Status
Conditions
Treatments
Study type
Funder types
Identifiers
About
The general aim of the project is training and testing a predictive model of inpatient rehabilitation stay after hip and knee replacement for osteoarthritis. This specific part focuses on data collection and analysis for the model validation.
Full description
The following variables will be investigated to evaluate their predictive value towards length of rehabilitation stay (primary outcome), place of discharge, burden of care at discharge and/or functional recovery (secondary outcomes):
Due to retrospective study design, patients will not be recruited ex-novo and will be used data already stored in the internal databases.
Data retrieved from clinical records and other sets internal to the hospital will be collected on a dedicated Excel sheet and divided per section of interest (baseline, intervention, outcomes).
Data from 1710 patients who underwent total hip or knee replacement and subsequent inpatient rehabilitation in our hospital were collected, in the previous part of this project, to train the model.
In this second part, the same variables from 400 patients will be collected to test the model predictivity. Put together, these samples represent around 80% and 20% of the total population (2100) respectively, which is a common cohort distribution to test for validity.
To verify the validity (discrimination and calibration) of the newly developed prediction and stratification models, we will conduct a temporal external validation based on a sample of newly collected patients undergoing hip and knee at the same hospital but at a different point in time (2018, as compared to 2019 for the development sample).
To this aim, we plan to include a convenience sample of at least 400 patients: the sample size was determined based on a minimum sample size calculation, using Hoeffding's inequality (DOI: 10.1016/j.jspi.2012.09.013), which is sufficient to identify differences in performance smaller than 10% performance points with confidence higher than 95% (minimum sample size=390).
Basic modelling methodologies such as linear and logic regression will also be taken into consideration.
The best performing model for each outcome of interest will be finally selected.
Enrollment
Sex
Ages
Volunteers
Inclusion criteria
Exclusion criteria
400 participants in 1 patient group
Loading...
Central trial contact
Coordinator of the Scientific Direction; Grant office operator
Data sourced from clinicaltrials.gov
Clinical trials
Research sites
Resources
Legal