1 Arlene Ash QMC - Third Tuesday September 21, 2010 (as amended, Sept 23) Analyzing Observational Data: Focus on Propensity Scores.

Slides:



Advertisements
Similar presentations
The Application of Propensity Score Analysis to Non-randomized Medical Device Clinical Studies: A Regulatory Perspective Lilly Yue, Ph.D.* CDRH, FDA,
Advertisements

Andrea M. Landis, PhD, RN UW LEAH
1.3 Data Collection and Experimental Design
A workshop introducing doubly robust estimation of treatment effects
M2 Medical Epidemiology
V.: 9/7/2007 AC Submit1 Statistical Review of the Observational Studies of Aprotinin Safety Part I: Methods, Mangano and Karkouti Studies CRDAC and DSaRM.
Aftercare Attendance Partially Moderated by History of Physical Abuse and Gender Louise F. Haynes 1 ; Amy E. Herrin 1 ; Rickey E. Carter 1 ; Sudie E. Back.
Predictors of Change in HIV Risk Factors for Adolescents Admitted to Substance Abuse Treatment Passetti, L. L., Garner, B. R., Funk, R., Godley, S. H.,
Differences in Characteristics of Heroin Inhalers and Injectors at Admission to Treatment J. C. Maxwell, R. T. Spence, & T. M. Bohman UT Center for Social.
12 June 2004Clinical algorithms in public health1 Seminar on “Intelligent data analysis and data mining – Application in medicine” Research on poisonings.
Delay from Testing HIV Positive until First HIV Care for Drug Users: Adverse Consequences and Possible Solutions Barbara J Turner MD, MSEd* John Fleishman.
Observational Studies Based on Rosenbaum (2002) David Madigan Rosenbaum, P.R. (2002). Observational Studies (2 nd edition). Springer.
Journal Club Alcohol, Other Drugs, and Health: Current Evidence July-August 2007.
CHILDREN’S MENTAL HEALTH PROBLEMS IN RHODE ISLAND: THE PREVALENCE AND RISK FACTORS Hanna Kim, PhD and Samara Viner-Brown, MS Rhode Island Department of.
1 Arlene Ash QMC - Third Tuesday September 21, 2010 Analyzing Observational Data: Focus on Propensity Scores.
Who are the participants? Creating a Quality Sample 47:269: Research Methods I Dr. Leonard March 22, 2010.
Journal Club Alcohol, Other Drugs, and Health: Current Evidence November-December 2007.
Journal Club Alcohol and Health: Current Evidence May–June 2005.
Journal Club Alcohol, Other Drugs, and Health: Current Evidence May-June 2007.
1 Journal Club Alcohol, Other Drugs, and Health: Current Evidence September–October 2010.
Journal Club Alcohol, Other Drugs, and Health: Current Evidence July–August 2008.
1 Journal Club Alcohol, Other Drugs, and Health: Current Evidence July–August 2011.
Journal Club Alcohol, Other Drugs, and Health: Current Evidence January–February 2011.
Journal Club Alcohol, Other Drugs, and Health: Current Evidence November–December 2008.
Marshall University School of Medicine Department of Biochemistry and Microbiology BMS 617 Lecture 12: Multiple and Logistic Regression Marshall University.
Cohort Study.
Unit 6: Standardization and Methods to Control Confounding.
Multiple Choice Questions for discussion
Advanced Statistics for Interventional Cardiologists.
Chapter 5 Data Production
Chapter 1: Introduction to Statistics
1 Journal Club Alcohol, Other Drugs, and Health: Current Evidence January–February 2014.
Evidence-Based Medicine 3 More Knowledge and Skills for Critical Reading Karen E. Schetzina, MD, MPH.
Disparities in the Adequacy of Depression Treatment in the United States Jeffrey S. Harman, Ph.D. University of Florida Mark J. Edlund, M.D., Ph.D. John.
FRAMING RESEARCH QUESTIONS The PICO Strategy. PICO P: Population of interest I: Intervention C: Control O: Outcome.
S-005 Intervention research: True experiments and quasi- experiments.
Obtaining housing associated with achieving abstinence after detoxification in adults with addiction Tae Woo Park, Christine Maynié-François, Richard Saitz.
Exploring The Determinants Of Racial & Ethnic Disparities In Total Knee Arthroplasty: Health Insurance, Income And Assets Amresh Hanchate, PhD Health Care.
Article Review Cara Carty 09-Mar-06. “Confounding by indication in non-experimental evaluation of vaccine effectiveness: the example of prevention of.
ANOVA and Linear Regression ScWk 242 – Week 13 Slides.
Estimating Causal Effects from Large Data Sets Using Propensity Scores Hal V. Barron, MD TICR 5/06.
Arnold School of Public Health Health Services, Policy, and Management 1 Drug Treatment Disparities Among African Americans Living with HIV/AIDS Carleen.
Analysis Section Research Design. Protocol Overview Background4-5 pages Question/Objective/Hypothesis4 lines Design4-20 lines Study Population0.5-1 page.
Abstinence Incentives for Methadone Maintained Stimulant Users: Outcomes for Those Testing Stimulant Positive vs Negative at Study Intake Maxine L. Stitzer.
+ Terrell Preventable Readmission Project Jeylan Buyukdura & Natalie Davies.
The Health Consequences of Incarceration Michael Massoglia Penn State University.
MBP1010 – Lecture 8: March 1, Odds Ratio/Relative Risk Logistic Regression Survival Analysis Reading: papers on OR and survival analysis (Resources)
Shane Lloyd, MPH 2011, 1,2 Annie Gjelsvik, PhD, 1,2 Deborah N. Pearlman, PhD, 1,2 Carrie Bridges, MPH, 2 1 Brown University Alpert Medical School, 2 Rhode.
1 Lecture 6: Descriptive follow-up studies Natural history of disease and prognosis Survival analysis: Kaplan-Meier survival curves Cox proportional hazards.
1 Multivariable Modeling. 2 nAdjustment by statistical model for the relationships of predictors to the outcome. nRepresents the frequency or magnitude.
A Claims Database Approach to Evaluating Cardiovascular Safety of ADHD Medications A. J. Allen, M.D., Ph.D. Child Psychiatrist, Pharmacologist Global Medical.
Finding a Predictive Model for Post-Hospitalization Adverse Events Henry Carretta 1, PhD, MPH; Katrina McAfee 1,2, MS; Dennis Tsilimingras 1,3, MD, MPH.
Analytical Example Using NHIS Data Files John R. Pleis.
IMPORTANCE OF STATISTICS MR.CHITHRAVEL.V ASST.PROFESSOR ACN.
Using Propensity Score Matching in Observational Services Research Neal Wallace, Ph.D. Portland State University February
Predictors of study retention in addiction treatment trials KORTE JE 1, MAGRUDER KM 1,2, KILLEEN TK 1, SONNE SC 1, SAMPSON RR 1 and BRADY KT 1,2 1. Medical.
1 Statistical Review of the Observational Studies of Aprotinin Safety Part II: The i3 Drug Safety Study CRDAC and DSaRM Meeting September 12, 2007 P. Chris.
Impact of Perceived Discrimination on Use of Preventive Health Services Amal Trivedi, M.D., M.P.H. John Z. Ayanian, M.D., M.P.P. Harvard Medical School/Brigham.
NURS 306, Nursing Research Lisa Broughton, MSN, RN, CCRN RESEARCH STATISTICS.
Trends in Access to Substance Abuse Treatment for Women and Men: Jeanne C. Marsh, PhD, Hee-Choon Shin, PhD, Dingcai Cao, PhD University of Chicago.
2 NURS/HSCI 597 NURSING RESEARCH & DATA ANALYSIS GEORGE MASON UNIVERSITY.
Matching methods for estimating causal effects Danilo Fusco Rome, October 15, 2012.
Predictors of study retention in drug abuse treatment trials
Constructing Propensity score weighted and matched Samples Stacey L
Experiments Why would a double-blind experiment be used?
Impact Evaluation Methods: Difference in difference & Matching
The European Statistical Training Programme (ESTP)
Chapter: 9: Propensity scores
Case-control studies: statistics
Effect Modifiers.
Presentation transcript:

