![]() | |
#21
| |||
| |||
|
#22
| |||
| |||
|
|
OK guys, Thanks so far for the replies. I have thought of another scenario which may provide the solution to problem I face. What if I don't enforce referential integrity? That would mean I would be modelling exactly the real life scenarios we are likely to encounter, with a child not being recruited to the trial, whose parents details we don't possess (for whatever reason) or who may not have either living parents etc, etc). In this case, we have a child table (PK ChildID, FK MotherID, FK FatherID) a father table (PK FatherID) and a mother table (PK MotherID). I know there's duplication going on there. Crucially, if I didn't enforce referential integrity, this would more closely match what happens in the trial and would allow me to create a data entry form which exactly matches the paper CRF (case report form). This is another thing the PI (Principal Investigator) has insisted on. I know that there are dangers with creating orphan records here, but if we did regular queries to check all the orphan records, and knew which ones were meant to be orphan records from the paper CRFs, we could then investigate those which weren't and hopefully remedy the situation. I'd really appreciate some input from you as I'm sure they'll be a thousand and one reasons you'll think of that I shouldn't go down this route. TIA |
#23
| ||||||
| ||||||
|
|
OK guys, Thanks so far for the replies. I have thought of another scenario which may provide the solution to problem I face. What if I don't enforce referential integrity? |
|
That would mean I would be modelling exactly the real life scenarios we are likely to encounter, with a child not being recruited to the trial, whose parents details we don't possess (for whatever reason) or who may not have either living parents etc, etc). |
|
In this case, we have a child table (PK ChildID, FK MotherID, FK FatherID) a father table (PK FatherID) and a mother table (PK MotherID). I know there's duplication going on there. |
|
Crucially, if I didn't enforce referential integrity, this would more closely match what happens in the trial and would allow me to create a data entry form which exactly matches the paper CRF (case report form). This is another thing the PI (Principal Investigator) has insisted on. Again, nothing in the design we proposed prevents you from adding a child |
|
I know that there are dangers with creating orphan records here, but if we did regular queries to check all the orphan records, and knew which ones were meant to be orphan records from the paper CRFs, we could then investigate those which weren't and hopefully remedy the situation. |
|
I'd really appreciate some input from you as I'm sure they'll be a thousand and one reasons you'll think of that I shouldn't go down this route. Why use a relational database? It sounds like the PI wants to continue using |
#24
| |||
| |||
|
#25
| |||
| |||
|
|
I think what we have failed to point out is the need to provide a form to enter a person record for a parent. In the subform, there should be a "New Parent" button that launches a modal ParentEntry form to enable the addition of a Person record for the parent. When this form is closed, the subform needs to be refreshed to allow the combo box to display the newly entered Person. Better yet, when the ParentEntry form is closed, the ID of the new Person |
#26
| |||
| |||
|
|
Bob Barrows wrote: I think what we have failed to point out is the need to provide a form to enter a person record for a parent. In the subform, there should be a "New Parent" button that launches a modal ParentEntry form to enable the addition of a Person record for the parent. When this form is closed, the subform needs to be refreshed to allow the combo box to display the newly entered Person. Better yet, when the ParentEntry form is closed, the ID of the new Person record is automatically entered in the combo box. |
![]() |
| Thread Tools | |
| Display Modes | |
| |