Dear all,
thanks in advance for your help and your time.
I have a dataset with approximately 1500 patients.
For each of these patients, I have the following variables:
My initial approach would be a multilevel logistic model to model the odds of mortality, and include splines to capture non-linearity and insert the interaction spline*department.
However, I am asked to evaluate the incidence (not the odds!) of mortality.
My problem is that to obtain incidences, I need to collapse the events using a sum within a certain time frame (let's say a month or a week).
The issue is that in this way, with aggregated data, I wouldn't be able to adjust the estimates of a count model (Poisson, negbinomal...)for the different severity scores (or anyother variable) of the single patients.
At the moment, I only think of stratifying by age, sex, score.
Does anyone have an idea of how to proceed?
Please find belw how my dataset appears:
Thanks again.
Gianfranco
thanks in advance for your help and your time.
I have a dataset with approximately 1500 patients.
For each of these patients, I have the following variables:
- outcome at discharge (variable "outcome," alive=0, dead=1),
- the hospital department (variable "department," oncology, hematology),
- the age class (variable "age," over70, under70),
- a clinical score of the patient (variable "score"),
- discharge date (variable "date"),
- days of hospitalization (variable "days").
My initial approach would be a multilevel logistic model to model the odds of mortality, and include splines to capture non-linearity and insert the interaction spline*department.
However, I am asked to evaluate the incidence (not the odds!) of mortality.
My problem is that to obtain incidences, I need to collapse the events using a sum within a certain time frame (let's say a month or a week).
The issue is that in this way, with aggregated data, I wouldn't be able to adjust the estimates of a count model (Poisson, negbinomal...)for the different severity scores (or anyother variable) of the single patients.
At the moment, I only think of stratifying by age, sex, score.
Does anyone have an idea of how to proceed?
Please find belw how my dataset appears:
Code:
* Example generated by -dataex-. For more info, type help dataex clear input float id str14 department str7 age float(admissiondays score date outcome) 846 "oncology" "over70" 14 3 23378 0 1343 "oncology" "over70" 6 3 23378 0 1348 "oncohematology" "over70" 5 3 23378 0 272 "oncology" "over70" 18 2 23378 1 594 "oncology" "over70" 9 2 23378 1 700 "oncohematology" "over70" 19 2 23378 1 1037 "oncology" "over70" 8 3 23378 0 1110 "oncohematology" "over70" 5 0 23378 0 1216 "oncology" "over70" 1 0 23379 0 1224 "oncology" "over70" 12 4 23379 0 end format %td date
Thanks again.
Gianfranco
