Presentation, data and programs at:

Slides:



Advertisements
Similar presentations
How would you explain the smoking paradox. Smokers fair better after an infarction in hospital than non-smokers. This apparently disagrees with the view.
Advertisements

Apr-15H.S.1 Stata: Linear Regression Stata 3, linear regression Hein Stigum Presentation, data and programs at: courses.
Department of Public Health and Primary Care, Cardiovascular Epidemiology Unit, Strangeways Research Laboratory, Cambridge, UK Mendelian randomization:
Lecture 28 Categorical variables: –Review of slides from lecture 27 (reprint of lecture 27 categorical variables slides with typos corrected) –Practice.
Associations between Obesity and Depression by Race/Ethnicity and Education among Women: Results from the National Health and Nutrition Examination Survey,
EPID Introduction to Analysis and Interpretation of HIV/STD Data Confounding Manya Magnus, Ph.D. Summer 2001 adapted from M. O’Brien and P. Kissinger.
HSRP 734: Advanced Statistical Methods July 24, 2008.
Confounders and Interactions: An Introduction
Observational Studies Based on Rosenbaum (2002) David Madigan Rosenbaum, P.R. (2002). Observational Studies (2 nd edition). Springer.
Chance, bias and confounding
Causal Diagrams: Directed Acyclic Graphs to Understand, Identify, and Control for Confounding Maya Petersen PH 250B: 11/03/04.
FINAL REVIEW BIOST/EPI 536 December 14, Outline Before the midterm: Interpretation of model parameters (Cohort vs case-control studies) Hypothesis.
Causal Diagrams for Epidemiological Research
Sampling and Experimental Control Goals of clinical research is to make generalizations beyond the individual studied to others with similar conditions.
Hein Stigum courses E8 DAGs intro 2h, Answers Hein Stigum courses 17. apr. H.S.
Jul-15H.S.1 Short overview of statistical methods Hein Stigum Presentation, data and programs at: courses.
Jul-15H.S.1 Linear Regression Hein Stigum Presentation, data and programs at:
Intermediate methods in observational epidemiology 2008 Confounding - I.
A Longitudinal Study of Maternal Smoking During Pregnancy and Child Height Author 1 Author 2 Author 3.
1 PH 240A: Chapter 8 Mark van der Laan University of California Berkeley (Slides by Nick Jewell)
 Pg : 3b, 6b (form and strength)  Page : 10b, 12a, 16c, 16e.
Stratification and Adjustment
Causal Graphs, epi forum
Concepts of Interaction Matthew Fox Advanced Epi.
Tim Wiemken PhD MPH CIC Assistant Professor Division of Infectious Diseases University of Louisville, Kentucky Confounding.
What You Will Do Identify changeable risk factors that affect your levels of health and personal fitness. Describe lifestyle choices that can improve overall.
Oct-15H.S.1Oct-15H.S.1Oct-151 H.S.1Oct-15H.S.1Oct-15H.S.1 Causal Graphs, epi forum Hein Stigum
Oct-15H.S.1Oct-15H.S.1Oct-15H.S.1 Directed Acyclic Graphs DAGs Hein Stigum
Statistics and Quantitative Analysis U4320 Segment 8 Prof. Sharyn O’Halloran.
October 15H.S.1 Causal inference Hein Stigum Presentation, data and programs at:
Amsterdam Rehabilitation Research Center | Reade Multiple regression analysis Analysis of confounding and effectmodification Martin van de Esch, PhD.
G Lect 21 G Lecture 2 Regression as paths and covariance structure Alternative “saturated” path models Using matrix notation to write linear.
Introduction to confounding and DAGs
Introduction to Logistic Regression Rachid Salmi, Jean-Claude Desenclos, Alain Moren, Thomas Grein.
October 15. In Chapter 19: 19.1 Preventing Confounding 19.2 Simpson’s Paradox 19.3 Mantel-Haenszel Methods 19.4 Interaction.
An Introductory Lecture to Environmental Epidemiology Part 5. Ecological Studies. Mark S. Goldberg INRS-Institut Armand-Frappier, University of Quebec,
Jun-16H.S.1 Confounding and DAGs (Directed Acyclic Graphs) Hein Stigum.
Analytical epidemiology Disease frequency Study design: cohorts & case control Choice of a reference group Biases Alain Moren, 2006 Impact Causality Effect.
Selection Bias Concepts
Section 3.3: The Story of Statistical Inference Section 4.1: Testing Where a Proportion Is.
Tim Wiemken PhD MPH CIC Assistant Professor Division of Infectious Diseases University of Louisville, Kentucky Confounding.
Instructor Resource Chapter 14 Copyright © Scott B. Patten, Permission granted for classroom use with Epidemiology for Canadian Students: Principles,
Case-Control Study Duanping Liao, MD, Ph.D
1 Follow the three R’s: Respect for self, Respect for others and Responsibility for all your actions.
01/20151 EPI 5344: Survival Analysis in Epidemiology Confounding and Effect Modification March 24, 2015 Dr. N. Birkett, School of Epidemiology, Public.
Matching. Objectives Discuss methods of matching Discuss advantages and disadvantages of matching Discuss applications of matching Confounding residual.
POPLHLTH 304 Regression (modelling) in Epidemiology Simon Thornley (Slides adapted from Assoc. Prof. Roger Marshall)
Matched Case-Control Study Duanping Liao, MD, Ph.D Phone:
Rerandomization to Improve Covariate Balance in Randomized Experiments Kari Lock Harvard Statistics Advisor: Don Rubin 4/28/11.
Variable selection in Regression modelling Simon Thornley.
Ten things about Experimental Design AP Statistics, Second Semester Review.
Meta-analysis of observational studies Nicole Vogelzangs Department of Psychiatry & EMGO + institute.
Figure 1. Illustrating confounders with a directed acyclic graph. A.A. Akinkugbe et al. J DENT RES 2016; Copyright © by International &
Using Directed Acyclic Graphs (DAGs) to assess confounding Glenys Webster & Anne Harris May 14, 2007 St Paul’s Hospital Statistical “Rounds”
Lecture Slides Elementary Statistics Twelfth Edition
Epidemiology 503 Confounding.
The Centre for Longitudinal Studies Missing Data Strategy
Validity Generalization
DAGs intro with exercises 3h DirectedAcyclicGraph
DAGs intro with exercises 6h
Hein Stigum courses DAGs intro, Answers Hein Stigum courses 28. nov. H.S.
Scale, Causal Pies and Interaction 1h
DAGs intro without exercises 1h Directed Acyclic Graph
Modeling the Causal Effects of Assisted Reproductive Technology (ART)
Counterfactual models Time dependent confounding
Summary of Measures and Design 3h
Causal diagram showing assumed associations between baseline smoking status, ESRD, and baseline characteristics in the Study of Heart and Renal Protection.
Journal reviews 이승호.
Effect Modifiers.
Summary of Measures and Design
Presentation transcript:

