NLM logo

National Information Center on Health Services Research and Health Care Technology (NICHSR)

HSRProj (Health Services Research Projects in Progress)

Information about ongoing health services research and public health projects


Development of a causal inference toolkit for patient-centered outcomes research
Investigator (PI): Zhang, Yi
Performing Organization (PO): (Current): Medical Technology and Practice Patterns Institute / (301) 652-4005
Supporting Agency (SA): Patient-Centered Outcomes Research Institute (PCORI)
Initial Year: 2014
Final Year: 2018
Record Source/Award ID: PCORI/ME-1303-6031
Funding: Total Award Amount: $1,043,456
Award Type: Contract
Award Information: PCORI: More information and project results (when completed)
Abstract: One problem central in using real-world data such as Medicare, Medicaid, electronic health records, and clinical registry data to prove cause and effect for treatments and patient outcomes is that confounding occurs that arises because the prior treatment affects the outcome, which, in turn, affects the future treatment, etc. Commonly used statistical methods are not well equipped to appropriately handle this phenomenon and tend to produce biased estimates. In this proposal, we want to introduce researchers to two cutting-edge approaches to address this problem: inverse-probability (IP) weighting of marginal structural models and the parametric g-formula approach. However, these important tools for patient-centered outcomes research (PCOR) are not well understood, due to lack of applications to real-world data. In particular, how to tailor the use of these advanced methods to large administrative data is largely unknown. Existing computer software facilitating the use of IP weighting and g-formula has not been developed to use on such large databases. Furthermore, such software has primarily been accessible to elite statisticians in a few universities and not to general researchers or clinicians. The goal of this research is to develop a specialized toolkit, called the causal inference toolkit (CI-Toolkit) to provide a comprehensive, practical, and accessible guide to implementing these advanced statistical techniques so researchers can ask and answer questions about cause and effect of various treatments-answers patients are most concerned about. Patients want to understand causation, not just the association, when making complicated health-related decisions. Specifically, the CI-Toolkit has three major components: (1) structured/tailored guidelines and recommendations for clinical researchers regarding use of IP weighting and g-formula methods in large observational data; (2) a web-based, user-friendly, open-source software package that contains a set of functions and procedures to facilitate the use of IP weighting and g-formula approach; and (3) three case studies to illustrate the process of adapting guidelines and software developed in 1 and 2 to specific contexts and to show step-by-step implementation. The clinical utility, user-friendliness, and effectiveness of the CI-Toolkit will be evaluated among multiple stakeholders so that it can be easily shared by applied researchers in need. The successful development of CI-Toolkit represents a major step towards filling a methodological gap regarding statistics that can be used to produce cause-and-effect results in the PCORI draft Methodology Report. We hope the CI-Toolkit will promote broader use of these advanced statistical methods in future studies and provide patients and their caregivers and families with the scientifically valid information they need to make informed decisions about their health care.
MeSH Terms:
  • Algorithms
  • Electronic Health Records
  • Humans
  • Internet
  • Medicaid
  • Medicare
  • Models, Statistical
  • * Outcome Assessment, Health Care
  • * Patient-Centered Care
  • Probability
  • Registries
  • Software
  • Statistics as Topic
  • United States
Country: United States
State: Maryland
Zip Code: 20816
UI: 20143592
Project Status: Completed
Record History: ('2017: Project extended to 2018.',)