcomes Improvement Packet for aims, interventions, and resources related to each of these areas of focus.
Discharge
• Provide antibiotics and other discharge medications
• Establish pain management plan • Instruct patient/family on wound
treatment, monitoring signs and symptoms
• Make follow-up appointments
• % of patients with patient/family education documented
• % of patients with follow-up appointment scheduled.
• % of patients discharged to home versus care facility
Sepsis has the highest mortality rate and cost of any condition treated in U.S. hospitals, and causes significant morbidity. It ranks high on Health Catalyst’s key process analysis of opportunty based on financial and volume metrics from a large normalized data set. This high-level Care Process Improvement Map illustrates key evidence-based best practices and associated improvement opportunities that drive metrics and data visualizations in the Sepsis application. It is not intended as an inclusive care process model or protocol, but rather to catalyze iterative care process analysis and outcomes improvement.
KEY OUTCOME METRICS
Financial• Variable cost per case
• Electronic Medical Record (EMR)• Billing: Professional and Hospital• Financial
KEY DATA SOURCES
Clinical• Hospital mortality rate• Length of stay (ICU, overall)
• Readmissions
D/C criteria
met
DEFINITIONS
• Sepsis: The presence (probable or documented) of infection, together with systemic manifestations of infection.
• Severe sepsis: Sepsis plus sepsis-induced organ dysfunction or tissue hypoperfusion.• Septic shock: Sepsis-induced hypotension persisting despite adequate fluid
resuscitation.
Notes about evidence sources:• Unless otherwise indicated, key best
practices are from the Surviving Sepsis Campaign: international guidelines for management of severe sepsis and septic shock. (Dellinger RP, Levy MM, et al. 2012. Intensive Care Med. 2013 Feb; 39(2):165-228. )
• Practices with the strongest recommendation level (Grade 1) and high to moderate quality of evidence (Class A or B), based on the grading system used in the above guidelines, are noted as such.
2: Compliance to 3-hour
bundle3: Compliance
to 6-hour bundle
Pre-Hospital/Transport
• Use an early detection and treatment protocol (EMS personnel)
Copyright © 2015 Healthcare Quality Catalyst, LLC. All Rights Reserved Sepsis Care Process Improvement Map | 3/24/2015
1: ED early recognition (screening and alert
processes)
Key time-measured processes
LEGEND
Key initial improvement/aim focus areas
Phase or site of care
• Key best practice
• Key process metrics
• Key outcome metrics
Key outcome metrics
• Ventilator days• Time on vasopressors • Discharges to home (vs. long-term care)
Key recommended knowledge assets (standardization tools)
Routine ED sepsis screening & alert protocol
Sepsis order set (include 3- and 6- hour bundles)
Care Facility
• Length of stay at care facility.
Sepsis order set (include 3- and 6- hour bundles)
Figure 4: Care Improvement Map for Sepsis and Septic Shock (click to enlarge)
Copyright © 2016 Health Catalyst 5
Identify Common Problems and Potential Improvements
Organizational leaders can use the map to identify which of those problem areas (or storm clouds) will be most important to work on first. They can look to historical data for variation as an indication of opportunity for improvement. Having a list of common problems that are often not fully effective saves time and allows the team to come to a consensus around what to work on first.
For example, with heart failure, having a list of potential process aim statements can help expedite discussion on choosing an area to focus on first. If a team considers heart failure readmissions as the primary improvement area, the team could look at three aim statements: 1. Improve medication reconciliation; 2. Improve follow-up visit scheduling; and 3. Improve follow-up phone call within 48 hours of discharge. The team will come up with specific percentage improvements and target dates based on their own data, but these starter sets can really help accelerate the discussion. As an organization writes—and achieves—aim statement after aim statement, it will see an effect on the outcome improvement goal. The specific numbers will be determined by the team, but these are sample aim statements to help the team get started.
To paraphrase W. Edwards Deming, PhD, the great process improvement consultant, “aim defines a system.” Deciding what part of the care to improve is a critical step.
Scope the Problem and Define Precise Patient Registries
Next, an organization needs to scope the problem and define the precise patient registry for a given aim statement. The process can start with a standard registry from administrative codes, but quickly will need to move beyond that. This will mean specific clinical inclusion and exclusion criteria for the sub-cohort of patients.
