Identifying a minimum of several dates in study definition

[annaschultze]

I understand a fix for this was recently implemented (opensafely-core/cohort-extractor#571), I've just attempted to use this but I'm not sure if I'm specifying it correctly. Code and error message below.

    any_vte_gp=patients.with_these_clinical_events(
        vte_codes_primary_care,
        returning="date", 
        date_format="YYYY-MM-DD",
        on_or_after="index_date",
        find_first_match_in_period=True,
        return_expectations={"date": {"earliest": "index_date"}},
    ),
    any_vte_hospital=patients.admitted_to_hospital(
        returning="date_admitted",
        with_these_diagnoses=vte_codes_secondary_care,
        on_or_after="index_date",
        date_format="YYYY-MM-DD",
        find_first_match_in_period=True,
        return_expectations={"date": {"earliest": "index_date"}},
    ), 
    any_vte_death=patients.with_these_codes_on_death_certificate(
       vte_codes_secondary_care,
       on_or_after="index_date",
       match_only_underlying_cause=True,
       returning="date_of_death", 
       date_format="YYYY-MM-DD",
       return_expectations={"date": {"earliest": "index_date"}},
    ), 
    any_vte=patients.minimum_of("any_vte_gp", "any_vte_hospital", "any_vte_death"),

[wjchulme]

You should drop the return_expectations arguments. The variable is always calculated from other variables, whether in dummy data or real data, so expectations aren't needed, and using them throws an error

Oh hang on, sorry, there is no return_expectations argument in the minimum_of call

[alexwalkerepi]

Is it locally that you're getting an error ? If so have you done an opensafely upgrade and opensafely pull recently?

[annaschultze]

Ah, that solved it! I'd run opensafely upgrade but had forgotten to also run opensafely pull.