Causal Inference for Comprehensive Cohort Studies.

In a comprehensive cohort study of two competing treatments (say, A and B), clinically eligible individuals are first asked to enroll in a randomized trial and, if they refuse, are then asked to enroll in a parallel observational study in which they can choose treatment according to their own preference. We consider estimation of two estimands: (1) comprehensive cohort causal effect -- the difference in mean potential outcomes had all patients in the comprehensive cohort received treatment A vs. treatment B and (2) randomized trial causal effect -- the difference in mean potential outcomes had all patients enrolled in the randomized trial received treatment A vs. treatment B. For each estimand, we consider inference under various sets of unconfoundedness assumptions and construct semiparametric efficient and robust estimators. These estimators depend on nuisance functions, which we estimate, for illustrative purposes, using generalized additive models. Using the theory of sample splitting, we establish the asymptotic properties of our proposed estimators. We also illustrate our methodology using data from the Bypass Angioplasty Revascularization Investigation (BARI) randomized trial and observational registry to evaluate the effect of percutaneous transluminal coronary balloon angioplasty versus coronary artery bypass grafting on 5-year mortality. To evaluate the finite sample performance of our estimators, we use the BARI dataset as the basis of a realistic simulation study.

[1]  L. Fielding,et al.  Patients who are eligible but not randomised should be included as additional comparative arm in study , 1999, BMJ.

[2]  James M. Robins,et al.  MINIMAX ESTIMATION OF A FUNCTIONAL ON A STRUCTURED , 2016 .

[3]  J. Robins,et al.  Estimating causal effects from epidemiological data , 2006, Journal of Epidemiology and Community Health.

[4]  M Schumacher,et al.  Analysis of randomized and nonrandomized patients in clinical trials using the comprehensive cohort follow-up study design. , 1992, Controlled clinical trials.

[5]  James M. Robins,et al.  Unified Methods for Censored Longitudinal Data and Causality , 2003 .

[6]  B. Caffo,et al.  A Unifying Framework for Marginalised Random‐Intercept Models of Correlated Binary Outcomes , 2014, International statistical review = Revue internationale de statistique.

[7]  James M. Robins,et al.  Observational Studies Analyzed Like Randomized Experiments: An Application to Postmenopausal Hormone Therapy and Coronary Heart Disease , 2008, Epidemiology.

[8]  P. Vestergaard,et al.  Hormone Replacement Therapy Dissociates Fat Mass and Bone Mass, and Tends to Reduce Weight Gain in Early Postmenopausal Women: A Randomized Controlled 5‐Year Clinical Trial of the Danish Osteoporosis Prevention Study , 2003, Journal of bone and mineral research : the official journal of the American Society for Bone and Mineral Research.

[9]  M. Dewey,et al.  Assessing effectiveness of treatment of depression in primary care. Partially randomised preference trial. , 2000, The British journal of psychiatry : the journal of mental science.

[10]  H. Barnhart,et al.  Angioplasty or surgery for multivessel coronary artery disease: comparison of eligible registry and randomized patients in the EAST trial and influence of treatment selection on outcomes. Emory Angioplasty versus Surgery Trial Investigators. , 1997, The American journal of cardiology.

[11]  R. Califf,et al.  Coronary revascularization in diabetic patients: a comparison of the randomized and observational components of the Aypass Angioplasty Revascularization Investigation (BARI). , 1999, Circulation.

[12]  R. Califf,et al.  Predictors of mortality and mortality from cardiac causes in the bypass angioplasty revascularization investigation (BARI) randomized trial and registry. For the BARI Investigators. , 2000, Circulation.

[13]  K. Do,et al.  Efficient and Adaptive Estimation for Semiparametric Models. , 1994 .

[14]  Mark J. van der Laan,et al.  Asymptotic Theory for Cross-validated Targeted Maximum Likelihood Estimation , 2010 .

[15]  W. Newey,et al.  Semiparametric Efficiency Bounds , 1990 .

[16]  J. Robins,et al.  Adjusting for Nonignorable Drop-Out Using Semiparametric Nonresponse Models , 1999 .