For example, a children’s hospital decided to focus on improving care for an asthma patient population, and it wanted to define “acute exacerbation of asthma in the inpatient setting.” It started by finding patients with the ICD9 code for asthma, 493.xx who had two steroids or beta-agonists administered within eight house of admission and found about 29,000 patients. To create a precise registry, the hospital looked beyond the ICD9 code to patient problem lists and found an additional 22,955 patients. Next, the organization pulled supplemental ICD9 codes listed on patient encounters (such as 786.07 for wheezing and two steroids or beta-agonists administered within eight house of admission) and identified 38,250 patients. Finally, it looked at medication lists for associated prescriptions such as albuterol and gathered 72,581 patients. All of these elements and many others were evaluated, which expanded the final cohort in some ways and condensed it substantially in other ways. In the end, the hospital had a precise asthma patient population registry to focus
Copyright © 2016 Health Catalyst 6
on for its improvement efforts that comprised around 8,000 patients.
Adopt Standardization Aids
Then, the health system needs to adopt standardization aids. These can include checklists, order sets, or protocols. Sometimes these aids are called knowledge assets. These will make it easy for clinicians to choose the best action.
There are three major types of standardization aids (shown in figure 6):
Utilization knowledge assets, which describes which patients should receive care. This would include triage criteria indicating if a patient should be admitted to the hospital or cared for by his primary care physician. It also includes intervention criteria such as specifying if a patient should get an invasive procedure versus physical therapy. Referral criteria is another
SupplementalICD9 (38,250)
Medications(72,581)
ProblemList
(22,955)
ICD9 493.XX(29,805) Additional
Potential Rules(101,389)
Standard Registry
Precise Patient Registry Total Count of Distinct Patients = 106,714
Figure 5: An Example of Creating a Precise Asthma Patient Registry
KnowledgeAsset Type
Question toask
Cost per caseNursing hours by unitOR minutesL&D minutesCycle timesCost per ancillary testEnvironmental services
Cost/caseCost/procedureOR minutesL&D minutesOther LOS
Admits/1000 membersIP days/1000 membersOP visits/1000 membersProcedures/1000 membersED visits/1000 membersReadmissions/1000 members
Examples Possible Measures
Who shouldget the care?Utilization
What careshould beincluded?
Order Sets
How can carebe deliveredefficiently ?
Workflow
Transfer Checklist Clinical Ops Procedure Guidelines
Discharge Checklist Standardized Follow-up Checklist
Risk Assessment Bedside Care Practice Guidelines
Patient Injury Prevention Protocol
Health Maintenance and Preventive Guidelines
Treatment and MonitoringAlgorithms
Indications for Referral
Substance Selection
Admission Order Sets
Pre-Procedure Order Sets
Post-procedure Order Sets
Diagnostic algorithms
Triage Criteria
Indications for Intervention
Clinical Supply ChainManagement
Post-acute care order setsIP (SNF, IRF)
Home health, Hospice
Supplementary Order Sets
Figure 6: The Three Types of Assets, including Questions to Ask, Examples, and Possible Measures for Each
Copyright © 2016 Health Catalyst 7
example of a utilization knowledge asset; showing whether a patient should be referred to a specialist or stay with her primary care physician for care. Typically, decisions such as these are not standardized and are often based on the treating physician’s prior experience and training, which may not always be the latest evidence or medical knowledge.
Order sets come into play after a provider decides that a patient should receive care. Order sets dictate what should be included in that care. This includes admission order sets, pre-procedure order sets, and supplementary order sets. It also includes guidelines for which medications or substances should be selected and what supply chain components will be a part of this patient’s care.
Workflow standardization aids help deliver the care in the most efficient way. These include checklists (transfer checklists, discharge checklists, etc.), risk assessments, standard work, and actions performed by providers and support staff, over and over again.
