Traumatic Brain Injury and Depression: A Survival Analysis Study in R (Part 0)
December 30, 2024
Featured
Research
Tutorials
Introduction
This retrospective observational study examines whether depression influences all-cause mortality risk within five years following the first-year interview in individuals with moderate-to-severe traumatic brain injury (msTBI). The analysis will be conducted both with and without adjustment for established mortality risk factors.
Data are sourced from the Traumatic Brain Injury Model Systems (TBIMS), a multi-center longitudinal cohort study encompassing a large dataset of individuals with msTBI throughout the United States. Cox proportional hazards modeling will be employed for time-to-event analysis.
The study cohort includes individuals enrolled in the TBIMS who experienced a msTBI between October 1, 2006, and October 1, 2012. Follow-up extended through October 1, 2022.
Measures
Primary Exposure Measure: Depression Level at Year 1
Depression levels were computed from participants' responses to the Patient Health Questionnaire (PHQ-9) at the first follow-up interview following the index injury (i.e., Year 1). The PHQ-9 consists of 9 items that can be scored from 0 (Not at All) to 3 (Nearly Every Day), producing a total score that ranges from 0 to 27.
In the present study, participants' depression severity levels are classified as "No Depression," "Minor Depression," or "Major Depression" based on their endorsement of positive symptoms and cardinal symptoms of depression for the two-week period that preceded the Year 1 interview.
First, the positive symptoms of depression were calculated for each participant as the total number of PHQ-9 items scored at 1 or higher, indicating that the symptom was present for several days or more over the two-week period prior to the interview. The categorization process considers both the total count of positive symptoms (i.e., total number of PHQ-9 items endorsed at a frequency of several days or more within the previous two weeks), as well as the presence of cardinal symptoms of depression (i.e., anhedonia and depressed mood, represented by the first and second items of the PHQ-9, respectively).
The No Depression label is reserved for cases in which the total number of positive symptoms is 0 or 1 and/or for cases in which neither cardinal symptom is endorsed (regardless of the total number of positive symptoms endorsed);
The Minor Depression label is reserved for cases in which the total number of positive symptoms is 2-4 inclusive, and one or both cardinal symptoms are endorsed; and
The Major Depression label is reserved for cases in which the total number of positive symptoms ranges from 5-9 inclusive, and one or both cardinal symptoms are endorsed.
Proposed Measures
The following TBIMS variables were proposed as measures in the survival analysis due to their previously observed associations with mortality among TBI patients in the literature.
Additional information about each measure may be accessed through the TBIMS Data Dictionary.
Depression Level at Year 1
Sociodemographic Measures
Mental-Health Related Measures
Functional Status Measures
Additional Variables
Notebooks and Scripts
The provided scripts should be run in the sequential order specified by the file names to ensure the successful execution of the analysis workflow. The scripts are designed to perform the following operations:
Preprocessing (
01_Preprocessing.qmd): This initial script prepares the dataset for subsequent analyses via data cleaning, handling missing values, and imputing Year 1 values to previous and subsequent observations. To ensure reproducibility of results, request and use theTBIMS_2023Q2_SPSSForm 1 and Form 2 files as the raw data input for this script.Exploratory Data Analysis (
02_Exploratory_Data_Analysis.qmd): This script generates exploratory data visualizations and summaries to provide insights into the nature and severity of missing data within the preprocessed dataset.Descriptive Tables (
03_Generate_Descriptive_Tables.qmd): This script generates descriptive tables that provide summaries of the study sample characteristics with respect to key variables.Univariate and Bivariate Plots (
04_Generate_Univariate_and_Bivariate_Plots.qmd): This script generates univariate and bivariate plots of the measures.Cox Regression Analysis (
05_Perform_Cox_Regression_Analysis.qmd): This script performs univariable and multivariable survival analyses using the Cox proportional hazards model. Assumption tests (e.g., proportional hazards and time-invariant covariates, linearity of continuous covariates, multicollinearity) are conducted to ensure the validity of results.
Data Access
Access and use of the TBIMS data is governed by the TBIMS standard operating procedures.
Comments
Newer