[17]  Martin Schumacher,et al.  Evidence from nonrandomized studies: a case study on the estimation of causal effects. , 2008, American journal of epidemiology.

[18]  Mark J. van der Laan,et al.  Higher-order Targeted Minimum Loss-based Estimation , 2014 .

[19]  Sarah E. Robertson,et al.  Transporting inferences from a randomized trial to a new target population , 2018 .

[20]  S. Kerry,et al.  Routine referral for radiography of patients presenting with low back pain: is patients' outcome influenced by GPs' referral for plain radiography? , 2000, Health technology assessment.

[21]  J. Pfanzagl,et al.  CONTRIBUTIONS TO A GENERAL ASYMPTOTIC STATISTICAL THEORY , 1982 .

[22]  J. Robins,et al.  Doubly Robust Estimation in Missing Data and Causal Inference Models , 2005, Biometrics.

[23]  R. Frye,et al.  Long-Term Clinical Outcome in the Bypass Angioplasty Revascularization Investigation Registry Comparison With the Randomized Trial , 2000 .

[24]  P. van den Broek,et al.  Generalizability of trial results based on randomized versus nonrandomized allocation of OME infants to ventilation tubes or watchful waiting. , 2001, Journal of clinical epidemiology.

[25]  J. Robins,et al.  Double/Debiased Machine Learning for Treatment and Causal Parameters , 2016, 1608.00060.

[26]  Aad van der Vaart,et al.  Higher order influence functions and minimax estimation of nonlinear functionals , 2008, 0805.3040.

[27]  J. Mark,et al.  Targeted estimation of nuisance parameters to obtain valid statistical inference. , 2014 .

[28]  M. Olschewski,et al.  Comprehensive Cohort Study: An Alternative to Randomized Consent Design in a Breast Preservation Trial , 1985, Methods of Information in Medicine.

[29]  P. Bower,et al.  Randomised controlled trial of non-directive counselling, cognitive-behaviour therapy and usual general practitioner care in the management of depression as well as mixed anxiety and depression in primary care. , 2000, Health technology assessment.

[30]  M. J. van der Laan.  Targeted Estimation of Nuisance Parameters to Obtain Valid Statistical Inference , 2014, The international journal of biostatistics.

[31]  M. J. Laan,et al.  Doubly robust nonparametric inference on the average treatment effect , 2017, Biometrika.

[32]  Sivaraman Balakrishnan,et al.  Sharp instruments for classifying compliers and generalizing causal effects , 2018, The Annals of Statistics.

[33]  Marie Davidian,et al.  Doubly robust estimation of causal effects. , 2011, American journal of epidemiology.

[34]  K. Nicolaides,et al.  Comparison of chorionic villus sampling and amniocentesis for fetal karyotyping at 10-13 weeks' gestation , 1994, The Lancet.

[35]  R. Califf,et al.  Coronary revascularization in diabetic patients: a comparison of the randomized and observational components of the Aypass Angioplasty Revascularization Investigation (BARI). , 1999, Circulation.

[36]  Robert L. Frye,et al.  Comparison of coronary bypass surgery with angioplasty in patients with multivessel disease. , 1996, The New England journal of medicine.

[37]  M Schumacher,et al.  Randomized and non-randomized patients in clinical trials: experiences with comprehensive cohort studies. , 1995, Statistics in medicine.

[38]  A. Tsiatis.  Semiparametric Theory and Missing Data , 2006 .

[39]  J. Horowitz.  Semiparametric and Nonparametric Methods in Econometrics , 2007 .

[40]  James T. Willerson,et al.  Appointment of New Associate Editor for Circulation , 1996 .

[41]  Tyler J. VanderWeele,et al.  Concerning the consistency assumption in causal inference. , 2009, Epidemiology.

[42]  S. Marcus,et al.  Assessing non-consent bias with parallel randomized and nonrandomized clinical trials. , 1997, Journal of clinical epidemiology.

[43]  R. Henshaw,et al.  Comparison of medical abortion with surgical vacuum aspiration: women's preferences and acceptability of treatment. , 1993, BMJ.