10
Access Database Design Access Database Design for Clinical Research for Clinical Research Studies Studies October 17, 2003 October 17, 2003 Paul A. Harris Paul A. Harris Julie McFarlane Julie McFarlane GCRC Research Skills Workshop GCRC Research Skills Workshop

Access Database Design for Clinical Research Studies

  • Upload
    fraley

  • View
    29

  • Download
    3

Embed Size (px)

DESCRIPTION

Access Database Design for Clinical Research Studies. October 17, 2003 Paul A. Harris Julie McFarlane GCRC Research Skills Workshop. Clinical Database Design. Task #1 – Determine the discrete measurement types and the times of occurrence (1 hour + ?) - PowerPoint PPT Presentation

Citation preview

Page 1: Access Database Design for Clinical Research Studies

Access Database Design Access Database Design for Clinical Research for Clinical Research StudiesStudies

October 17, 2003October 17, 2003

Paul A. HarrisPaul A. Harris

Julie McFarlaneJulie McFarlane

GCRC Research Skills WorkshopGCRC Research Skills Workshop

Page 2: Access Database Design for Clinical Research Studies

Clinical Database Clinical Database DesignDesign Task #1 – Determine the discrete Task #1 – Determine the discrete

measurement types and the times of measurement types and the times of occurrence (1 hour + ?)occurrence (1 hour + ?)

Excel is a great tool to perform this 1Excel is a great tool to perform this 1stst cut view of the data as your final product cut view of the data as your final product should be a grid AND if possible, the should be a grid AND if possible, the vertical axis can be used later to actually vertical axis can be used later to actually create the Access tables.create the Access tables.

Page 3: Access Database Design for Clinical Research Studies

Same Measurement Frequency1 Time Per Study or PatientIn a one-time study, we can group these withthe Demographic Information to reduce tables(ie one-to-one table relationships aren’t that helpful.

Time

DiscreteMsmts

Drugs areDifficult toCapture (Dose/Type/Start/Stop)

Note units and examplesgiven whenever possible

Page 4: Access Database Design for Clinical Research Studies

Clinical Database Clinical Database DesignDesignTask #2 – Ask Task #2 – Ask QuestionsQuestions1) Who/what/where – This will help you think of where the 1) Who/what/where – This will help you think of where the database file should be stored, security, interface, etc. database file should be stored, security, interface, etc. (JM – (JM – 1 user; JM office – network server)1 user; JM office – network server)

Who will enter data?, Where will data be entered?, Will more Who will enter data?, Where will data be entered?, Will more than one person need to add data?, Where will data be than one person need to add data?, Where will data be stored?, Who will need to access the data and when? stored?, Who will need to access the data and when? (JM (JM entering data / NB access periodic – cannot see server)entering data / NB access periodic – cannot see server)

2) What sorts of questions will you try to answer with the 2) What sorts of questions will you try to answer with the database – this will help in designing tables and queries database – this will help in designing tables and queries (Does ACE Inhibitor correlate with decrease in AFIB? Initially (Does ACE Inhibitor correlate with decrease in AFIB? Initially blinded – pharmacy to give key later by vial label number)blinded – pharmacy to give key later by vial label number)

Will we need real-time query tools? Will we need the final Will we need real-time query tools? Will we need the final numbers packaged in a certain way? numbers packaged in a certain way? (Individual (Individual measurement groups and/or individual measurements – all to measurement groups and/or individual measurements – all to Excel file)Excel file)

Page 5: Access Database Design for Clinical Research Studies

Clinical Database Clinical Database DesignDesignTask #3 – Table Task #3 – Table StructureStructureUse information gained in the Excel exercise and in the Use information gained in the Excel exercise and in the question and answer session to determine the optimum question and answer session to determine the optimum table design structure.table design structure.

Work on paper before committing time to programming. If Work on paper before committing time to programming. If the Excel file is in good shape, you can often manipulate the Excel file is in good shape, you can often manipulate and import rather than typing in Access.and import rather than typing in Access.

After setting up tables, always set up formal relationships After setting up tables, always set up formal relationships between tables to enforce referential integrity.between tables to enforce referential integrity.

If possible, write queries to repackage and ask questions If possible, write queries to repackage and ask questions about your data before going on – this will ensure you about your data before going on – this will ensure you have a table design that will eventually work (use dummy have a table design that will eventually work (use dummy data for this exercise)data for this exercise)

Page 6: Access Database Design for Clinical Research Studies

Clinical Database Clinical Database DesignDesignTask #3 – Table Task #3 – Table StructureStructure

Think about final user interface. Does the table Think about final user interface. Does the table structure look appropriate?structure look appropriate?

Do all tables have a unique (primary key) field? Is it Do all tables have a unique (primary key) field? Is it logical? If not, add a tablename_id autonumber logical? If not, add a tablename_id autonumber field. You really strive to never design a table that field. You really strive to never design a table that doesn’t have a unique field.doesn’t have a unique field.

Are all tables related through a primary/foreign key Are all tables related through a primary/foreign key relationship? You should be able to start at the relationship? You should be able to start at the lowest data table and work back to high level lowest data table and work back to high level details following foreign/primary key links if table details following foreign/primary key links if table relationships are properly set up (think one to relationships are properly set up (think one to many relationships).many relationships).

Page 7: Access Database Design for Clinical Research Studies

Clinical Database Clinical Database DesignDesignReal-World ExerciseReal-World Exercise

Look at the example spreadsheet and think about table Look at the example spreadsheet and think about table structures for this project.structures for this project.

Demographics (include one-shot msmts unless this data Demographics (include one-shot msmts unless this data will be one set of a larger group of data). The study will be one set of a larger group of data). The study number or patient MRN is a logical choice for PK here.number or patient MRN is a logical choice for PK here.

Most measurements that are not one-shot in this study Most measurements that are not one-shot in this study are repeated with sufficient frequency that we might are repeated with sufficient frequency that we might think about dumping all data in one other table. The think about dumping all data in one other table. The PK choice here is not as obvious, but we could PK choice here is not as obvious, but we could perhaps combine the study number with an perhaps combine the study number with an underscore and a time code (visit number) to assure underscore and a time code (visit number) to assure we don’t add the same data for the same patient.we don’t add the same data for the same patient.

Page 8: Access Database Design for Clinical Research Studies

Clinical Database DesignClinical Database DesignReal-World Real-World DemonstrationDemonstration

Page 9: Access Database Design for Clinical Research Studies

Clinical Database DesignClinical Database DesignReal-World Real-World DemonstrationDemonstration

Rework Excel File into sheets for each table with field names and values.

This should really be done only after 2-3 cuts at Excel file, screening form and flow sheets.

Import sheets for demographicsand study measurements into Access.

Clean and check results.

Set up relationships.

Page 10: Access Database Design for Clinical Research Studies

Clinical Database Clinical Database DesignDesignUser InterfaceUser Interface

Your goals in designing a user interface should be:Your goals in designing a user interface should be:

1) Provide an intuitive interface for users that 1) Provide an intuitive interface for users that matches the logical workflow process.matches the logical workflow process.

2) Completely protect your database from 2) Completely protect your database from inappropriate data entry. Guide the process of inappropriate data entry. Guide the process of data entry.data entry.

Help those poor overworked souls do their jobs.

Don’t let those misfits put garbage into this application.