Case Review

Case Review enables users to drill down into case-level data. Because it displays protected health information (PHI), the only area of Ursa Studio to do so, access is limited to those users with appropriate authorization.

Field Groups

In Measure Workshop, each published field can be assigned to a field group. If no field group is selected, a default is applied based on where the field is located in the measure. The Case Review table will show all the fields grouped by field group, and the field groups are ordered as per the ranking defined in Metadata Manager.

Navigating the Table

Users can sort, resize, pin, and hide columns via icons in the header row. Pinned columns are fixed to the left of the screen and do not disappear upon horizontal scrolling. Pinning and hiding can also be managed in the Choose Fields screen, which is available via the dictionary topbar icon button or the alt-C hotkey. Using either mechanism, hidden and pinned fields are remembered for future visits to the measure. The table can be reset to its original state via a menu item in the hamburger menu.

Aggregate Statistics

Case review and case board tiles can return aggregate field-level statistics. Users can click on the information icon in the case review header to calculate the aggregate statistics. Once the statistics are calculated the icon will change color and the results are displayed on hover. Statistics include the min, max, distinct count, and % null for the field. Numeric fields will also calculate sum and average. Fields with less than 500 distinct values will furthermore display the ten most common distinct values. The calculations are either run on the raw row-level measure data or on the case-level data, whichever is the more relevant, and that choice will be appropriately described in the statistics hover panel.

Drag-Reordering Columns

Users can drag-reorder columns of Case Review using a gripper in the column header. The new column ordering will be remembered upon refresh and will be honored in any boards or news feed posts. Users cannot drag into or out of the "pinned columns" column group, nor can they drag unpooled columns, although the unpooled columns will follow any reordering of its companion pooled column. If any columns are dragged into another field group, the second "field group" super-header will merge together into a single "fields" field group.

Target Met and Target Not Met

For applicable measures, successes and failures are calculated on a per-case level. Measures are considered for target-met and target-not-met if they have a target direction and an inferable target. The target is considered inferable in three cases:

• It is explicitly defined.
• The measure range is binary.
• The target direction is down, and the measure range is non-negative.

By selecting the pertinent tab of the Case Review table, users can apply a filter to see only cases of that type. 

Accordioning

Some measure types might have multiple rows of raw data that roll up into a particular case. For example, the case grain size is frequently on the patient level, but multiple events might roll up to the patient if, for example, the measure is of the form "at least one event happened within a certain amount of time following another event." For such cases, the underlying raw data can be accordioned out.

Copy/Paste

By default, users will be not allowed to copy/paste values from the Case Review table. Users can choose to disable this security control on a temporary, per-screen basis by choosing the "Enable copy/paste" menu dropdown option. This action will be logged to the PHI audit log for full traceability.

Clicking on the first column of a row in case review will highlight the row. Users can remove the highlight by clicking again on the first column. By clicking on any other cell in a highlighted row, users can opt to copy that cell's contents into their clipboard. This action will also be logged to the PHI audit log.

Observation and Case Roll-Up

The numerator and denominator columns in Case Review, and the case value filter driven by these two values, are determined by the observation-level numerator and denominator values on the measure. In Case Review, unlike in Measure Explorer, these values will not be re-rolled-up from the component level for long-form measures if there is a filter in place that cuts out some components within an observation.

For example, let's say we have a measure for the maximum pain score a week after surgery. Both the case and the observation represent the surgery, and the long-form components represent each pain score. If you added a filter to exclude any pain scores taken on a weekend, this might screen out the highest pain score, but the numerator column in Case Review will not update to reflect the second-highest score. Moreover, if you were to add a case value filter (e.g., case value = 8), then the filter would capture observations with a maximum pain score of 8, even if that score was taken on the weekend. The rationale here is that Case Review is analogous to a spreadsheet, and users would most likely expect these columns to be stable for the observation.

Even though the component-to-observation roll-up will not be recalculated for long-form measures in Case Review, the observation-to-case roll-up will in fact be recalculated for Case Review if filters are in place that cut out some observations of the case, as they are in Measure Explorer. 

For example, let's say that you have a PMPM measure. The case will be a member, and an observation will be a member-month. The target for the measure is <= $10,000. If a user were to apply a date range filter, they would want to have the successes and fallouts tab represent the patients who incurred those costs within the newly specified date range. Doing so requires the case-level measure value to be recalculated from the available observations.

Exporting

Users with sufficient authorization can export data to CSV. In the default configuration, the CSV is not downloaded to the user’s workstation, and architects or admin users can export to a storage container for retrieval. These defaults can be changed via application-wide environment variables.

The export honors the filters currently in place. For example, to export all data, a user must first clear out all existing filters. Data can be exported at either the case level (as it is rendered on the table) or at the raw-data level, which is equivalent to a fully accordioned view.

Preparing the data file for export can take some time, during which the user’s avatar turns into a spinner but the user can navigate the app as usual. When the data is ready, the user will be prompted to download it.