Age - Patient Resource

[joshua-hiltunen]

Documentation for age in the Patient resource is clear, however elicits a broader question about the Age extension. The Age extension indicates that it's presenting "the age of the patient at a specific time", but the extension doesn't provide an element to indicate the specific time.

For Encounter, you have the reference date within the resource itself, but for Patient the specification is that "the Data Submitter has to compute the age based as of December 31st of the previous year for inclusion in the UDS+ report." That date doesn't show up anywhere in the Patient resource, but provides necessary context for interpretation of Age. Should the Age extension include a means of indicating ageOnDate?

[nbashyam]

For de-identified resources, the dates are not allowed. So we cannot add that extension, however your understanding and interpretation is correct.

[joshua-hiltunen]

All elements of dates (except year) for dates that are directly related to an individual, including birth date, admission date, discharge date, death date, and all ages over 89 and all elements of dates (including year) indicative of such age, except that such ages and elements may be aggregated into a single category of age 90 or older

Is date that Age is calculated against "directly related to an individual"? It's an arbitrary date in the year (end of CY in the case of Patient resource), rather than related to the patient directly.

If you allow for arbitrary dates, you give an indirect method for ascertaining birthdate, but doesn't this apply to supplying Age in Encounter as well? Someone looking at Encounter data could find where the Age value ticks up compared to Encounter date to backsolve their way into Date of Birth +/- the time between Encounters for each Patient in the set.

[joshua-hiltunen]

An alternative to reference date might be having a type/category with a fixed set of reference points through the year. It avoids allowing for arbitrary dates.

[joshua-hiltunen]

To jump on that Encounter thing a bit more. If you have a patient that has qualifying encounters once a month, you get that patient's birthdate +/- a month. If "birthdate within a month" was acceptable deidentification, we'd be working with date of birth de-resolved to mm/yyyy, right?

Are there any numbers around average frequency of visit for patients that UDS reporting tracks?

What's the exact value that you get out of having Age in Encounter? Over the course of the 1/1-12/31 reporting period, you'll only ever have x and x+1 as values in the field. Is that degree of Age resolution necessary to a degree that it warrants the roundabout exposure of d.o.b.?

[nbashyam]

Joshua - Good questions.

However all dates are truncated to a YEAR in UDS+, so the submission would be rejected if the dates are more precise than a year.