Choosing models for health care cost analyses: issues of nonlinearity and endogeneity.

OBJECTIVE To compare methods of analyzing endogenous treatment effect models for nonlinear outcomes and illustrate the impact of model specification on estimates of treatment effects such as health care costs. DATA SOURCES Secondary data on cost and utilization for inpatients hospitalized in five Veterans Affairs acute care facilities in 2005-2006. STUDY DESIGN We compare results from analyses with full information maximum simulated likelihood (FIMSL); control function (CF) approaches employing different types and functional forms for the residuals, including the special case of two-stage residual inclusion; and two-stage least squares (2SLS). As an example, we examine the effect of an inpatient palliative care (PC) consultation on direct costs of care per day. DATA COLLECTION/EXTRACTION METHODS We analyzed data for 3,389 inpatients with one or more life-limiting diseases. PRINCIPAL FINDINGS The distribution of average treatment effects on the treated and local average treatment effects of a PC consultation depended on model specification. CF and FIMSL estimates were more similar to each other than to 2SLS estimates. CF estimates were sensitive to choice and functional form of residual. CONCLUSIONS When modeling cost or other nonlinear data with endogeneity, one should be aware of the impact of model specification and treatment effect choice on results.

[1]  Enno Mammen,et al.  ANALYZING THE RANDOM COEFFICIENT MODEL NONPARAMETRICALLY , 2009, Econometric Theory.

[2]  W. Manning,et al.  Estimating Log Models: To Transform or Not to Transform? , 1999, Journal of health economics.

[3]  J. Terza,et al.  Assessing the impact of drug use on hospital costs. , 2009, Health services research.

[4]  J. Gill,et al.  Generalized Linear Models: A Unified Approach , 2000 .

[5]  Partha Deb,et al.  Specification and Simulated Likelihood Estimation of a Non-Normal Treatment-Outcome Model with Selection: Application to Health Care Utilization , 2006 .

[6]  M Alan Brookhart,et al.  Instrumental variables II: instrumental variable application-in 25 variations, the physician prescribing preference generally was strong and reduced covariate imbalance. , 2009, Journal of clinical epidemiology.

[7]  Milton C Weinstein,et al.  Comparative effectiveness and health care spending--implications for reform. , 2010, The New England journal of medicine.

[8]  N. Duan.  Smearing Estimate: A Nonparametric Retransformation Method , 1983 .

[9]  P. Deb,et al.  Hospital-based palliative care consultation: effects on hospital cost. , 2010, Journal of palliative medicine.

[10]  Melinda Beeuwkes Buntin,et al.  Too much ado about two-part models and transformation? Comparing methods of modeling Medicare expenditures. , 2004, Journal of health economics.

[11]  Willard G. Manning,et al.  Issues for the Next Generation of Health Care Cost Analyses , 2009, Medical care.

[12]  J. Trogdon,et al.  The economic implications of influenza vaccination for adults with asthma. , 2010, American journal of preventive medicine.

[13]  Paul J Rathouz,et al.  Two-stage residual inclusion estimation: addressing endogeneity in health econometric modeling. , 2008, Journal of health economics.

[14]  Jack Hadley,et al.  Comparative effectiveness of prostate cancer treatments: evaluating statistical adjustments for confounding in observational data. , 2010, Journal of the National Cancer Institute.

[15]  Joseph V. Terza,et al.  Estimating count data models with endogenous switching: Sample selection and endogenous treatment effects , 1998 .

[16]  C. Morris,et al.  A Comparison of Alternative Models for the Demand for Medical Care , 1983 .

[17]  Steven D. Pizer,et al.  An intuitive review of methods for observational studies of comparative effectiveness , 2009, Health Services and Outcomes Research Methodology.

[18]  Dylan S Small,et al.  Two‐stage instrumental variable methods for estimating the causal odds ratio: Analysis of bias , 2011, Statistics in medicine.

[19]  Yijian Huang,et al.  Cost analysis with censored data. , 2009, Medical care.

[20]  D. Meier,et al.  Cost savings associated with US hospital palliative care consultation programs. , 2008, Archives of internal medicine.

[21]  Elliott S Fisher,et al.  Analysis of observational studies in the presence of treatment selection bias: effects of invasive cardiac management on AMI survival using propensity score and instrumental variable methods. , 2007, JAMA.

[22]  W. Manning,et al.  The logged dependent variable, heteroscedasticity, and the retransformation problem. , 1998, Journal of health economics.

[23]  James J. Heckman,et al.  Alternative methods for evaluating the impact of interventions: An overview , 1985 .

[24]  Joshua D. Angrist,et al.  Identification of Causal Effects Using Instrumental Variables , 1993 .

[25]  D. Pierce,et al.  Residuals in Generalized Linear Models , 1986 .

[26]  M C Hornbrook,et al.  Modeling risk using generalized linear models. , 1999, Journal of health economics.

[27]  C. Wilcox,et al.  The economics of upper gastrointestinal bleeding in a US managed-care setting: a retrospective, claims-based analysis , 2010, Journal of medical economics.

[28]  Partha Deb,et al.  Maximum Simulated Likelihood Estimation of a Negative Binomial Regression Model with Multinomial Endogenous Treatment , 2006 .

[29]  Partha Deb,et al.  Medicare Spending and Outcomes After Postacute Care for Stroke and Hip Fracture , 2010, Medical care.

[30]  A James O'Malley,et al.  Estimating cost-offsets of new medications: Use of new antipsychotics and mental health costs for schizophrenia , 2011, Statistics in medicine.

[31]  W. Newey,et al.  Nonparametric estimation of triangular simultaneous equations models , 1999 .

[32]  A. Moskowitz,et al.  Process of Care Events in Transplantation: Effects on the Cost of Hospitalization , 2010, American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons.

[33]  Han Hong,et al.  Flexible Estimation of Treatment Effect Parameters , 2011 .

[34]  S. Lee.  Endogeneity in Quantile Regression Models: A Control Function Approach , 2004 .

[35]  T. Wickizer,et al.  Evaluation of the Washington State Screening, Brief Intervention, and Referral to Treatment Project: Cost Outcomes for Medicaid Patients Screened in Hospital Emergency Departments , 2010, Medical care.

[36]  P. Haan,et al.  Estimation of Multinomial Logit Models with Unobserved Heterogeneity using Maximum Simulated Likelihood , 2006 .