When organizations consider these three standardization aids, metrics can be used to fully understand the use of these aids in the care delivery process and how to manage each one effectively throughout the continuum of care. For example, when looking at utilization knowledge assets, a hospital can
Home (patient portal)
* To Invasive Care Processes
Clinic Care non-recurrent
Clinic Care chronic
Acute Medical IP Med-Surg
Acute Medical IP ICU
Invasive Medical
Invasive Surgical
Diagnostic work-up
Bedside care
Triage to treatment venue
Substance preparation
Invasive* subspecialist
Chronic disease
subspecialist
Screening & preventive Symptoms
CareProcessModels
ValueStreamMaps
Diagnostic algorithms
Indications for referral
Indications for intervention
Triage criteria
Preventive, Diagnostic,Triage and Clinic Care, Algorithms; Referral & Intervention Indications (scientific flow)
PopulationUtilization
Knowledge Assets
Substance selection
Substance selection
Clinical supply chain management
Procedure Treatment and
monitoring algorithms
Admission order sets Admission order sets
Supplementary order sets
Pre-procedure order sets
Post-procedure order sets
Discharge
Bedside care practice guidelines, risk assessment and patient injury prevention protocols, bedside care procedures, transfer and discharge protocols
Treatment and monitoring algorithms
Health maintenance and preventive guidelines
Standardized follow-up
Post-acute care order sets IP (SNF, IRF) Home health Hospice
Order sets and indications for selection of substances and clinical supplies (scientific-flow focus)
MD Per CaseKnowledge
Assets
Implementation of protocols based on MD orders and clinical operations-initiated activities (Lean/TPS workflow focus)
Clinical Ops PerCase Knowledge
Assets
Clinical ops procedure guidelines and patient injury prevention
Post-procedure care
Figure 7: The Anatomy of Healthcare Delivery, colors correspond with figure 6, the Three Types of Assets. Orange denotes a utilization asset, blue is an order set asset, and green indicates a workflow asset
(Click to enlarge)
Copyright © 2016 Health Catalyst 8
consider metrics such as: Admits/1000 members, IP days/1000 members, Readmissions/1000 members, etc. For order sets, a health system can use cost per case/procedure or Operating Room minutes as metrics. And for workflow metrics, a hospital can measure nursing hours by unit or cost per ancillary test.
In figure 7, the anatomy of healthcare delivery is shown, which places the different types of knowledge assets in their logical location within the continuum of care delivery. Different standardization aids are used in the clinic and the ambulatory setting, for example, than in the acute medical unit or the invasive setting.
Payment Structure Considerations
As shown in figure 8, payment structures can really impact improvement work. Sometimes doing the right thing for the patient can have a negative financial impact on the care delivery organization as a whole. For example, a hospital in a discounted fee-for-service payment arrangement that makes a change to an intervention indication, resulting in fewer patients receiving a particular intervention, can see negative impacts in its bottom line. Whereas, if that hospital were in a full capitation or a condition capitation arrangement, the bottom-line impact would be positive.
The implications, of course, are that care delivery systems should consider the type of standardization change (seen as knowledge asset types in the right hand column of figure 8) in relation to its payment arrangements and consider approaching payers with a full capitation or condition capitation model. This way, hospitals, clinics, payers, and the patient community can all benefit from improving outcomes through utilizing the system appropriately, including the right components as the care is ordered, and decreasing waste and wait times through efficiency improvements.
Workflow
Diagnostic Variation
Standing Orders
Substance Selection
Triage Criteria
Patient Safety
Indications for Referral
Indications forIntervention
= Negative Impact = Positive or Negative = Positive Impact
Treatment andMonitoring Algorithms
Knowledge AssetType
DiscountedFFS Per Diem
Per Case
CommercialCMS CommercialCMS
Bundled Per Case ConditionCapitation
FullCapitation
Figure 8: Payment Structure Considerations Based on Standardization Types; the colors in the “Knowledge Asset Type” column correspond
with the colors in figure 6, the Three Types of Assets, and figure 7, the Anatomy of Healthcare Delivery. Orange denotes a utilization asset, blue
is an order set asset, and green indicates a workflow asset.
Copyright © 2016 Health Catalyst 9
Produce Actionable Visualizations
The final component of best practices is actionable visualizations. It is just not enough for a health system to know that, on the whole, it’s managing its diabetic patients or heart failure patients in a good manner. The organization needs to know exactly what actions or steps should be taken with specific patients.
A good example of this type of actionable visualization is shown in figure 9. Here, recently discharged heart failure patients are listed by whether or not there is a follow-up visit scheduled. The list is sorted by risk of readmission. So, care managers could make phone calls to schedule follow-up visits with those patients who have the highest risk scores (the highest on this example is 81) and move down the list. Since follow-up visits are part of best practice protocol for heart failure patients, ensuring high-risk patients are prioritized for follow-ups should help decrease readmission rates.
Patient Flight Path
Another example of an actionable visualization is the Health Catalyst Patient Flight Path (shown in figure 10). With this tool, a physician could sit down with an individual patient and talk about changes or actions the patient should take in his day-to-day life, whether it be trying to get his BMI to a better place or reducing LDL. The physician can
An optimized patient flight path requires layers of predictive functionality, including 1) disease specific risk scores, 2)statistical impact of best practice care recommendations, and 3) “what if” functionality to visualize impact on avoidingcomplications and lowering cost.
