ContentsDATA GOVERNANCE HANDBOOK TEAMWhy Does Data Governance Matter?. . . . . . . . . . 1SA Kushinka, MBA, CCI Program DirectorUsing the Data Governance HandbookJerry Lassa, MS, Principal Data Matt3rsAmy Jean Ham, PCMH CCECherbon VanEtten, MBI, Independent ConsultantDana Herrick, Dana Kay DesignFUNDED BYCalifornia Health Care Foundation as part of theSafety Net Analytics ProgramLaying the Foundation . . . . . . . . . . . . . . .3Purpose and GoalsLeadership Support and Executive SponsorshipData StrategyTools for Laying the FoundationAssembling the Team. . . . . . . . . . . . . . . . 7Data Governance CommitteeData StewardsData ServicesTools for Assembling the TeamPutting Governance in Motion . . . . . . . . . . .Training and CommunicationPolicies and ProceduresCenter for Care Innovations (CCI)1438 Webster St., Suite 101Oakland, CA ools for Putting Governance in Motion17

Why Does Data Governance Matter?Many years ago – even before electronic health records permeated the exam rooms of health centers – we firstheard the saying “you can’t make good decisions with bad data.” Unfortunately, good data seems to be hardto come by even as our information systems grow more sophisticated every day. Moore’s Law for data tells usthat the data available to us doubles every 18 months. Without a way to manage it all, we’ll drown in it.1And that’s what data governance is all about: managing data. Like the techniques, policies, and proceduresused to leverage any other valuable asset of your health center – such as people, capital, or facilities –right-sized data governance helps you to provide accurate, timely, trusted and complete information toexecutives and front line staff alike. Taking that idea further, Health Catalyst describes the “Triple Aim ofData Governance” that nicely sums up what data governance should do:MAXIMIZEDATA USEDATAGOVERNANCEIMPROVEDATAQUALITYSCENARIO 1. A team of “super users” at a healthcenter was responsible for building EHR templatesand setting system parameters. When one of thecare teams asked the EHR team to create more userfriendly names for the cancer screening labs, theyreadily complied. Unfortunately, the EHR team wasunaware of the need to map results coming in froman external laboratory to the lab results field in theEHR by using the same LOINC codes. The healthcenter began seeing a dramatic decrease in cervicalcancer screening rates. When investigating, thedata analyst found all labs being pulled onto thereports, but had no idea that results had stoppedinterfacing to the EHR. Better education of the EHRteam and stricter controls on critical system changes– both a function of data governance – would haveprevented this.SCENARIO 2. A clinic manager added a new visitINCREASEDATALITERACYIn the video segment Data Governance for High Functioning Health Centers, Dale Sanders notes that datagovernance is nothing new; other industries have embraced these ideas and procedures long ago. It’s new inhealth care because we’re just becoming digital and are now producing electronic data – lots and lots of data!Add to that the rapid growth and uncertain operating environment health centers find themselves in and it’seasy to understand how overwhelming data governance can seem. Like most vexing challenges, the flip sideof the coin is the opportunity to provide high value at relatively low cost or effort. See if you can recognize yourhealth center in the scenarios presented in the sidebar.type to the appointment scheduling system to helpproviders understand the reason for the patients’visit. Within a month, visit volume reports started toshow a dramatic decline in productivity. The DataServices team was asked to investigate the cause.It was found that the productivity reports wereprogrammed to run off a list of visit-types from amaster file. The new visit type had not been addedto this file. Good data governance procedures andcommunication would have prevented this situationby ensuring that the interconnectedness of dataelements and the consequences of making changesare understood by all.1. Point B issue brief How to Turn Your Data Governance Project into a Long-Term Success, 2015DATA GOVERNANCE HANDBOOK1

Using the Data Governance HandbookWhether you identify with one or many of the stories shared throughout, this handbook will help you beginto treat your data as organizational currency. The Data Governance Handbook and its companion, Buildinga Data Driven Culture ( video learning series provide practical tools and guidance forimplementing effective data governance.Outlined are critical building blocks for an effective data governance program that are organized into threephases:Data governance is a journey.Start small, produce value andgrow the data governance Laying the Foundation. Focuses on identifying the problems you are trying to solve, what you would liketo achieve, and establishing leadership support. Assembling the Team. In addition to leadership support, you