Home / Learn

What is Epic SlicerDicer, and where does it stop?

Epic SlicerDicer, also written as Slicer Dicer, is a self-service tool in Hyperspace for exploring populations through counts and trends without a report writer. It reads Caboodle through available data models. When a question needs unavailable fields, a new measure or records from other systems, the next route is usually a Clarity report or a warehouse.

A department manager checks a hunch, narrows a population and sees a change worth exploring. The follow-up asks which patients are involved, whether a different definition changes the result, or how staffing relates to it. SlicerDicer answered the first count, and each of these needs records the count didn't show.

The useful next step depends on what is missing. A field already available in the model calls for another selection. A patient list, a custom calculation or staffing records may require different access or a reporting request. Check which case you're in before you take the first count to a meeting.

SlicerDicer builds a question from a data model

The model determines which records you can explore. A patient question and an encounter question start from different units: a person can have several encounters. That distinction affects what the total means before you select a period or compare groups.

The Boston University workshop slides describe a sequence from data model to population, slices, measures, dates and visualization. You select the record type, define the population, divide it into groups, choose the measure and dates, then display the result. The slides group models into access, finance, operations and clinical subjects.

UVA's Epic Tools page gives examples including patient records, surgical cases, hospital accounts and medication orders. These subjects help explain why the model choice comes first. A question about accounts needs records and measures suited to accounts, even when the same patients appear elsewhere.

UC Davis lists totals, percentages, averages, variance and minimum or maximum values, displayed through bar charts, line charts, maps or tree maps. Its training resources cover saving and sharing sessions. Keep the selected population, measure and dates with a shared result so a colleague can understand what the chart represents. A saved session is useful only when its recipient understands the question it asks.

Clinicians, managers and researchers use it differently

A clinician reviewing a trend and a researcher estimating a study population may use the same interface under different permissions. Their intended use determines which detail they need and which local approvals apply. Access to SlicerDicer alone does not describe everything they may view or export.

Yale's guidance describes de-identified YNHHS data for assessing cohort feasibility, developing protocols and preparing research without IRB approval for those activities. Yale requires Research Basic Access and completion of the recommended training. Researchers can define populations, compare groups and generate summary counts within that arrangement.

For clinical and management work, UC Davis describes exploring trends and reaching line-level detail. It permits patient details and data exports for quality improvement with appropriate approvals and permissions. That is a specific local use, so it should not be read as permission for every researcher to export patient records.

Training also follows the local setting. The Medical College of Wisconsin course covers clinical, access and revenue subjects, including saving and sharing queries. Before comparing what colleagues can see, establish whether they have the same approved purpose, data access and training. Different access can change the detail available to each person.

SlicerDicer counts come from Caboodle's last load

The time represented by a result depends on the data loaded into its source. Vanderbilt describes SlicerDicer querying Caboodle with a one-day delay. The Boston University workshop likewise describes Caboodle as updated daily. Under those arrangements, running another query does not add records entered after the source cutoff.

Drexel's page describes real-time access to clinical data and says changes in Epic are reflected in SlicerDicer. The institutions therefore describe freshness differently. Check the source and completed load in your institution before using a result to answer a question about today's activity.

Coverage matters alongside timing. Vanderbilt says its Caboodle contains most of the data available in Clarity and includes links to external data. That does not establish whether your model contains the field or external record needed for a particular question. A recent load cannot supply a field that the model does not include.

When results disagree, compare their reporting periods, source cutoffs and population definitions before drawing a conclusion about the change. Our guide to Epic's databases explains the relationship between the operational record and the reporting stores. That relationship helps distinguish a missing record from a record still waiting for the next load.

SlicerDicer stops at counts, models and permissions

A population total does not automatically provide a route to export its underlying records or define a new measure. Vanderbilt explicitly rules out record-level extracts through SlicerDicer. The UC Davis quality-improvement permission described above shows why that restriction needs an institution's name attached to it.

The available fields, measures and groupings also set the questions a model can answer. Changing selections can explore those records further. It cannot establish a definition that requires fields absent from the model. Before making another selection, identify whether the missing part is a record, a calculation or permission to see detail.

OHSU's research guidance says requests for model additions, new metrics and template changes will not be addressed for research purposes. It also treats cohorts of 11 or fewer as potentially identifiable and requires IRB approval to view those small counts. The document notes that SlicerDicer does not impose a lower cohort limit itself.

OHSU requires IRB approval when researchers use these tools to view protected health information. It also describes prompts for research-only users who seek line-level detail, including an IRB-number prompt. These are OHSU's research rules, published in 2022. They illustrate why a visible number and permission to use that number are separate questions. Your institution's research and reporting owners determine the approved route for the proposed use.

Pick the Epic tool by the question

Choose the route that can supply the required answer and support the action that follows. A trend, a current patient list and a study extract each need different records and different ways to use them. Keeping the intended output explicit makes the choice easier to explain to the reporting team.

