How is actualVentdays in apachePatientResult table calculated?

[mtwayne]

How is the actualVentdays in the apachePatientResult table calculated?

When we compare ventilation using the actualVentdays to ventilated using the vent from apacheApsVar, there are almost 5000 patients who are ventilated per the vent from apacheApsVar that do not have an actualVentdays ventilation duration.

[jraffa]

Hi @mtwayne, I believe the vent variable is defined as if the APACHE respiratory rate was taken while on ventilation, not if the patient had been ventilated. I think this maybe the reason for the discrepancy.

[mtwayne]

If actualVentdays uses the Vent variable to define itself, then shouldn't these variables capture the same number of patients?

Said another way, if the vent variable must be 1 for actualVentdays to capture a value, then shouldn't every patient with vent variable of 1 have a value for actualVentdays? Or at least a value of 1?

Thanks

[jraffa]

The vent variable and ventdays variables are generated separately (one from the APACHE assessment on the first day and the other from the EHR at the conclusion of the unitstay), and should not be thought to be dependent on each other.

With that said, I think your assumption is right. Logically, if a patient was ventilated for their RR (vent), then they should have some (non-zero) vent days, but because they are determined independently they still may not match.

I've been told that the ventday calculations for APACHE are not the most accurate. Each hospital charts the ventilation in a different way, and unfortunately, the only way to get this data out is to determine which set of hospitals have reliable data and to extract the data from that subset of hospitals.

Our lab will likely release some code for this at some point, but it's a challenging task, and requires a lot of careful vetting of the data. Unfortunately, I don't have an ETA for this, but we would happily take pull requests in the meantime.