Figure 10: An Example of an Actionable Visualization: The Patient Flight Path for a Patient with Diabetes
Figure 9: An Example of an Actionable Visualization: Heart Failure Patients Needing Follow-up Phone Calls Prioritized by Risk
Copyright © 2016 Health Catalyst 10
actually create a simulation based on similar patients with similar conditions and demographics and see how to move from a high-risk category down to a lower-risk category. The physician can also show the patient the impact improving (or not improving) his condition will have on his out-of-pocket costs. This is an actionable visualization that can help promote positive activities by both clinicians and patients.
HOW ARE THE ORGANIZATION AND INDIVIDUALS WITHIN THE ORGANIZATION DOING?: ANALYTICS
Analytics give the ability to measure how the organization and the individuals providing care within that organization are performing. Good analytics will consider compliance with best practices and whether care is being provided in the most effective and safest way. If an organization is unable to measure these actions, it proves really hard to actually improve.
An organization also needs to be able to find patterns in the data to show correlation and causation. This means integrating clinical data, financial data, and patient experience data. Eventually, clinicians should be able to use this integrated data to predict outcomes and prescribe actions that would be most appropriate for an individual patient.
The biggest issue organizational leaders must address is the amount of time it takes to get information. Strong analytics will limit the amount of time data analysts and data architects spend hunting down and compiling data. The majority of their time should be spent interpreting the data and answering questions.
How to Start Building Analytics: Establish an Infrastructure to Bring Data Together
The first step to building a strong analytics is establishing an infrastructure that is capable of securely bringing all the data from EMRs, billing systems, etc. together. Every hospital has multiple source systems and data is typically
Weak Analytics
Strong AnalyticsThe majority of time is spent analyzing
and interpreting data
Understanding the questionHunting for data
Interpreting data
Data distribution
Gather, compiling or running
Understanding the question
Hunting for data
Interpreting data
dataData distribution
Gather, compiling or running
Non value-add Value-add
Figure 11: A Graph Showing How Strong Analytics Can Add Value Through More Productive Data Analysts and Architects
Copyright © 2016 Health Catalyst 11
siloed in different areas. An enterprise data warehouse can combine that data into a single source of truth.
There are three data modeling approaches often seen in healthcare.
The Enterprise Data Model
Technology vendors porting products from other industries and healthcare systems who have hired data modelers from other industries often adopt the enterprise data modeling technique. The model works well when data rarely changes, which is why it has worked in other industries like manufacturing and retail. Using this approach, data is organized and defined, and once it’s in the model, applications can use the data easily.
But enterprise data modeling has real challenges and limitations in healthcare. All data must be bound early and tightly to the predefined model. Changes in requirements, such as a regulatory definition modification, require a complete redesign or remapping of the model. Plus, developing the data model in the first place is time consuming. Using this method, it is difficult to model complex healthcare concepts.
Dimensional Data Modeling
Another common approach is the dimensional data model. EMR vendors and healthcare point solutions tend to favor this approach. It is quick to get started and to add value using this model. Only data to be used for the given analysis is extracted. Many traditional visualization tools prefer this type of model.
However, it carries some flaws. As additional data is needed, many redundant data feeds must be created. Changes to the underlying source systems cause a maintenance nightmare. Additionally, comprehensive atomic-level detail is not stored in the model meaning deeper questions can’t be answered without pulling more data. Dimensional data modeling is difficult to use in the healthcare environment.
Adaptive Data Modeling
One approach that has gained real traction at many successful health systems is the adaptive data model. The organization does not need to predetermine how all the data will be used because data is bound to definitions late in the process. Data is pulled from transactional systems into source marts with very little transformation (thus, it is a quick and simple process). This model can grow incrementally and adapts easily to changing healthcare requirements. In addition, adaptive data modeling contains both atomic- and summary-level detail and can successfully support complex healthcare quality improvement initiatives.
Copyright © 2016 Health Catalyst 12
However, the model is unfamiliar to many IT professionals because it’s been most commonly used in healthcare settings and military logistics; a lot of IT professionals come from data warehousing in retail, manufacturing, or finance environments. IT professionals evaluating data models can’t review the data model in this approach because it is always changing. Using an adaptive data model requires a mindset shift.
