Create rule DDF00033: A StudyAmendmentReason is linked to a StudyAmendment either as a primaryReason or as a secondaryReason, but not both.

[ASL-rmarshall]

Create rule defined for #114

[ASL-rmarshall]

Rule id updated from DDF99987 to DDF00033.

[ASL-rmarshall]

@DianeWold This one is also ready for review. I created test data as follows:

• Negative: a total of 10 records, only 8 of which are expected to be reported as errors. There are 4 study amendments, each with a different combination of reasons:
  • StudyAmendment_1: 2 records for a primary reason and one secondary reason, both of which have the same code.code value (even though the decodes are different). Both records should be reported as errors.
  • StudyAmendment_2: 3 records for a primary reason and two secondary reasons. The primary reason is "Other" and the secondary reasons have the same code.code value (even though the decodes are different). Only the secondary reasons should be reported as errors.
  • StudyAmendment_3: 2 records for a primary reason and one secondary reason, both of which have the same code.code value the the otherReason values are different. Both records should be reported as errors.
  • StudyAmendment_4: 3 records for a primary reason and two secondary reasons. The primary reason is "IRB/IEC Feedback" and the secondary reasons have the same code.code value for "Other", but with different otherReason values. Only the secondary reasons should be reported as errors.
  • Positive: a total of 5 records, none of which should be reported as an error because the reason codes within each of the two study amendments are unique. (Note that the same code used with different decodes for two study amendments - C99904x2 - is correctly not reported as an error by this rule - but this might suggest the need for another check for a one-to-one relationship between code and decode, probably within the same code system version used across codes in all classes within a study version).

[DianeWold]

@ASL-rmarshall Should discuss whether the rule needs to be modified. Negative and positive test data ran as expected. However, rule may need to be modified to allow multiple instances of the code for "other" if values of otherReason are different.
Also note comment above about checking needed for the same code used for different study amendments.

[DianeWold]

Agreed that multiple instances of the code "other" should trigger the rule, as in this case otherReason would have to be checked by hand. Comparing otherReason values electronically could easily miss slight representations of the same reason.

[BSnoeijerCD]

@DianeWold and @ASL-rmarshall I would say that reasons can be the same across different amendments. So the check should be within an amendment. Not within a study definition. I cannot see whether this now done based on the Core ruled definition. Not sure whether multiple instances of Other should trigger the rule. Although I agree with your point Diane, I believe the check description text should then be different - indicating that it could be right but should be checked. While twice filling the same reason is really an error.