Step Count Example 1 (Ametris)

The following should be considered when viewing this resource: 

• The Device Identification Parameter "Device Type" is defined as "Generic category of medical device within the device classification established by a regulatory authority."  This suggests that terms should be taken from a vocabulary established by a regulatory authority. It is not clear what vocabulary should be used.
• Step count data was represented in the Musculoskeletal System Findings (MK) domain. Ambulation is not the product of the musculoskeletal system alone, so this choice of domain could be changed in future.
• The TESTCD and TEST values "STEPSTN" and "Number of Steps Taken" from the Findings About Test Code and Findings About Test Name codelists were used to represent the number of steps, although It is not clear that the meaning of the "awake wear filtered steps" fits the definition of this test, which is, "The number of steps taken by an individual."  The "Number of Steps Taken" test name and code are not in the Musculoskeletal Findings Test Name and Musculoskeletal Findings Test Code codelists.
• The root variable --REFID is defined as, "A sequence of characters used to uniquely identify a source of information." In this example, MKREFID is the "DEVICEID" from the dataset provided by the device vendor. An alternative approach would have been to define a separate identifier (SPDEVID) for each individual device.
• In this example, three specific algorithms involved in the processing of the raw data were identified and were used to populate MKANMETH. SDTM terminology includes "Analytical Method Calculation Formula" codelists associated with the --ANMETH variable for certain other SDTM domains, but there is not currently a codelist associated with the MKANMETH domain.
• Algorithms can also be modeled as properties of the device, but that would not allow the identification of the different algorithms used for a specific test.
• The root variable --LOC is defined as, "The anatomical location of the intervention, event, or assessment." MKLOC was used to represent the location at which the device was worn, under the assumption that the location at which the device was worn is the location of the assessment.
• The study referenced by DiMe Endpoint 399 used a "hip-worn" device, but the sample data used to construct this example was collected by a wrist-worn watch, so MKLOC was populated with "WRIST." The term "WRIST JOINT" is in the SDTM Anatomical Location (LOC) codelist, but "WRIST" is not.
• Note that although there is a unit "STEPS" in the SDTM UNIT codelist, this is a measure of distance, as in "The tree is 12 steps from my front door."  The SDTM test "Number of Steps Taken" includes what is being counted in the name of the test, so in a record for this test the SDTM variables --ORRESU and --STRESU are null. * Some terminology used in the biomedical concepts is present in the NCI Thesaurus, as reflected in C-codes included in biomedical concepts and SDTM specializations.  If an appropriate C-code is not available, the placeholder "Cnew" is used.
• Some SDTM variable values used in the example SDTM datasets are included in CDISC SDTM terminology, but in other cases where a codelist is associated with an SDTM variable, the values used in the example are not in the codelist.

Known Issue

The measurement provided by the device, "awake wear filtered total steps" is a count of steps during the time period which the device determines that the device is being worn and the subject is awake. It is clear that the amount of time the device was worn over the course of a day affects how the number of steps recorded while the device was worn. It is less obvious how the subject being awake or asleep affects the number of "awake aware filtered steps." In this example, it seemed necessary to provide "awake aware minutes" to assist in understanding "awake aware filtered steps". The approach used in this example was to use two MK domain tests, annot be usefully interpreted unless one knows was represented as two tests in the MK domain, "Number of Steps" with an evaluation interval of "AWAKE WEAR MINUTES IN DAY" and "Awake Wear Minutes" with an evaluation interval of "CALENDAR DAY". This solution is less than satisfying, as the awake wear minutes are primarily a measure of how much the device was worn, which is not a finding about the subject's musculoskeletal system. It is more a measure of the device's operation, but is hard to represent in the current Device Exposure (DE) domain, which is designed to represent intervals of time in which the subject is exposed to the device, and this is a summary of the durations of periods of non-exposure. Because of the questions about whether "Awake Wear Minutes" is an appropriate Musculoskeletal Finding the test "Awake Wear Minutes" was not developed as a biomedical concept. 

This approach chosen in this example is similar, in some ways, to the use of the test, "Technical Quality" in the ECG and Holter ECG test name codelists.