Information Management: Reoccurring Data Tasks
In addition to specific use cases, there are common tasks that hospitals will repeatedly do with data. For example, defining patient cohorts and registries, attributing patients to providers, and understanding the severity of the illnesses and comorbidities. To create the most effective system, the organization will need to have three distinct components working with its Late-Binding Data Warehouse.
The first is data capture. Here, application administrators acquire key data elements, assure that data’s quality, and ensure that data capture is integrated into the operational workflow. The second task involves data provisioning. Data architects and analysts move data from the transactional systems into the data warehouse, then build visualizations and generate external reports. The third component is data analysis. Data analysts and subject matter experts interpret
Acquire key data elements electronicallyAssure data qualityIntegrate data capture into operational workflow
DATA CAPTURE
Interpret dataDiscover new information in the data (data mining)Evaluate data quality
DATA ANALYSIS
Move data from transactional systems into the Data WarehouseBuild visualizations for use by cliniciansGenerate external reports (e.g., CMS)
DATA PROVISIONING
= Subject Matter Expert
= Data Capture
= Data Provisioning
= Data Analysis
Application Administrators(optimization of source systems)
Data Architects(Infrastructure, visualization, analysis, reporting)
Knowledge Managers(Data quality, data stewardship
and interpretation)
Figure 12: The Information Management Cycle
Copyright © 2016 Health Catalyst 13
the data, discover new information in the data, and evaluate the data’s quality (relaying that final point to the application administrators).
It is a cyclical process, as seen in figure 12, with capture leading to provisioning, leading to analysis, which naturally circles back to capture.
What to Do with the Data
There are several ways an organization can decide what to do with the data—that is, which area to focus on for improvement and how to go about bringing that improvement.
One method involves setting a minimum standard of quality and then focusing the improvement effort on those not meeting that minimum standard. This might be called the “rank and spank” or “punish the outliers” approach. While those physicians who were below the minimum standard will rise to the challenge and improve their performance, the rest of the providers who were already above that standard will not have changed.
A better approach—and one that will produce improvement throughout the organization for all providers—will focus on reducing variation rather than just dictating a minimum standard. Organizational leaders share best practices and establish a “shared baseline” for typical cases then encourage physicians to be consistent about using those best practice guidelines when it clinically makes sense and deviating from that shared baseline when a patient requires it (yet records their reason for the deviation so that the shared baseline can be improved over time). Even the best performing physicians often improve using this approach.
Excellent outcomes Poor outcomes
# ofCases
Mean
Excellent outcomes
# ofCases
Poor outcomes
Focus onBest PracticeCare Process
Model
Option 2: Identify Best Practice “Narrow the curve and shift it to the right”
Strategy- Identify evidenced based “Shared Baseline”- Focus improvement effort on reducing variation by following the “Shared Baseline”- Often those performing the best make the greatest improvements1 box = 100 cases in a year
Current Condition- Significant Volume- Significant Variation
Figure 14: A Better Approach to Improvement: Shift the Curve and Everyone Improves
# ofCases
Option 1: “Punish the Outliers” or“Cut Off the Tail”
Strategy- Set a minimum standard of quality- Focus improvement effort on those not meeting the minimum standard
Focus onminimumstandard
metric
Excellent outcomes Poor outcomes
1 box = 100 cases in a year
# ofcases
Mean
Excellent outcomes Poor outcomes
Current Condition- Significant Volume- Significant Variation
Figure 13: A Less Effective Approach to Improvement: Punishing the Outliers
Copyright © 2016 Health Catalyst 14
Improvement Prioritization
To be successful and make a visible impact as soon as possible, an organization needs to prioritize its improvement projects. One way to do this is to consider resource consumption and potential to reduce variability. As seen in figure 15, the best place to start is on projects that will address care consuming large amounts of resources with high variability.
This method to prioritization follows the Pareto principle or 80/20 Rule, where a very small number of projects account for a majority of the opportunity for improvement.
HOW DOES THE ORGANIZATION TRANSFORM?: ADOPTION
Adoption means understanding an organization’s capacity for change, implementing excellent governance principles, standardizing improvement methodology, and providing great training. Adoption is all about how an organization transforms. This means getting an innovation or improvement deployed broadly through the health system, including every clinician providing care.
Improvement Capacity Assessment
First, the organization and its leaders need to understand the current capacity for improvement by evaluating the organization’s current capabilities, challenges, and gaps. This involves a self-assessment of an organization’s performance for best practices, analytics, and adoption. Organizational leaders answer 70+ questions in one of three ways: 1. Just Starting; 2. Mid-Journey; or 3. Mature.
