Skip to main content

Sentinel Routine Querying System

    Basic Details
    Date Posted

    Sentinel routine querying tools are SAS programs designed to run against the Sentinel Common Data Model (SCDM). They allow rapid implementation of standard queries across the Sentinel Distributed Database (SDD). The programs can be customized using input parameters that define medical product exposures, outcomes, covariates, diagnoses, date ranges, age ranges, and other implementation details. Tools can perform simple cohort characterization and descriptive analyses, but may also be used to perform more complex adjustment for confounding and to support prospective surveillance activities. 

    The Cohort Identification and Descriptive Analysis (CIDA) program is the foundation of the routine querying system. CIDA is responsible for identifying, extracting, and characterizing cohorts of interest from the SDD based on the specification of a number of requester-defined options (e.g., continuous enrollment requirements, incidence criteria, inclusion/exclusion criteria). CIDA may be used to calculate simple background rates of health outcomes of interest (HOIs) (e.g., prevalence of acute myocardial infarction), or rates of medical product use (e.g., new warfarin use), or it may be used for more complex queries that identify the occurrence of HOIs during exposure to a medical product of interest (e.g., number of incident diagnoses of rhabdomyolysis during new treatment with a statin). 

    The CIDA program, by default, will output summary-level counts (e.g., number of new users, number of HOIs) stratified by various parameters (e.g., age group, sex, year, year-month). CIDA will also output metrics on eligible members and eligible member days associated with each result stratum, allowing for the calculation of proportions and rates, and an attrition table, to determine the number of eligible members removed from consideration after application of various cohort selection criteria.

    The CIDA program is integrated with additional, optional modules that perform more complex analyses and confounder adjustment:

    • Propensity score matching: estimates a propensity score based on predefined covariates and/or via a high-dimensional propensity score approach.  Members in an exposed cohort are matched to members in a comparator cohort by propensity score; hazard ratios, incidence rate differences and 95% confidence intervals are calculated.
    • Self-controlled risk interval design: extracts a cohort based on exposure(s) of interest, defines a risk and control window for each patient relative to exposure, and counts the occurrence of health outcomes during the risk and control periods.
    • Comorbidity score stratification: calculates a combined Charlson/Elixhauser comorbidity score and stratifies output by requester-defined ranges of values.
    • Medical utilization stratification: calculates the number of medical visits during a defined time period and stratifies output by requester-defined ranges of values.

    The functionality and modules used for a particular routine query request is dependent on the question of interest and required output.

    The Sentinel Routine Querying System Documentation is located externally on Sentinel's Git Repository. The Git Repository serves as Sentinel's version control tracking system for analytic packages and technical documentation. A history of modifications for the Sentinel Routine Querying System is located here.