Diabetic Ketoacidosis Adverse Events

For each DKA event, the relevant data collected on the study CRFs were represented in the AE dataset. The records with AETERM = "Diabetic Ketoacidosis" are the trial disease milestones previously defined in the TM dataset. These AE records include the MIDS variable with the value "DKA(n)". Because these are records for disease milestones rather than associated records, the variables RELMIDS and MIDSDTC are not needed. The variables AESTDTC and AESPID, corresponding with the AE record number from the CRF, were used by the sponsor to determine the chronological order of the events. The severity of the DKA event was based on the set of criteria used by the investigator. The name of the set of criteria used was collected on the CRF and is recorded in the NSV AESSEVCN. The severity criteria are used to standardize the assessment of DKA severity by investigators. The NSV AESTDSEV is used to represent the severity of the each DKA as determined by the set of criteria represented in the AESSEVCN NSV. The use of these NSVs is being discussed and may be changed. Note that the permissible variable, AESEV, is included, but it is blank for all DKA events. 

In this study, the sponsor also collected whether cerebral edema occurred with the DKA event. Because the AE dataset only contains information on events that have occurred, the sponsor created an FAAE dataset to represent the information collected on the example CRF regarding the occurrence of cerebral edema. Any cerebral edema that occurred was represented in the AE dataset. The severity of any cerebral edema was represented in AESEV. 

Note that:

• In this example, the sponsor did not represent the information on whether a DKA occurred. 
• Some expected variables in the AE domain (e.g., AELLT, AEPTCD, AEHLT) have not been shown in the example to save space. Refer to the SDTMIG for examples of how to use these expected variables.

The AEACNDEV variable was used to represent any actions taken with a device with respect to an adverse event; the NSV AERLDEV was used to represent the relationship of the adverse event to a device. If a single device was evaluated with respect to the adverse event, the evaluated device is identified by the value of the SPDEVID variable. When more than 1 device was evaluated with respect to the adverse event, both the AEACNDEV variable and the AERLDEV NSV were assigned the value "MULTIPLE" and the associated data for each question for each device provided using the FAAE dataset. In this case, each evaluated device is identified by the value of the SPDEVID variable in the FAAE dataset and the SPDEVID variable in the AE dataset is blank.

The FA domain was used to represent information for both of the following:

• The actions taken and relationships with multiple devices with respect to individual DKA events. For these records, the value of the FATESTCD variable is "RLDEV" or "ACNDEV" and the SPDEVID variable identifies the device that was evaluated for action taken or relationship with the DKA event.
• The occurrence of the pre-specified adverse event, cerebral edema, in association with each DKA event. For these records, the value of the FATESTCD variable is "OCCUR". The pre-specified adverse event is identified in the FAOBJ variable and the value of the FAORRES variable indicates whether or not the adverse event occurred. (More guidance on the representation of pre-specified adverse events can be found in the SDTMIG.) The SPDEVID variable is blank on these records because the occurrence of cerebral edema was not associated with any device.
  

For all records, the MIDS variable was used to associate the information with a particular DKA event. The RELMIDS variable is populated with a textual description of the temporal relationship between the observation and the disease milestone named in MIDS. Note that controlled vocabulary has not yet been developed for RELMIDS. In this example, the sponsor elected to populate RELMIDS as "ENTIRE EVENT" to reflect that the sponsor evaluated the observation of interest throughout the DKA event. Sponsors may use more specific textual terms to reflect the relationship if needed (e.g., BEFORE, AFTER, DURING).

The FALNKID variable was included in the dataset and populated for the device evaluation records (where FATESTCD is "RLDEV" or "ACNDEV") so that the collected relationships between multiple device evaluations and specific AE records could be represented in the same way for both DKA events and regular adverse events. The FALNKID variable is not populated on the cerebral edema occurrence records (where FATESTCD = "OCCUR") because the sponsor considered the MIDS-defined association with the DKA event to be sufficient. If the MIDS variable had not been used in the study, the collected relationships between the DKA event record in the AE dataset, the cerebral edema occurrence record in the FA dataset, and (when present) the cerebral edema adverse event record in AE would be represented using RELREC.

A contingent visit was assigned based on the specific DKA event occurrence for each subject. The VISITNUM variable was assigned the appropriate contingency visit number.

The RELREC dataset is used to store relationships between the different SDTM datasets. The disease milestone timing variable, MIDS, provides a way to link records associated with a disease milestone without the need for additional linking via RELREC. 

This RELREC table shows the relationship of the records in FAAE to the AE domain.