For example, an assessment criterion for best practices evaluates the standard protocols for population management in the organization by asking, “What types of standardized content have you implemented to support Population Health Management?” Organizations respond one of three ways: Just Starting (We have not standardized content to support Population Health Management. Our clinicians use their best judgement based on their individual
Y-ax
is =
Inte
rnal
var
iatio
n in
reso
urce
s co
nsum
ed
Bubble size = Resources consumed Bubble color = Clinical domain
X-axis = Resources consumed
1
2
3
4
Figure 15: A Variation and Resource Consumption Chart Showing the Best Place to Start for Improvement Efforts
Copyright © 2016 Health Catalyst 15
training.); Mid-Journey (We have begun to standardize some content [e.g. CPOE to implement standardized order sets provided by our EMR vendor]. We have not yet created standard content for both workflow and clinical domains across the continuum of care.); or Mature (We have implemented standardized content to manage ambulatory and inpatient care management [e.g., ambulatory treatment algorithms, order sets, bedside care protocols] and utilization criteria [e.g., diagnostic algorithms, triage criteria, indications for referral and intervention] regardless of what unit or facility a patient enters the same workflow and care delivery content is followed and measured.)
For adoption, one of the elements the assessment considers is data governance and data quality process by asking, “Who manages the quality of data?” Again, organizations can respond one of three ways: Just Starting (No team owns data quality); Mid-Journey (Quality is managed at the report level. Individual analysts scrub data before reports are distributed); or Mature (Data governance has been established by giving clinical and business owners the role of data stewards to identify source system errors and correct problems at the source).
These are just two questions from the comprehensive assessment organizations can undertake to see if they are set-up for sustainable success—and to show the areas where they might need to do some work. Figure 16 is a visualization showing a health system’s assessment results. This system has the greatest amount of work to do in its analytics. However, it got a bulls-eye in the best practices element, standardized protocols for population health.
Governance Teams
Next, the organization needs to tackle one of the most challenging elements of the improvement process: establishing a good governance model that allows for both data governance and data stewardship, as well as advanced organizational governance and prioritization. Getting the governance structure right will accelerate the improvement with population health.
Analytics
Best Practices
AdoptionData Access Process
Registry Definition Process
Data Governance and Data Quality Process
Sustained Care ImprovementProcess
Cost allocationmethodology
Standardized protocols for Population Health
Standardized calculations and definitions
for internal reporting
Prescriptive Modeling
Data Driven Prioritization
Missing Data Element
Capture
Data Integration Infrastructure
Standardized criteria for treatment and venue
Just
Star
ting
Mid
Jour
ney
Mat
ure
(Achievement Level)
Figure 16: An Example of a Self-assessment Visualization for Best Practices, Analytics, and Adoption
Copyright © 2016 Health Catalyst 16
Getting it wrong will stall all efforts. Figure 17 shows the set up of teams that will result in success.
First, an executive leadership team must sponsor any kind of outcomes improvement effort. This group will oversee the permanent guidance teams, approve board-level outcomes goals, review progress, and remove road blocks. The leadership team will also determine and create guidance teams to tackle prioritization efforts based on the opportunities for greatest improvement, as discussed above.
Small Teams (Designs Innovation) • Meet weekly in iteration planning meeting • Build DRAFT processes, metrics, interventions • Present DRAFT work to Broader Teams OB
Innovators
Guidance Team (Prioritizes Innovations)
• Meet quarterly to prioritize allocation of technical staff • Approves improvement AIMs • Reviews progress and removes road blocks
OB Newborn GYN
W&N
W&N
Innovators
Innovators
Early Adopters
Broad Teams (Implements Innovation)
• Broad RN and MD representation across system • Meet monthly to review, adjust and approve DRAFTs • Lead rollout of new process and measurement
OB
W&N
W&N
W&N
Innovators
Early Adopters
Early Adopters
Executive Leadership Team
• Prioritizes sequence of formation of Guidance Teams • Approves Board Level Outcomes Goals • Reviews progress and removes road blocks
Figure 17: The Permanent Teams Required for Successful, Sustainable Care Improvement
The guidance teams will prioritize innovations and can be formed around a clinical support service or a clinical program. The team should have expert and clinicians that will meet quarterly to determine aim statements, guide the small teams that are drafting the changes, and review progress.
