Case Study:
Cohort Identification

Requirements

• Ability to collect, normalize and store patient clinical data from electronic sources such as EMRs
• Ability to store and retrieve multiple forms of molecular data (e.g. gene sequence data where mutations are simply called present or absent)
• Ability to search the data in the data warehouse to identify cohorts

Solution Narrative

One of XB-BIS' main strengths is its ease of use. Researchers and clinicians are not required to learn SQL or other programming languages in order to perform complex queries and basic analysis. XB-BIS provides a simple and intuitive user experience for querying and viewing data.

Dr. Stevens queries XB-BIS for all patients who were diagnosed with suppurative pancreatitis and received a CT scan of the abdomen between 1998 and 2008.

1. Dr. Stevens selects 577.0- suppurative pancreatitis for the diagnosis filter with the date range constraints.

2. Dr. Stevens 74160-CT Scan of Abdomen for the procedure filter with the date range constraints and applies the filter.

3. XB-BIS presents the patients that match the filter in the query results grid. Dr. Stevens may view or report on this set of patients individually or as a group.

Sample Vignette #2: Use Groups for Comparing Cohort Membership

Dr. Stevens is interested in patients diagnosed with pyrexia and how they compare with patients diagnosed with other convulsions.

1. Dr. Stevens queries the patients diagnosed with 780.6-pyrexia, creates and labels a group containing these patients.

2. Dr. Stevens queries the patients diagnosed with 780.39-other convulsions, creates and labels a group containing these patients.

3. Dr. Stevens selects the two groups and creates a Venn diagram of the two groups. Using this tool, Dr. Stevens can create cohort groups of patients who had both diagnoses, only one of the diagnoses or all patients in the Venn diagram. These new groups can in turn be processed with the Venn tool. The result is that PIs and clinicians are able to perform powerful queries without requiring any knowledge of the SQL programming language. XB-BIS enables a self-service model for cohort identification.

Sample Vignette #3: Query on Clinical and Molecular Attributes

Dr. Stevens is interested in all male patients diagnosed with mesothelioma who've had a U95A experiment run against their tumor tissue for a study.

1. Dr. Stevens selects Male on the subject tab.

2. Dr. Stevens selects mesothelioma on the diagnosis tab.

3. Dr. Stevens selects U95A on the molecular experiment tab and applies the filter.

4. XB-BIS presents all users who meet all three filter criteria.

Sample Vignette #4: Longitudinal Filtering

The three previous vignettes queried phenotype and molecular data across all time. Many times, researchers and clinicians need to locate patients who match a particular pattern of events. The sequence and relative duration between events are key input to identifying members of the cohort. Dr. Stevens wishes to identify all PICU patients who were diagnosed with septic shock while in the PICU, then readmitted to the PICU within the next 14 days.

1. Dr. Stevens specifies the sequence of events to XB-BIS which form the longitudinal filter: PICU visit admission, septic shock diagnosis, PICU visit discharge followed by another PICU visit. Dr. Stevens specifies the readmission must occur within 14 days in the Proximity parameter.

2. XB-BIS searches its data mart containing 14,000 PICU patients and returns the 18 who match this longitudinal filter and displays the results in the longitudinal viewer. The first patient below (Patient 000-093-034) shows three diagnoses of septic shock followed with a readmission a few days later.

3. Dr. Stevens clicks on the Grid tab to view the details. Below shows the specific dates for the events for Patient 000-000-335. After reviewing the longitudinal data, Dr. Stevens creates a cohort group consisting of these patients and contacts the protocol coordinator to request a study on this cohort.