1 Arlene Ash QMC - Third Tuesday September 21, 2010 (as amended, Sept 23) Analyzing Observational Data: Focus on Propensity Scores

2 The Problem Those with the intervention and those without have markedly different values for important measured risk factors & Outcome is related to the risk factors that are imbalanced between the groups & It is not clear how the risk factors and outcome are related Why may standard analyses be misleading?

3 True and Modeled Relationship Between Risk and Outcome

4 Is Imbalance in Risk a Problem? If we correctly model the relationship between risk factors and outcome, we correctly estimate effect of the intervention With many risk factors, hard to know if the relationship between risk factors and outcome is correctly modeled Propensity score - a way to reduce the effect of imbalance in measured risk when models may be inadequate

5 Propensity Score Method (Key Idea) The propensity score (PS) for an observation is the probability that the observation is “exposed” or “got the intervention” Use the PS model in pre-processing the data –To draw a sub-sample where the exposed and non- exposed groups are fairly balanced on risk factors. Then –Use standard techniques to analyze the sub-sample

6 Simple Propensity Score Approach Estimate a model to predict the “probability of intervention/exposure” –This is “the propensity score” Divide the population into PS quintiles Create a subsample by taking equal numbers of exposed and unexposed observations from each quintile Use a subsequent regression model to estimate the effect of the intervention in the subsample

7 Propensity Score Sampling Example PS Quintile# Cases# Controls# Sampled Lowest nd Middle th Highest Total

8 Propensity Score Sampling Example: Treatments for Drug Abusers Patients seeking substance abuse detoxification in Boston receive either Residential detoxification Lasts ~ one week + encouragement for post- detox treatment, or Acupuncture Acute (daily) detox months of maintenance with acupuncture and motivational counseling