will need to gather together otherstakeholders and resources. Putting Governance in Motion. Having completed the work of the first two phases, you will be ready toexecute your data governance plan. This phase provides guidance on training and communication,as well as policy and procedure development.function as your organizationIn addition to the building blocks and tools, we’ve also directed you to specific videos in the learning centerthat give more context and depth to the topics covered. Look for these featured sidebars throughout thehandbook.and information needs grow. TEMPLATES, EXAMPLES AND RESOURCES KNOWLEDGE CENTER VIDEOSAs you review the data governance building blocks, don’t be surprised to learn that you already have somepieces of data governance in place. Use this handbook to acknowledge and celebrate what you are alreadydoing and identify the work that remains. The tools, templates, and principles described in each buildingblock will help you put the right amount of structure and process in place without becoming burdensometo producers or consumers of data. The tools and samples are meant to help communicate roles, establishaccountability, highlight interdependence, and promote efficiency. Remember, data governance is a journey.Start small, produce value and grow the data governance function as your organization and informationneeds grow. Above all, let us know what works for you and what tools you have to share so this handbook canrobustly support all health centers.2 CENTER FOR CARE INNOVATIONS

Laying the FoundationLeaders need to model data-drivenPurpose and Goalsand without guidance from leaders aboutData governance is like any other project or process, it needs to have a purpose – a reason to exist; in otherwords, a problem to solve. The problems that need to be addressed by data governance are often surfacedby those closest to the problems, such as a Data Analyst or Quality Improvement Manager or an end userof the data, and a typical starting place is a desire to solve data quality problems. Other times, the need fordata governance evolves from an existing committee or project, such as an EHR implementation team or anexecutive team meeting where organizational priorities are set. Regardless of how the need for governancesurfaces, the purpose should always include a clear value proposition and keep in mind the “Triple Aim ofData Governance:” improving data quality, increasing data literacy and maximizing the use of data. This willkeep the effort focused on value and not governance for governance sake. Successful governance startssmall – don’t try to fix the data in all departments and systems at once. To get started, define how governancesupports the organization’s mission with a compelling initial purpose and SMART (Specific, Measurable,Attainable, Results-oriented, Time-bound) focusing efforts on the wrong areas. AtLeadership Support and Executive SponsorshipWhile data analysts can provide data andStrong leadership support and engaged executive sponsors are critical success factors for governance evenif the need for governance often arises from other levels in the organization. Leaders are in a unique positionto communicate the degree to which the organization views analytics as a strategic imperative and supportsa structured approach to managing data resources. They serve as role models for how to take a data-drivenapproach to decision making, support the adoption of data governance processes, and influence datapriorities to meet organizational goals. Leaders establish or endorse the Data Governance Committee thattypically includes an executive sponsor, other departmental leaders, data stewards and data analysts. Withoutleadership support and leaders who model data driven behavior, governance efforts are likely to stall.Data StrategyA data strategy is a documented plan that defines resource allocation, activities, and timeframes for addressingdata acquisition, completeness, accuracy, timeliness and use. Documenting key components of your datastrategy – which include things like data sources, data quality, data “ownership”, data privacy and security, datatimeliness, level of detail needed, type of analyses needed, data storage and retention – can ensure that datadeficiencies are remedied and that the organization has the right information to achieve its goals. The planshould be widely understood and considered a “living” document that responds to organizational priorities.behavior. The Data Services team is almostalways inundated with information requestsorganizational priorities, the data team couldone health center, a data analyst was taskedwith ensuring providers’ eligibility for theMeaningful Use (MU) program. In additionto conducting training on the MU measures,he created a monthly spreadsheet to show ifproviders were in danger of missing the MUthreshold. As providers were missing the goal,he continued to reach out to the CMO. Atthe end of the year, several providers didn’tmeet the eligibility goals, including the CMO.recommendations, it takes the organization’sleaders to act upon that data for improvementand model data-driven behavior.DATA STRATEGY TIPSTo keep things manageable, try developing adata strategy for an organizational priority thathas the attention of your senior leaders and has“boundaries.” This could include: A key quality initiative A priority objective from yourorganization’s strategic plan A pay-for-performance program An important grantDATA GOVERNANCE HANDBOOK3

