Alistair James O'Malley1, Bruce E Landon2,3, Lawrence A Zaborski2, Eric T Roberts4, Hazar H Khidir5, Peter B Smulowitz6,7, John Michael McWilliams2,8. 1. Department of Biomedical Data Science and The Dartmouth Institute for Health Policy and Clinical Practice, Geisel School of Medicine at Dartmouth, Lebanon, New Hampshire, USA. 2. Department of Health Care Policy, Harvard Medical School, Boston, Massachusetts, USA. 3. Division of General Medicine, Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA. 4. Department of Health Policy and Management, University of Pittsburgh, Pittsburgh, Pennsylvania, USA. 5. National Clinician Scholars Program, Yale University School of Medicine, New Haven, Connecticut, USA. 6. Department of Emergency Medicine, University of Massachusetts Medical School, Worcester, Massachusetts, USA. 7. Emergency Department, Milford Regional Medical Center, Milford, Massachusetts, USA. 8. Department of Internal Medicine, Brigham and Women's Hospital, Boston, Massachusetts, USA.
Abstract
OBJECTIVE: To examine whether the correlation between a provider's effect on one population of patients and the same provider's effect on another population is underestimated if the effects for each population are estimated separately as opposed to being jointly modeled as random effects, and to characterize how the impact of the estimation procedure varies with sample size. DATA SOURCES: Medicare claims and enrollment data on emergency department (ED) visits, including patient characteristics, the patient's hospitalization status, and identification of the doctor responsible for the decision to hospitalize the patient. STUDY DESIGN: We used a three-pronged investigation consisting of analytical derivation, simulation experiments, and analysis of administrative data to demonstrate the fallibility of stratified estimation. Under each investigation method, results are compared between the joint modeling approach to those based on stratified analyses. DATA COLLECTION/EXTRACTION METHODS: We used data on ED visits from administrative claims from traditional (fee-for-service) Medicare from January 2012 through September 2015. PRINCIPAL FINDINGS: The simulation analysis demonstrates that the joint modeling approach is generally close to unbiased, whereas the stratified approach can be severely biased in small samples, a consequence of joint modeling benefitting from bivariate shrinkage and the stratified approach being compromised by measurement error. In the administrative data analyses, the estimated correlation of doctor admission tendencies between female and male patients was estimated to be 0.98 under the joint model but only 0.38 using stratified estimation. The analogous correlations for White and non-White patients are 0.99 and 0.28 and for Medicaid dual-eligible and non-dual-eligible patients are 0.99 and 0.31, respectively. These results are consistent with the analytical derivations. CONCLUSIONS: Joint modeling targets the parameter of primary interest. In the case of population correlations, it yields estimates that are substantially less biased and higher in magnitude than naive estimators that post-process the estimates obtained from stratified models.
OBJECTIVE: To examine whether the correlation between a provider's effect on one population of patients and the same provider's effect on another population is underestimated if the effects for each population are estimated separately as opposed to being jointly modeled as random effects, and to characterize how the impact of the estimation procedure varies with sample size. DATA SOURCES: Medicare claims and enrollment data on emergency department (ED) visits, including patient characteristics, the patient's hospitalization status, and identification of the doctor responsible for the decision to hospitalize the patient. STUDY DESIGN: We used a three-pronged investigation consisting of analytical derivation, simulation experiments, and analysis of administrative data to demonstrate the fallibility of stratified estimation. Under each investigation method, results are compared between the joint modeling approach to those based on stratified analyses. DATA COLLECTION/EXTRACTION METHODS: We used data on ED visits from administrative claims from traditional (fee-for-service) Medicare from January 2012 through September 2015. PRINCIPAL FINDINGS: The simulation analysis demonstrates that the joint modeling approach is generally close to unbiased, whereas the stratified approach can be severely biased in small samples, a consequence of joint modeling benefitting from bivariate shrinkage and the stratified approach being compromised by measurement error. In the administrative data analyses, the estimated correlation of doctor admission tendencies between female and male patients was estimated to be 0.98 under the joint model but only 0.38 using stratified estimation. The analogous correlations for White and non-White patients are 0.99 and 0.28 and for Medicaid dual-eligible and non-dual-eligible patients are 0.99 and 0.31, respectively. These results are consistent with the analytical derivations. CONCLUSIONS: Joint modeling targets the parameter of primary interest. In the case of population correlations, it yields estimates that are substantially less biased and higher in magnitude than naive estimators that post-process the estimates obtained from stratified models.
Authors: Jianhui Hu; Jack Jordan; Ilan Rubinfeld; Michelle Schreiber; Brian Waterman; David Nerenz Journal: Am J Med Qual Date: 2016-12-21 Impact factor: 1.852
Authors: Harlan M Krumholz; Zhenqiu Lin; Patricia S Keenan; Jersey Chen; Joseph S Ross; Elizabeth E Drye; Susannah M Bernheim; Yun Wang; Elizabeth H Bradley; Lein F Han; Sharon-Lise T Normand Journal: JAMA Date: 2013-02-13 Impact factor: 56.272
Authors: Eric T Roberts; Alan M Zaslavsky; Michael L Barnett; Bruce E Landon; Lin Ding; J Michael McWilliams Journal: JAMA Intern Med Date: 2018-11-01 Impact factor: 21.873
Authors: Alistair James O'Malley; Bruce E Landon; Lawrence A Zaborski; Eric T Roberts; Hazar H Khidir; Peter B Smulowitz; John Michael McWilliams Journal: Health Serv Res Date: 2021-11-03 Impact factor: 3.402
Authors: Orestis A Panagiotou; Kirsten R Voorhies; Laura M Keohane; Daeho Kim; Deepak Adhikari; Amit Kumar; Maricruz Rivera-Hernandez; Momotazur Rahman; Pedro Gozalo; Roee Gutman; Vincent Mor; Amal N Trivedi Journal: JAMA Netw Open Date: 2021-02-01
Authors: Thomas R Radomski; Robert Feldman; Yan Huang; Florentina E Sileanu; Carolyn T Thorpe; Joshua M Thorpe; Michael J Fine; Walid F Gellad Journal: JAMA Netw Open Date: 2020-09-01
Authors: Alistair James O'Malley; Bruce E Landon; Lawrence A Zaborski; Eric T Roberts; Hazar H Khidir; Peter B Smulowitz; John Michael McWilliams Journal: Health Serv Res Date: 2021-11-03 Impact factor: 3.402