9 Data From Boston’s publicly-funded substance abuse treatment system All cases discharged from residential detox or acupuncture between 1/93 and 9/94 Client classified (only once) as residential or acupuncture based on the modality of first discharge

10 Outcome Is client re-admitted to detox within 6 months? (Y/N) Study question: Are acupuncture clients more likely to be re-admitted than residential detox clients? –Exposure = assigned to accupuncture

11 Client Characteristics Available At Time Of Admission Gender Race/ethnicity Age Education Employment status Income Health insurance status Living situation Prior mental health treatment Primary drug Substance abuse treatment history

12 Residential Detox & Acupuncture Cases: % with Various Characteristics Characteristic Residential (n = 6,907) Acupuncture (n = 1,104) Gender: female 2933 Race/ethnicity: black 46 Hispanic White4143 Education: HS grad5659 College graduate413

Employment: unemployed Insurance: uninsured Medicaid Private insurance Lives: with child In shelter Characteristic Residential (n = 6,907) Acupuncture (n = 1,104) Characteristics of Residential Detox & Acupuncture Clients (2)

14 Prior mental health treatment Primary drug: alcohol Cocaine Crack Heroin Characteristic Residential (n = 6,907) Acupuncture (n = 1,104) Characteristics of Residential Detox & Acupuncture Clients (3)

15 Substance abuse admits in the last year Residential detox: Short-term residential: 0 Long-term residential: 0 Outpatient: None Acupuncture: None Characteristic Residential (n = 6,907) Acupuncture (n = 1,104) Characteristics of Residential Detox & Acupuncture Clients (4)

16 Results Of Standard Analysis Percentage of clients re-admitted to detox within 6 months Among 1,104 acupuncture cases, 18% re-admitted Among 6,907 residential detox cases, 36% re-admitted Raw odds ratio = 0.40 From a multivariable stepwise logistic regression model: Odds ratio for acupuncture:0.71 (CI = )

17 What’s the Worry? How Do We Address It? Given how different the two groups are, can we trust a model to correctly estimate the effect of acupuncture? PS methods generalize (long-standing) matching-within- strata methods that work well with 1 or 2 predictors PS can address imbalances in many important predictors simultaneously Both traditional and PS matching allow for –A pooled estimate (across all strata) or –When N is large enough, stratum-specific estimates

18 Propensity Score Application Use stepwise logistic regression to build a model to predict whether a client “is exposed”(i.e., receives acupuncture) Select sub-samples of exposed and non-exposed with similar distributions of the “propensity score” (predicted probability of being exposed) Model (as before) on the sub-sample

19 Sampling Results Able to match 740 who received acupuncture (out of 1,104) with 740 people who did not (out of 6,907) The risk factors in this subsample of 1480 are much more balanced between the two groups

20 Characteristic Residential Acupuncture College graduate Employed Private Insurance Lives with child or adult Lives in shelter Prior mental health Rx 7% 41% 9% 72% 5% 21% (4%) (13%) (3%) (55%) (30%) (12%) 7% 42% 6% 77% 4% 21% (13%) (57%) (15%) (76%) (3%) (28%) Characteristics of Clients in Subsample (vs. Full Sample)

21 Comparing Standard and Propensity Score Findings From the multivariable model fit to all cases: Odds Ratio for acupuncture: % Confidence Interval: From multivariable model fit to more comparable sub- sample: OR for acupuncture: % CI:

22 Summary In this case, results were similar - Why? Original model was very good (C-statistic = 0.96) What we learned from the PS analysis: –Could find a subset of (about 10% of) patients who got residential detox who look very similar to those who got acupuncture –Skeptics were more receptive to findings from the PS analysis

23 Which X’s Belong in the PS Model? The goal is to estimate the effect of exposure E on outcome Y Confounders (Brookhart’s X 1 variables)? –Directly affect both E and Y Simple predictors (X 2 s)? –Affect Y but not E Simple selectors (X 3 s)? –Affect E but not Y

24 Example The goal is to estimate the effect of E = CABG surgery on Y = 30-day mortality following admission for a heart attack –Confounder (e.g., disease severity) –Simple predictors (e.g., home support) –Simple selectors, aka “instrumental variables” (e.g., random assignment)

Variable typeDirectly affects Belongs in which model Outcome (Y) Exposure (E)PS Subsequent Regression X1Confounder11Yes X2Predictor10?Yes X3Selector01No? 25 ? = inclusion should neither harm nor help

Discussion The “pre-processing” that occurs when sub- sampling to create “PS-balanced” comparison groups protects against bias from confounding variables Putting selector variables in the PS model will hurt accuracy (by reducing the numbers of good matches) without making the groups more comparable Subsequent regression improves accuracy 26