Skip to main content

2020-4521

Research Proposal

Project Title: 
Causally interpretable meta-analysis with missing data: Generalizing evidence from bipolar disorder and schizophrenia trials to a target population
Scientific Abstract: 

Background: Many RCTs evaluate the effects of antipsychotic drugs on patients with schizophrenia or bipolar disorder. There is interest in synthesizing evidence across the different trials to improve precision of estimators of treatment efficacy. Furthermore, trial participants often differ from the underlying target population. This raises the question of how to combine information from multiple trials in a way that is interpretable in the context of the target population of interest. Such analysis are often complicated by data being systematically missing between trials (i.e., information on a certain variable is only collected in some trials).
Objective: We are going to develop and evaluate the performance of new statistical methods for handling systematic missing data in causally interpretable meta-analysis. A part of the evaluation will be done by applying the methods developed to the datasets requested in this data request.
Study Design: We will conduct a causally interpretable meta-analysis using the datasets requested.
Participants: The analysis will be restricted to all participants that are 18 years or older with a DSM-IV diagnosis of schizophrenia or Bipolar I disorder.
Main Outcome Measure(s): For participants with schizophrenia we will focus on Positive and Negative Syndrome Scale total score and for participants with bipolar disorder we will focus on the Young Mania Rating Scale.
Statistical Analysis: The methods developed will be semi-parametric efficient and are extensions of our previously developed methods to the setting of missing data [1].

Brief Project Background and Statement of Project Significance: 

Clinical trial results are commonly used to justify treatment options in different populations than the trial was conducted in (e.g., all participants who are eligible for a clinical trial rather than those that agreed to participate, a different geographic region, or a different clinical setting). Such justification requires generalization/transportation of clinical trial results to the target population for which the treatment is intended. A significant barrier to the practical utility of methods for generalizing/transporting treatment effects is the lack of methods for handling missing data. Data can be missing within trials or target and/or systematically missing where some covariates are not collected in some trials and/or the target. Within trial or target missing data has been extensively studied [2], but methods for addressing systematic missing data, a problem unique to the setting of having data from multiple data sources, have not yet been developed in the context of transporting treatment effects to a target population. We propose to develop and validate methods for handling systematic missing data when transporting treatment effects from one or more clinical trials to a target population, substantially improving the clinical utility of previously developed methods [1] for transportability of treatment effects from randomized controlled trials. The methods developed will be evaluated using the schizophrenia and bipolar disorder datasets requested and account for the differences in the populations underlying each trial. This will lead to results that are more generalizable to clinical practice and will provide understanding of the amount of treatment effect heterogeneity between different trials and how representative they are of more practical settings.

Specific Aims of the Project: 

AIM 1: Develop methods for handling systematic missing data when transporting treatment effects from a single or multiple trials to a target population. This involves: a) developing conditions under which treatment effects can be transported from one or more clinical trials to a target population in the presence of systematic missing data; and b) deriving and developing properties of estimators for transporting treatment effects from one or more clinical trials to a target population in the presence of systematic missing data.
AIM 2: Apply and empirically evaluate the methods developed in Aim 1 using the schizophrenia and bipolar disorder trials requested as a part of this proposal.

What is the purpose of the analysis being proposed? Please select all that apply.: 
Confirm or validate previously conducted research on treatment effectiveness
Participant-level data meta-analysis
Participant-level data meta-analysis using only data from YODA Project
Develop or refine statistical methods
Research on clinical trial methods
Software Used: 
R
Data Source and Inclusion/Exclusion Criteria to be used to define the patient sample for your study: 

In the analysis we will include all randomized trials that evaluate the effect of antipsychotic treatments on schizophrenia and bipolar disorder. To implement the analysis we need individual level data on all participants. This includes outcome information, treatment information, and individual level characteristics that allow us to account for differences in the underlying populations (e.g., demographic information, family history, severity of mental illness at baseline etc.). The analysis will include participants who are older than 18 and are diagnosed with schizophrenia or bipolar I disorder using the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV) diagnosis of schizophrenia or bipolar I disorder. We are requesting individual level data from all trials that involve paliperidone, paliperidone palmitate and/or risperidone (identified through the trial search on the YODA website).

Main Outcome Measure and how it will be categorized/defined for your study: 

For participants with schizophrenia we will focus on Positive and Negative Syndrome Scale total score and for participants with bipolar disorder we will focus on the Young Mania Rating Scale. For the analysis we will focus on the difference in these measures from baseline to end of study.

Main Predictor/Independent Variable and how it will be categorized/defined for your study: 

We are requesting information on all covariates collected in the studies. This includes demographic variables (age, race, ethnicity, etc.), clinical information (BMI, drug abuse, alcohol abuse etc.), and severity of symptoms at baseline. These variables will be used to account for differences in the covariate distributions underlying the studies and will allow us to generalize the treatment efficacy to different populations.

Statistical Analysis Plan: 

For Aim 1 we will use semi-parametric efficiency theory developed in the context of missing data and causal inference to develop methods for generalizing/transporting treatment effects from multiple clinical trials to a target population in the presence of systematic missing data. This involves extending the groups prior work to handling systematic missing data. This also involves identifying conditions under which such transportability analysis can be done and using falsification test of these assumptions to decide which datasets to include in the analysis. For Aim 2 we will use the methods developed in Aim 1 on the trials requested in this application.

Narrative Summary: 

Clinical trials are almost always conducted with a specific target population in mind, but data collected is rarely a random sample of that population. This project aims to develop new statistical methods that allow us to transport/generalize treatment effects from multiple clinical trials to the target population of interest. The methods will be evaluated using data from several bipolar disorder and schizophrenia randomized trials creating generalizable evidence about treatment efficacy.

Project Timeline: 

These analysis are a part of a PCORI funded proposal with a funding period from 1/1/21-1/1/23. In year 1 we expect to develop the methods described in Aim 1 of the proposal and clean and harmonize the datasets. In year 2 we expect to analyze the data using the methods developed in Aim 1.

Dissemination Plan: 

The results from this project will be journal publications and presentations at research meetings. The target audience will be statisticians, clinical trialists, and psychiatrists and the publication and presentation venues will focus on these groups (e.g, submit to more methodological work in biostatistics journal and more applied work in psychiatry journals).

Bibliography: 

[1] Dahabreh, I. J., Petito, L. C., Robertson, S. E., HernĂ¡n, M. A., & Steingrimsson, J. A. (2020). Toward Causally Interpretable Meta-analysis: Transporting Inferences from Multiple Randomized Trials to a New Target Population. Epidemiology, 31(3), 334-344.

[2] Little, Roderick J., Ralph D'Agostino, Michael L. Cohen, Kay Dickersin, Scott S. Emerson, John T. Farrar, Constantine Frangakis et al. "The prevention and treatment of missing data in clinical trials." New England Journal of Medicine 367, no. 14 (2012): 1355-1360.

General Information

How did you learn about the YODA Project?: 
Colleague

Request Clinical Trials

Associated Trial(s): 
What type of data are you looking for?: 
Individual Participant-Level Data, which includes Full CSR and all supporting documentation

Data Request Status

Change the status of this request: 
Ongoing