The Current Landscape in Biostatistics of Real-World Data and Evidence: Causal Inference Frameworks for Study Design and Analysis

Less than 1% of the US population lives in long-term care facilities, yet this subset of the population accounts for 22% of total COVID-19 related deaths.1Because of a lack of experimental evidence to treat COVID-19, analysis of real-world data to identify causal relationships between treatments/policies to mortality and morbidity among high-risk individuals is critical. We applied causal inference (CI) analysis to longitudinal patient-level health data of 4,091 long-term care high-risk patients with COVID-19 to determine if any actions or therapies delivered from January to August of 2020 reduced COVID-19 patient mortality rates during this period.Causal inference findings determined that certain supportive care interventions caused reduced mortality rates for nursing home residents regardless of severity of disease (as measured by oxygen saturation level, presence of pneumonia and organ failure), comorbidities or social determinants of health such as race, age,and weight.2While we do not address the biological mechanisms associated with specific medical interventions and their impact on mortality, this analysis suggests methods to validate and optimize treatment protocols using domain knowledge and causal inference analysis of real-world data across patient populations and care settings.

Download Full-text

Causal Inference From Real-World Data

Journal of the American College of Cardiology ◽

10.1016/j.jacc.2018.05.036 ◽

2018 ◽

Vol 72 (5) ◽

pp. 486-488

Author(s):

Peter Brønnum Nielsen ◽

Flemming Skjøth ◽

Mette Søgaard

Keyword(s):

Causal Inference ◽

Real World ◽

Real World Data ◽

World Data

Download Full-text

Analysis of cause-effect inference by comparing regression errors

PeerJ Computer Science ◽

10.7717/peerj-cs.169 ◽

2019 ◽

Vol 5 ◽

pp. e169 ◽

Cited By ~ 5

Author(s):

Patrick Blöbaum ◽

Dominik Janzing ◽

Takashi Washio ◽

Shohei Shimizu ◽

Bernhard Schölkopf

Keyword(s):

Causal Inference ◽

Least Squares ◽

Real World ◽

Causal Relation ◽

Data Sets ◽

Noise Distribution ◽

Real World Data ◽

World Data ◽

Causal Direction ◽

Inference Methods

We address the problem of inferring the causal direction between two variables by comparing the least-squares errors of the predictions in both possible directions. Under the assumption of an independence between the function relating cause and effect, the conditional noise distribution, and the distribution of the cause, we show that the errors are smaller in causal direction if both variables are equally scaled and the causal relation is close to deterministic. Based on this, we provide an easily applicable algorithm that only requires a regression in both possible causal directions and a comparison of the errors. The performance of the algorithm is compared with various related causal inference methods in different artificial and real-world data sets.

Download Full-text

Causal Inference for Observational Studies/Real-World Data

Real-World Evidence in Drug Development and Evaluation ◽

10.1201/9780429398674-7 ◽

2021 ◽

pp. 129-150

Author(s):

Bo Lu

Keyword(s):

Causal Inference ◽

Real World ◽

Observational Studies ◽

Real World Data ◽

World Data

Download Full-text

Methodological considerations when analysing and interpreting real-world data

Rheumatology ◽

10.1093/rheumatology/kez320 ◽

2019 ◽

Vol 59 (1) ◽

pp. 14-25 ◽

Cited By ~ 4

Author(s):

Til Stürmer ◽

Tiansheng Wang ◽

Yvonne M Golightly ◽

Alex Keil ◽

Jennifer L Lund ◽

...

Keyword(s):

Study Design ◽

Real World ◽

Propensity Scores ◽

Real World Data ◽

Design Elements ◽

World Data ◽

Treatment Regimens ◽

Immortal Time Bias ◽

Methodological Considerations

Abstract In the absence of relevant data from randomized trials, nonexperimental studies are needed to estimate treatment effects on clinically meaningful outcomes. State-of-the-art study design is imperative for minimizing the potential for bias when using large healthcare databases (e.g. claims data, electronic health records, and product/disease registries). Critical design elements include new-users (begin follow-up at treatment initiation) reflecting hypothetical interventions and clear timelines, active-comparators (comparing treatment alternatives for the same indication), and consideration of induction and latent periods. Propensity scores can be used to balance measured covariates between treatment regimens and thus control for measured confounding. Immortal-time bias can be avoided by defining initiation of therapy and follow-up consistently between treatment groups. The aim of this manuscript is to provide a non-technical overview of study design issues and solutions and to highlight the importance of study design to minimize bias in nonexperimental studies using real-world data.

Download Full-text

Estimates of Causality with Medical Image in Oncology

ANALES RANM ◽

10.32440/ar.2021.138.01.rev02 ◽

2021 ◽

Vol 138 (138(01)) ◽

pp. 16-23

Author(s):

Luis Martí-Bonmatí

Keyword(s):

Causal Inference ◽

Real World ◽

Observational Studies ◽

Clinical History ◽

Response To Treatment ◽

Control Analysis ◽

Real World Data ◽

World Data ◽

New Research ◽

Case Control Analysis

This work defines a research on data strategy focused on medical imaging and derived image biomarkers to critically assess the concept of causal inference and uncertainties. Computational observational studies will be valued to generate casual inference from real world data. Our main goal is to propose a scientific methodology that allows to estimate causalities from observational studies through quality control of large databases, definition of plausible hypotheses, using computational estimated models and artificial intelligence tools. The computational approach of radiology to precision medicine by using epidemiological strategies is based on causal inference studies relies on real-world data observational, longitudinal, case-control analysis designed (being case the presence, and control the absence of the event to be estimated). In this new research setting, we consider disease in classical epidemiology as phenotyping, response to treatment and final prognosis; and exposure equals to the presence of a radiomic, dynamic image biomarker or AI modeling solution. Research with data on which causality is to be inferred must control for recruitment of closed cases, in which the researcher does not intervene in the patient’s clinical history but works on databases, collecting data to be secondary used in generating consistent causalities.

Download Full-text