The analytic methods available in CLG have been developed in response to real needs of practicing physicians and researchers over many years. CLG studies are done by applying analytic methods to patient groups. The simplest CLG study uses the List method to create a listing of a patient group. Pivot methods are available on-demand to aggregate and summarize data. Beyond these traditional methods, CLG is differentiated from other Business Intelligence tools by analytic methods such as Time to Outcome and Time in Range and by its epidemiological sophistication. New analytic methods under development will focus on controlling analytic bias in group comparison. Analytic methods currently available in CLG are:
Incidence Density: Incidence density for a patient group shows the number of outcome events for the group divided by the total person-time that the group was under observation (at risk) for the outcome. Group difference in incidence density is presented as relative risk and risk difference with confidence limits for all estimates.
Time to Outcome: Based on a retrospective cohort study design, the Time to Outcome method uses subject-specific follow-up periods to query for user-defined outcomes. A Kaplan-Meier survival model is generated for graphical presentation and to allow users to request model estimates with confidence limits at specific time points.
Time in Range: The Time in Range method assesses the percent of time that a cohort is “under control” (e.g. for a given lab value). Originally developed to monitor anticoagulation therapy, CLG can apply this to any clinical parameter. Users define what “under control” means for the given lab value they are investigating. CLG will interpolate between values and roll-up the time each cohort member spends in each user-defined range. Confidence limits are generated for these percentage estimates using non-parametric bootstrap methods.
Analysis Events: CLG Analysis Definitions allow users to append any clinical data to patient groups for analytic purposes.
Pivot: Patient groups and appended analytic data can be viewed in aggregate in the form of a data cube or pivot. CLG produces this on-demand and provides users with an interactive viewer to allow them to slice, dice and filter data in both grid and chart formats.
List: This method produces a data grid for on-screen viewing and for export. By default, lists show patient detail without identifiers. Specific permissions are required to show identifiers and users are required to cite the authority by which they are empowered to see those identifiers for audit purposes each time they request to show them.