Small, integrated, working teams of physicians, nurses, technical staff, and analysts will actually design the innovation (implementation team). These teams meet weekly and share those changes with the larger, broad team, which provides guidance. These small teams can find way to drive adoption for those who will need to change their behavior.
Copyright © 2016 Health Catalyst 17
Improvement Types
There are three types of improvements that range from easier to difficult to achieve.
Opportunity identification improvements involve looking at variation and calculating how much might be saved by standardizing care. Process improvements address the root causes of improvement opportunities and take next steps to improving outcomes. Finally, outcomes improvements measure the project’s overall goals such as an improvement in health function, a reduction in mortality, or actual dollar savings.
Figure 18 follows a skiing metaphor, with green (easy runs) being the opportunity identification improvement, blue (intermediate runs) covering process improvements, and black (expert runs) denoting outcomes improvements.
Improvement Methodology
The organization needs to implement the right tools at the right time, meaning incorporating an improvement methodology that incorporates LEAN and PDSA, as well as Agile software development principles. Figure 19 shows how to systematically approach the real work of identifying the content and best practices, defining patient cohorts, choosing a goal with aim statements, selecting the metrics, and then redesigning the entire process to get closer to the goal.
Prerequisites
EDW Platform Recruit team Train team
7. Measure Progress
1. Best Practices
2. Define Cohort
3. AIM Statement
4. Select Metrics
5. Rollout Plan
6. Rollout
Work Streams
• Solicit front line planinput
• Finalize analytics dev, testing, and rolloutsupport • Finalize interventionrollout plan
• Guidance teamvalidation
Rollout
• Review initial results • Identify, approve any
modifications tointervention rollout
• Review lessonslearned
• Create next AIMstatement
• Repeat process
Results
• Finalize cohort • Identify intervention(s) • Direct observation • Solicit front line input
on AIM andintervention
• Define interventionrollout plan
• Guidance teamvalidation
Intervention
• Review visualizeddrafts of AIM cohortfindings
• Identify data qualityissues
• Direct observation • Prioritize and select
AIM #1 • Review cohort criteria
and visualizations • Guidance team
validation
AIM
• Confirm team mission, charter, roles
• Review AIM options • Gather best practices • Profile and visualize
preliminary data • Select 2-3 potential
AIMs • Guidance team
validation
Prioritize
Select Build and Refine Build and refine Build and refine
Rollout Date
Figure 19: How to Systematically Approach Improvement
OutcomesImprovement
Examples: Reduction in Mortality Rate; HardCost Savings; Time Savings (Soft Cost);Improved Health Function
Diff
icul
ty to
Ach
ieve
ProcessImprovement
Examples: Process Step: % of Patients withscheduled follow-up visit at discharge;Data Quality: % of Heart Failure Patients with Ejection Fraction captured in EMR
OpportunityIdentificationImprovement
Examples: Potential $ Savings from VariationReduction (Key Process Analysis) ; Potential$ Leakage reduction by encouragingproviders to refer patients in network
Figure 18: Three Improvement Types in Terms of Difficulty and Using a Ski Run Metaphor
Copyright © 2016 Health Catalyst 18
Traditional “Waterfall” Approach vs. Agile Approach
The agile approach to software development is critical to successfully driving improvement. The traditional way, the waterfall method, requires a lot of documentation upfront about what the IT team will build. First, the project plan, estimates, and requirements are gathered—creating documentation. Then, the possible use cases and functional specifications are hammered out–resulting in more documentation. Next, design specifications are created—causing even more documentation. Then, some coding actually starts, and finally the customer sees the product. The customer then tests the software resulting in more—documentation. Developers take the test results and may fix the code—this is all documented, of course. Finally, once all the bugs are worked out, the customer can actually use the software and realize some value from it. This waterfall process produces significant documentation about a system that doesn’t really work.
A more effective way—the agile approach seen in figure 20—lets the clinicians who will be using the system see it in development on a weekly basis. It relies on flexible tools so the developers can make adjustments and provide information while the kinks are being worked out. Each release results in value for the customer. This approach is about working software, not comprehensive documentation. Technical staff using agile principles can provide value from the very beginning and help keep momentum around improvement efforts.
Accelerated Practices Training
Finally, clinicians, technical team members, and organizational leaders will need advanced training to add skills and capabilities to accelerate outcomes improvements. Three types of training are particularly important.