Presentation, data and programs at: Simple Causal Graphs Simple Casual Graphs Hein Stigum Presentation, data and programs at: http://folk.uio.no/heins/ Apr-17 H.S.

Causal graphs Simple causal graphs Directed Acyclic Graphs (DAGs) Proper analysis (adjust or not) Direction of bias Directed Acyclic Graphs (DAGs) Formal tool Inventory of variables Causal inference Apr-17 H.S.

Exposure-Disease influenced by C C can be: Confounder Intermediate (in 2. Path) Collider Effect modifier Use graphs Determine C-type Choose analysis C E D Apr-17 H.S.

Example Exposure Disease Covariates Pysical Activity: PA Diabetes type 2: D2 Covariates Smoking: S Health Conscious: HC Overweight: Ov Blood Pressure: BP Apr-17 H.S.

Linear models Best model? Model choice can not be based on data only. Best model? Likelihood ratio tests or Akaike criteria  mod 4 All changes in PA effect considered important  mod 4 Claim mod 2. Model choice can not be based on data only. Need extra info or assumptions. All est significant (except PA=0) 20% D2 overall. PA from 0-5. Max PA is -10 pp. Apr-17 H.S.

No influence of C E D C E D C E D C Apr-17 H.S.

Confounder: Smoking Should adjust for Smoking Stratify Regression S Negative bias + - biased true PA D2 -3 -2 Should adjust for Smoking Stratify Regression Apr-17 H.S.