Tools for Laying the FoundationTips for Getting StartedKeep these tips in mind as you start to put data governance practices into place:TEMPLATESFOR LAYING THE FOUNDATION DataStrategy Worksheet Analytics Capability Assessment Choose a specific measure set to start with and build trust in the data by examining the data input and output. Startingsmall and keeping your efforts within one department will help to keep your work manageable, allow you to fine tuneprocesses, and help your team to show value right away. Make sure you have an executive sponsor, champion or leadership support from a clinical or operational area. A big mistakehealth centers make is to assume that data governance and management is the responsibility of IT. Use the opportunity to get one or two end users really excited about data they can trust; these staff can become datastewards in the future. Although you’re starting small and focused, avoid referring to data governance or management as a “project”. Instead, lineup your next data management effort early on so that everyone knows the work will continue. Build policies and procedures as needed. Present these as communication tools rather than rules. Some staff may mistakegovernance for restricted access; in fact, it’s just the opposite.Analytics Capability AssessmentAnalyticsInstructions: CapabilityEvaluate each questionAssessmentin the first column of the assessment matrix and select a score that reflects your organization’s capability by circling acorresponding number. Total your score in each of the three domains then divide by the number of factors in each one (People 4, Process 6, Technology 3) todetermine youryouraveragescore for thatcapabilitiesdomain. To assess yourorganization’scapabilityoverall, total forthe scoresof each domainanalytics,and divide by 3.GeneralTo reconsiderusing thecharacteristics of each level are described below.AnalyticsCapabilityAssessment(ACA) and administeringit on an annualspecifically forCapabilityLevelsReactiveResponsiveProactive basis. DevelopedPredictiveNo evidenceor very limitedSomeenvironmentdepartmental evidence andEvidenceof an emergingFully integratedand alignedGeneralCharacteristicshealthcentersand ssmentcan helpevidence of capability,but not integrated or aligned,integrated approach, clinical and organizationally, leading edgedecentralized efforts to getinitial data marts, standardizedbusiness process improvements tools and skills, data servicesfocus capacity building effortsand influencethe work of the datagovernance committee,especially as itdata, access to information forreporting through IT, improvedbased on analytics, analyticsprovide robust support acrossfirst time, situationaldata capture at departmentdriving change and strategy,the health center, automatedrelates to building data el, some historical trendingculture change, integration ofanalytic results are fed back intoand analysis.measure across domainsAnalytics(clinical, financial,operations,patient experience).predictive models for value-CapabilityAssessmentdriven health care.Factor ExampleA S S E S S M E N T(See sidebar to download complete template.)1. P E O P L ECapability LevelsReactiveResponsiveProactivePredictiveSenior Leader Sponsorship: Senior Leader Sponsorship assesses the degree to which leaders in the organization sponsor healthcare analytics efforts, advocatefor a structured approach to analytics and allocate resources to it.Senior leaders have responsibility Senior leaders sponsor efforts1A. To what extent are senior Managers typically firefight data Managers/Directors areissues as they arise; senior leaders responsible for departmental data for ensuring that data is available throughout the organization toleaders involved with gproblemsasfor driving decisions and allocate ensure healthy data and analyticssupportive of data efforts,efforts, and ensure thatsuch issues.they relate to operations.resources to ensure its quality,issues and analytics in youravailability and timeliness.departmental efforts are balancedorganization?and aligned to maximize the useof data as a strategic asset.SCORE01234567891011Data Stewardship: The role of the "data steward" may be formally defined or informally recognized and is typically the “go to” person within a department orsite for all the queries/issues and usability of the data. Data stewards ensure the data is complete, accurate, and timely and that it is useful to the department orsite in measuring performance and making improvement.CCI Center for Care Innovations41 CENTER FOR CARE INNOVATIONS