UC Davis's Reporting Workbench page describes reports that use current Chronicles data and generally suit shorter periods. That fits questions where staff need to act on an operational list. A historical analysis with custom fields may instead need a report writer to assemble detailed records.

An existing Caboodle subject area can already hold the required relationships. Vanderbilt offers a SQL interface to approved users. Check that route before asking someone to recreate a measure. For a question spanning staffing and clinical activity, establish which source supplies each part and how their dates and definitions will be aligned.

Your questionUseWhy
How many patients or encounters match, and what is the trend?SlicerDicerPopulation totals within an available model
Which patients need action today?Reporting WorkbenchCurrent Chronicles records, generally suited to short periods
Does an existing subject area already hold the measure?Caboodle SQL, where offeredReuse prepared reporting relationships
Do I need record-level fields or a new definition?Clarity through a report writer or data coreDetailed, identifiable reporting records refreshed nightly
Do I need Epic beside Workday, finance or research records?A warehouse built on ClarityShared definitions across the required systems

Questions past SlicerDicer usually move to Clarity

A request for a custom cohort needs someone who can connect the definition to detailed Epic records. At Vanderbilt, Clarity access requires certification or proficiency in Epic's Clinical Data Model, and researchers can request extracts through its data core. The report writer supplies the knowledge needed to turn the question into a query.

That knowledge can be concentrated among a few specialists. Requests then compete for their capacity, and researchers can wait months for IT. A recurring question creates more work when each follow-up requires another extract or another version of the calculation.

Databasin's team connects Clarity and useful Caboodle subject areas to governed tables, defines measures once and adds Workday staffing or REDCap study records. Our Epic connector was created at WashU Medicine. People can then ask questions in plain language with the source and SQL attached, or explore the measures in Data Apps.

Existing Epic reports keep serving the questions they already answer. Our Epic reporting work covers the reports and follow-ups that need shared definitions and detailed records. When the answer depends on several systems, the healthcare data warehouse brings those records into the same analysis under approved access.

Example: following one readmission question past SlicerDicer

This illustrative case follows a service line's heart failure readmission question from a monthly trend to patient follow-up, a research cohort and staffing analysis.

StepToolWhat you get
Count readmissions by monthSlicerDicerA trend from the last Caboodle load, using an available measure
List patients to call this weekReporting WorkbenchA list from current entries, with permitted actions
Build a research cohort with exclusionsClarity through a report writer or data core, with required research approvalRecord-level fields selected under the study's definition
Compare readmissions with unit staffingA warehouse joining Clarity and Workday, built by DatabasinA shared definition for examining clinical activity beside staffing

Each step changes the required answer, which changes the appropriate route. Agree the readmission period and exclusions before comparing results, and use the institution's approved route for patient detail and research records.

Five checks before you trust a SlicerDicer count

  1. Confirm the data modelCheck whether the selected model represents patients, encounters or another record type, and whether its fields cover the question.
  2. Check the last load timeCheck when the source last loaded before you read a count as today's activity. Records entered since then won't appear.
  3. Ask who may see patient detailConfirm the approved purpose and permissions for viewing or exporting records, including the research rules for small groups.
  4. Write down the definitionRecord the population, measure, dates and exclusions so colleagues can explain why the result includes each patient or encounter.
  5. Know where follow-up questions goIdentify the reporting team or data core that can handle missing fields, new definitions and approved record-level extracts.

Sources

UC Davis Health, Epic SlicerDicer, 2026 · health.ucdavis.edu/data/epic-slicer-dicer.html

UC Davis Health, Epic Reporting Workbench, 2026 · health.ucdavis.edu/data

Vanderbilt Clinical Informatics Center, Epic Data Resources, 2026 · vumc.org/vclic/epic-data-resources

Yale Biomedical Informatics & Computing, SlicerDicer, 2026 · medicine.yale.edu/ybic

OHSU, Guidelines for research use of Epic reporting tools, 2022 · ohsu.edu/sites

Boston University / RPN workshop, Demystifying TriNetX & Epic SlicerDicer for Research, 2025 · bumc.bu.edu/ohra

UVA School of Medicine, Epic Tools, 2026 · med.virginia.edu/clinical-trials-analytics

Medical College of Wisconsin, SlicerDicer Overview, 2026 · ocpe.mcw.edu/SlicerDicerOverview

Epic, Healthcare Intelligence, 2026 · epic.com/software/healthcare-intelligence

Drexel University College of Medicine, Slicer Dicer, 2026 · drexel.edu/medicine

Discuss the questions SlicerDicer can't answer

We can discuss the follow-up questions your team waits for, whether they need patient detail, another definition or records from other systems. In the demo, we show a working example and discuss what delivering it would involve.