Patient Timeline of Time-Varying Features

Definition

A collection of records capturing the changing status of a patient over time with respect to one or more "time-varying" patient features (i.e., patient features that can change over time). Features can be unique to the client, such as a custom patient risk score or engagement level, or more broadly used, such as the CMS RAF score. As with all timelines, each record represents the patient status at any time between the Period Start Date (inclusive) and Period End Date (exclusive) on that record.

Purpose

Used to capture important patient features that change over time; often for the ultimate purpose of filtering and/or stratifying patients more accurately in longitudinal analyses. Many organizations have a number of important home-grown features like this, and the Patient Timeline of Time-Varying Feature is a convenient structure for storing that information.

Related Objects

• Patient Timelines of Time-Varying Features
  • Patient Timelines of Time-Varying Features, Precursor 1 (All Source Records)
  • Patient Timelines of Time-Varying Features, Precursor 2 (Patient-Feature Group Timelines)

Detailed Narrative of Logic

• To facilitate the integration of records from distinct source systems into a single timeline covering diverse concepts, time-varying fields are organized into "Field Blocks".
  • Patient Timelines of Time-Varying Features, Precursor 1 (All Source Records) includes flags for Is Field Block CMS Risk Scoring (covering the CMS RAF Score field), Is Field Block Custom Clinical Features (covering the Risk Score, Risk Category Operational ID, and Risk Category Description fields), and Is Field Block Custom Operational Features (covering the Is Active Patient, Engagement Category Operational ID, and Engagement Category Description fields).
  • The assumption is that, in the context of integrating potentially conflicting information from multiple records, all the field values in a group must be taken together (from the same record) to ensure a coherent representation of the feature or features.
  • Conversely, if a record is not flagged as containing information about a particular block groups, all fields in that block should be ignored.
  • In practice, these conventions allow records containing information on one group of features to be merged with records containing information on a different group of features without one inadvertently overwriting the other.
• Interval records containing time-varying patient features from all contributing source systems, having been mastered and conformed to the appropriate semantic and structural standards, and with the appropriately set Field Block flags, are collected in Patient Timelines of Time-Varying Features, Precursor 1.
  • Taken as a whole, the records collected across all stacks in Patient Timelines of Time-Varying Features, Precursor 1 should represent patient-field block timelines. That is, the intervals should be non-overlapping for the same patient and field block. (This is naturally how the data will be structured if all the information about a particular field group come from a single source, which will be the typical case.)
• The next object in the sequence, Patient Timelines of Time-Varying Features, Precursor 2 (Patient-Field Group Timelines), uses the Simple Timeline object type to effectively guarantee compliance with this patient-field block timeline standard. Specifically, overlapping intervals for the same patient and field block are resolved to favor the interval with the more advanced start date.
• The terminal object in the sequence, Patient Timelines of Time-Varying Features, interleaves the patient-field block timeline intervals into a patient timeline.
• This sequence allows, for example, a record describing a patient's clinical risk status from one source to be combined with a record covering an overlapping period of time but describing a patient's engagement status, producing a single set of timeline records describing both the patient's clinical risk status and engagement status.