Suggested Approach for Using the ACA Data Governance Committee (DGC) members should score the ACA individually, scoring from the functional perspective ofall data stakeholders they are representing (e.g., medical, behavioral health, operations, finance, IT, QI, HR). Prior to scoringthe ACA, view the video “Introduction to the Analytics Capability Assessment” at the site under theTerms & Tools menu. This will provide a brief overview of the ACA tool.KNOWLEDGE CENTER VIDEOSFOR LAYING THE FOUNDATION Guide to Using the Data StrategyWorksheet Introduction to the Analytics CapabilityAssessment Assign a data analyst or other staff to consolidate results of the ACAs from all DGC members, averaging scores from eachmember for each item. A spreadsheet is ideal for consolidating results. What Is Data Governance How to Get Started with Data Governance Right-Sizing Data Governance Barriers to Effective Data Governance Schedule a time within a Data Governance Committee meeting (or a separate time) to review the aggregated results of theACA scores from each member. Allow 1-2 hours so the DGC can gain full understanding of the results and have thoughtfuldiscussion about different perspectives in scoring and to draw conclusions that will inform DGC efforts.Suggested agenda for review of ACA results: Review the aggregated scores and discuss. Recognize factors where scores were high and discuss how to best leveragethese strengths. Discuss factors where scores were low and where there was a lot of variability across members. Gain consensus on a final score for each factor. Provide a brief comment for each that gives a qualitative flavor for whythe factor was scored at the level it was. Discuss which factors hold the greatest opportunity for impact and how to best use the resultsto guide analytic capability development efforts by the DGC.DATA GOVERNANCE HANDBOOK5

Data Strategy WorksheetData Strategy WorksheetThe worksheet shown here contains a set ofquestions that can be used to build and reviewyour data strategy to align with your organization’skey performance metrics or a family of measuresfor a specific improvement effort. Not all questionsneed to be answered for each data point ormeasure; use this as a guide to highlight potentialdata integrity and data management issues.Diabetes MeasureSet ExampleComponentTypical QuestionsDataRequirements (See sidebar on page 4What core data elements do you need to start with?Which ones will you need in the future? What are the sources of that data?Current State:to download a blankworksheet template.)Data Governance oooStart with diabetic patient diagnosis, A1c value.Will need blood pressure reading in future.Sources of data are EHR and PHM systems.oooProviders own the diagnosis field, lab owns A1c, MAs own BP.Medical leadership, providers define meanings and valid values.Medical leadership ensures standard protocol for diagnoses.oDiabetes diagnoses not being documented consistently; # diabeticpatients in PHM system does not match EHR.BP readings are out of range. Improve data input procedures.Data QualityIntegrationThe DSW can be used What validity issues are there with the required data?Availability, accuracy, consistency, timeliness?What data fixes are required?Current State:ComponentTypical QuestionsStage and Store ooShadow in clinic to observe and document actual workflow; compareit to ideal workflow and adjust as necessary. Provide training.Review current training materials and update as needed.AnalysisPrivacyAssess each componentPlan of Action:ooMeet with EHR applications team for line by line mapping betweenEHR and PHM system. Establish standard data entry workflows.Develop reports to flag data quality issuesNeed clinic total, site, care team, individual provider level detail.Yes, need senior leaders to monitor at an organization level. Needmanagers and care teams to focus on local issues to intervene. How do you get the data? Does it need to be reformatted for consistency? Does it need to feed back to other systems?Current State:All data comes from EHR and PHM reports.PHM system will map some fields to identify appropriate diabeticpatients. It can also help flag outlier data.What is your data architecture ‐ specifically where is the data held?Will you have a central repository or data warehouse?Current State:oDevelop data quality and measure reports at the provider, care team,site and clinic total levels.Plan of Action:ooReview and update mapping in PHM system to align with EHR.Update workflows in EHR to show past diabetic dx; Create otherprompts to improve dx coding and upper/lower limit checks on BP. ofamily of measures. Data collected in EHR, flows to PHM system; stored on SQL server.oYes, data is aggregated from multiple sources (EHR/PHM).oooNeed data from EHR and PHM for analysis.Need