Confounder 2 Adjust for Smoking or for Health Consciousness + - HC S Negative bias biased true PA D2 -3 -2 Adjust for Smoking or for Health Consciousness Assume all following models are adjusted for smoking Problem: if effect from HC to D2 (via diet ) If we include both, the simple pos/neg bias calc will not work unless both pathways have the same sign Could adjust for HC if measured, or for both Smoking and Diet. DAGs will not give this info, a important shortcoming. Apr-17 H.S.

Intermediate (in 2. path): Overweight + - Alt 1: Ignore Overweight Total -2.0 PA D2 Alt 2: Two models: Direct c2 -1.5 Indirect b1*c1 -0.5 Total c2+ b1*c1 -2.0 Ov Ov c1 Assume all these model are adjusted for smoking b1=-0.25, c1=2, b1*c1=-0.5 b1 c2 PA PA D2 Simply adjusting for Overweight is not OK! Apr-17 H.S.

Select limping subjects Collider idea Two causes for limping Hip arthritis Limp Knee injury Select limping subjects + Limp + + Hip arthritis Knee injury - Conditioning on a collider induces an association between the causes Condition = (restrict, stratify, adjust) Bias direction? Hip arthritis and knee injury are not associated Hip arthritis and knee injury are rare events Apr-17 H.S.

Collider: Blood Pressure BP Positive bias if we adjust + - true biased PA D2 Should not adjust for Blood Pressure Problem if selection is connected to BP Study of subjects with high BP, decide to analyze PA-D2 association Or, based on invitation letter advertising BP-measurement, people with high BP select into study Apr-17 H.S.

Best model (so far) Model 2 is best Used extra info in graphs to decide All est significant (except PA=0) 20% D2 overall. PA from 0-5. Max PA is -10 pp. Apr-17 H.S.

Effect modifier: Sex Alt 3 : Ignore Sex Alt 1 : Two models PA D2 Alt 1 : Two models Easy No test for interaction Inefficient (12 estimates) p=number of covariates Estimates=2(p+1) versus p+2 Alt 2: Model with interaction Technical Test for interaction Efficient (7 estimates) Apr-17 H.S.

Effect modifier: Sex Model with interaction term Linear model Test for interaction Wald test on b3=0 If significant interaction Sex is coded 0 for Males and 1 for Females The effect of PA (1 unit increase) -2.5 -1.5 Apr-17 H.S.

Examples Apr-17 H.S.

Smoking and LRTI The truth is out there? LRTI=Lower Resperatory Tract Infections Want: effect of smoking in pregnancy on LRTI in children Have: 40% response, high education is overrepresented Best causal estimate: Crude smoke-LRTI under 100% response? Crude smoke-LRTI under 40% response? LRTI Smoke Educ - S Smoking in pregnancy Lower Respiratory Tract Infections Education a confounder? Strong selection on educ in data, effect? Education is a confounder Selection represents partial adjustment Apr-17 H.S.

Smoking and LRTI, ex 2 LRTI Smoke Educ S Education is a not a confounder Crude smoke-LRTI in population is unbiased Crude smoke-LRTI in sample is biased, S is a collider Adjusted smoke-LRTI in sample is unbiased Apr-17 H.S.

Ethnicity and lung function Exposure Ethnic group Outcome Lung function Covariates Hemoglobin, height Draw DAG Suggest analyzes/models Model with all covariates meaningful? Lung func Hemo Height Ethnic Apr-17 H.S.

Models Model 1 Lung func Ethnic Model 2 Lung func Hemo Height Hart rate Model 3 Model 4 Lung func Hemo Height Ethnic Model 1: total effect of ethnic group on lung function Model 2: adjusted effect of hemoglobin on lung function Model 3: Add all covariates, if effect of ethnic group, then some physio variable is missing Model 4: ethnic group as a effect modifier of hemoglobin on lung function Apr-17 H.S.

Summing up In a study of 2 variables, a 3. variable may have 4 effects: Confounder, Intermediate, Collider, Effect modifier Not distinguished from data, need extra info Casual graphs help use the extra info The 3 first situations can not be gleaned from the data Apr-17 H.S.