Mortality outcomes in AmsterdamUMCdb

[ningyile]

I have noticed discussions in the GitHub AmsterdamUMCdb community regarding patient mortality outcomes. Specifically, in a discussion about whether in-hospital mortality rates can be calculated. It is clear that the destination field in the admissions table represents ICU outcomes. Due to the lack of a field for in-hospital mortality outcomes, it cannot be determined in-hospital mortality. It should be noted that 30-day mortality could be calculated instead of in-hospital mortality at the end of this issue.

However, I came across another statement indicating that "the database does not contain information to be completely sure what the follow-up time was for a specific patient" and that "for billing purposes, for a proportion of patients we also have that data while a care path (Dutch: DBC) is active with the hospital. Typically those care paths will be finalized within the year when no chronic care is provided."

Therefore, I would like to clarify:

• Can mortality outcomes in AmsterdamUMCdb be calculated based on the dateofdeath field for 28-day, 90-day, 180-day, and 1-year mortality rates?
• What is the maximum allowable period for calculating mortality rates?
• If the above mortality rates cannot be calculated, why is the 30-day mortality rate deemed calculable?
• Does the dateofdeath field include data on deaths occurring outside the hospital?

Thank you for your attention to these questions. I look forward to your reply at your convenience.

[patrickthoral]

Good question. One of the challenges is losing patient to follow-up. Since our center is a tertiary referral center a number of patients have been discharge to other hospitals that prevented complete documentation on date of death.

Regarding your questions:
Q: Can mortality outcomes in AmsterdamUMCdb be calculated based on the dateofdeath field for 28-day, 90-day, 180-day, and 1-year mortality rates?
A: The primary source of the dateofdeath field was either the ICU specific system and the hospital information system. Like mentioned in one of the other issues, that data from the ICU and or the hospital admission (of the Amsterdam UMC hospital) was known and reflected in AmsterdamUMCdb. However, a proportion of the patients may have died in another hospital or outside the hospital. This was never reliably documented. If it was known (e.g. outpatient follow up due to missing appointments), the data was entered in the system and thus in AmsterdamUMCdb but unfortunately without comparing this with the national Personal Records Database the true value is not known. The 30-day mortality was suggested based on this information, but it should be very similar to hospital mortality, though technically the dateofdeath field contains also data outside of the hospital admission.

Q: What is the maximum allowable period for calculating mortality rates?
For benchmarking (i.e. comparing ICUs) I would not use any non-ICU mortality calculation, but depending on your use case 28/30 may be acceptable.

Q: If the above mortality rates cannot be calculated, why is the 30-day mortality rate deemed calculable?
See above.

Q: Does the dateofdeath field include data on deaths occurring outside the hospital?
A: Yes, but not collected standardized unfortunately, even though technically this could have been possible.

This is something that will hopefully be improved in the next version, since we may be able to use multiple source for a reliable mortality estimate outside the ICU.