skills in report writing in EHR and PHM systems.Analysis will help identify opportunities by site and provider.ooYes, PHI and provider names are sensitive data elements.Need to ensure PHI protected and have Data Use Agreements forsharing with other organizations.What information is required to perform the analysis?What skills are required to understand the data?What actions will result from the analysis? What are the criteria for those actions?Current State: Plan of Action:oUpdate SQL server as needed to align with updates to EHR and PHMsystem.Plan of Action:ooData analyst will generate reports from EHR and PHM systems withhelp from EHR application analyst.Data analyst will prepare a visual analysis of the data.oooWhen cleaning up data quality, need to be careful about PHI.For measure reporting, be sensitive about provider labels.No sharing of reports outside org without data use agreements.Are there any sensitive data elements?What are the HIPAA compliance requirements?Will this data be shared with third‐parties and what risks does that create?Current State:Plan of Action: Reportingor measure set you are Do you have a need to report your data to others? Do you need to alter the data to properly graph/report it? Who needs access and how will they get it?Current State:oAccess oYes, internal reporting to sites and providers; external reporting toUDS annually, and health plans quarterly.For internal reports, beginning to use Tableau at all levels.Current State:Versioning andRetentionPlan of Action:oNeed to have data quality reports and measure outcome reportsavailable in Tableau for all user levels (site, care team, provider).What are the requirements to make the right data available to the right people at the right time?oData quality and measure reporting is ideally accessible at time periodsthe user can specify (e.g., daily, weekly, monthly, annual) and shouldprovide trends on any data or measure reported.Plan of Action:ooEnsure Tableau report (above) allows user to select time period.Ensure appropriate access privileges for all staff. If data is regularly updated, what data changes do you need to capture? How do you track what version you are using? How long do you keep data? When do you archive it?Current State:Plan of Action:ooCCI Community Care Innovations6Identify correct number of DM pts using EHR, ensure match to PHM.If no match, find cause (e.g., data entry, coding, workflow, mapping)and address (e.g., training, fix EHR fields & workflow, mapping) What level of detail do you need? Does the data need to be at different levels of detail for different uses?Current State:Plan of Action:ooyour data strategy forfocusing on.oWho owns the data element(s)?Who defines meanings and valid values?What is the division of responsibilities between admin, clinical, and IT?Current State:Plan of Action:ooas relevant to the measureo Granularitya single measure orPlan of Action: oto build and review‐ DIABETES MEASURE SET EXAMPLEThis worksheet contains a set of questions that can be used to build and review your data strategy to align with your organization’s key performance metrics ora family of measures for a specific improvement effort. Not all questions need to be answered for each data point or measure; use this as a guide to highlightpotential data integrity and data management issues.Any changes made in EHR or PHM system are captured in a log.All reports have a version in their title.oAssess current versioning documentation procedures to ensurecompliance. Update as needed.2 CENTER FOR CARE INNOVATIONS

Assembling the TeamData Governance CommitteeA Data Governance Committee should be a multi-disciplinary group formed to increase data quality, improvedata literacy and ensure the organization maximizes the value of the data they collect. This team should haverepresentation from all staff who use the organization’s information systems. This committee develops datarelated policies and procedures that help ensure data can be turned into actionable information for end userswhile maintaining data security and integrity. These policies and procedures provide guidance to individualswho interact with data across the organization, performing specific data-related tasks such as data validation,workflow mapping, reporting, specification, integration and analysis. Often, data governance functions arecarried out in other standing committees rather than forming a new group. However an organization choosesto carry out this important function, the group needs to manage communication about data-related policies,standards and decisions to all stakeholders, and coordinate the activities of data stewards who are dispersedthroughout the health center. A Data Governance Charter helps to define the committee’s activities and serveas a communication tool making others aware of the role and activities of the committee.DATA GOVERNANCE TIPSData StewardsData stewardship refers to the processes