The first type of training an organization should do is an immersive quality training program. This program, targeted at those who will be training others, can be taught two to three days a month for multiple months. It should cover quality improvement theory and include an actual project with a two to four person team.
Project plan/ estimation
Agile
Traditional“Waterfall”
Requirements gathering
VisionHigh level stories
Use cases/functional specs
Designspecifications
Code
Test
Fix / Integrate
Documentation
Customersees theproduct
$$
Value to thecustomer
$$
$$
$$
$$
Figure 20: Software Development: Traditional, Waterfall Approach vs. Agile Approach
Copyright © 2016 Health Catalyst 19
Next, executive training will teach the executive team how hard it is to improve outcomes—and how important it is—if they do not already know. This program should cover high-level principles and give executives knowledge about the tools needed to drive long-term success.
Finally, just-in-time training programs are 10 to 15-minute modules used with an individual team working on a specific problem. These programs should be available to clinical, technical, and operational team members.
CONCLUSION: LIGHT THE OUTCOMES IMPROVEMENT FIRE WITH A SYSTEMATIC APPROACH TO POPULATION HEALTH MANAGEMENT
To drive scalable and sustainable outcomes improvement in population health, a healthcare organization needs to address all three questions with systematic approaches: What to do (best practices)? How is the organization performing (analytics)? And, how does the organization transform (adoption)?
As shown in figure 21, without all three, an organization has no chance at real, enterprise-wide improvement. For example, an organization that only has best practices is essentially an academic repository with no real practical application of those improvements. With only adoption, the organization falls victim to the dreaded “flavor of the month” syndrome, where clinicians (who see no evidence about best practices and have no way to measure progress) quickly disengage. If it has only the analytics and best practices, the organization is really good at one-off, siloed departments of excellence (science projects), but cannot spread that excellence across the entire organization. And as a final example, an organization with only adoption and best practices is missing the ability to automatically measure and track progress. Sustainable improvement becomes impossible.
To re-visit the fire analogy, all three—fire, oxygen, and fuel—can create an improvement fire to ignite scalable, sustainable change in an organization working on improving its population health management.
ANALYTICS
ADOPTION
BEST PRACTICES
OUTCOMESIMPROVEMENT
Information System Centric“If we build it they wil
come.” Focus on reducinginformation request queue.
Ignite ChangeScalable & Sustainable
Outcomes Improvement inPopulation Health
Science Project Centric Pockets of excellence, Limitedroll-out of improvement across
all facilities
Research CentricAcademic ideas with no
practical application. Lotsof published papers.
LEAN CentricUn-sustainable Improvements.
Can’t manually measureafter 2 or 3 projects.
Organization CentricManagement “Flavor of the month”
Clinicians disengage if evidenceand measurement are both
missing
Automation Centric“Paved Cow Paths”
(Process is automated but notimproved – many EMR
deployments)
Figure 21: What Happens When Any One of the Three Ingredients Is Missing in an Improvement Effort
Copyright © 2016 Health Catalyst 20
Thomas D. Burton, Co-Founder and Senior Vice President of Product Development
Mr. Burton is a co-founder and Executive Vice President of Health Catalyst. His leadership and decades of experience in business intelligence, analytics, and process improvement have helped many care delivery systems significantly improve clinical, operational, and financial outcomes. Mr. Burton was a member of the team that led Intermountain Healthcare’s nationally recognized improvements in quality of care delivery and
reductions in cost. He has taught courses on the Toyota Production System, Agile Software Development, value-based care, and data system design at various institutes including Intermountain Healthcare’s Institute for Health Care Delivery Research and Stanford’s Clinical Effectiveness Leadership Training. He has also given presentations at the Healthcare Analytics Summit and HIMSS. Mr. Burton holds an MBA and a BS in Computer Science from BYU.
Copyright © 2016 Health Catalyst
3165 East Millrock Drive, Suite 400 Salt Lake City, Utah 84121
ph. 800-309-6800
ABOUT HEALTH CATALYST
Health Catalyst is a mission-driven data warehousing and analytics company that helps healthcare organizations of all sizes perform the clinical, financial, and operational reporting and analysis needed for population health and accountable care. Our proven enterprise data warehouse (EDW) and analytics platform helps improve quality, add efficiency and lower costs in support of more than 50 million patients for organizations ranging from the largest US health system to forward-thinking physician practices.
For more information, visit www.healthcatalyst.com, and follow us on Twitter, LinkedIn,and Facebook.