and attention given to ensure that usable data and information isavailable throughout the organization. Data stewards are the individuals that make this happen. They areresponsible for the accuracy, reliability and completeness of data, usually within a specific department orfunctional area (medical, dental, women’s health, billing, etc.). They also work with the data services departmentand IT staff to prioritize data and information requests. Data stewardship is a responsibility given to an existingrole, typically a department supervisor or director. Data stewards understand and communicate key data andmetric definitions and guidelines for how data are analyzed and presented. Additionally, they train staff on howand where data should be entered in the EHR or other source systems and how to interpret data and use it fordecision-making. Data stewards work together to make sure changes to system parameters in one departmentdo not adversely affect another; they play a central role in carrying out data governance processes.Health Centers need to remember that data quality takes time and data will never be “perfect;” however, ifyou don’t start sharing the data, then you can’t make strides to improve the data. It’s the responsibility of thedata stewards to proactively explain the current limitations of the data, make sure everyone understands thespecifications of important metrics (numerator and denominator) and make recommendations to improve theaccuracy, completeness and timeliness of the data.DATA GOVERNANCE HANDBOOKData Governance Committees rarely start outwith that title – especially in small to mediumsized health centers. Instead, we’ve seen datamanagement activities emanate from thefollowing existing structures: An EHR implementation committee oftenmorphs into a data governance committeeby dropping some IT staff and adding clinicaland business users. A QI team often takes on many governancefunctions as they work on high priorityinitiatives. Some organizations carve out time duringmonthly executive team meetings to to setpriorities and rules for data management.7

CASE STUDIES FOR DATA STEWARDSHIP“Why can’t I get baseline data for our P-4-P program?” A health center was incentivized by theirlocal Managed Care plan to improve colorectal cancer screening rates. The Medical Director couldsee the data in the EHR and requested baseline data on screening rates. The Data Services Director,however, said that she couldn’t provide the data. When pressed, she explained that the data inthe EHR were sent from the local GI office. Since the results were free text and not structured data,they were not reportable. This is where the role of data stewards comes into play in educating endusers and managing change. They would identify the workflows that need to be changed in order tostructure data in a reportable format. In this case, providers would need to change the way they orderthe colonoscopies, and someone would need to be responsible for “abstracting” the results into areportable field.“I know our smoking cessation counseling rates should be higher than what the reports show.”The Prenatal Program staff found that using the structured smoking cessation template was challengingand time-consuming to access from their usual templates. They decided to begin typing the informationin their progress notes instead of using the template. However, the Meaningful Use reports were set-upto look for the data in the template. By free-typing the data, the clinic was losing “credit” for smokingcessation counseling on their quality reports. Furthermore, they were unable to analyze the effect ofcessation counseling on quit rates. Data stewards – a central role in carrying out data governanceprocesses – can help end users link their data capture efforts to the reports that show quality andeffectiveness of care so that this disconnect is avoided.“This data just looks wrong.” Even when data entry choices are limited by a pick list or drop downmenu, there’s still a risk of misinterpretation that impacts data accuracy. When the manager of theEmergency Department (ED) of a public hospital system noticed an important operational metric, LeftWithout Being Seen (LWBS), seemed to contradict his experience he investigated further. He learnedsome staff checked the LWBS box only if the patient registered but never made it to the exam bay.Others checked it only if there was no record of discharge (meaning the patient might have receivedtreatment but didn’t check out properly). Still others interpreted LWBS subjectively based on theirdefinition of “treatment.” A data steward’s responsibility is to make sure all users understand thedefinition of key

Data Governance:” improving data quality, increasing data literacy and maximizing the use of data. This will keep the effort focused on value and not governance for governance sake. Successful governance starts small – don’t try to fix the data in all departments a