MODEL SELECTION AND AVERAGING OF HEALTH COSTS IN EPISODE TREATMENT GROUPS

Episode Treatment Groups (ETGs) classify related services into medically relevant and distinct units describing an episode of care. Proper model selection for those ETG based costs is essential to adequately price and manage health insurance risks. The optimal loss model (or model probabilities) can vary depending on the disease. We compare four potential models (lognormal, gamma, log-skew-t, and Lomax) using four different metrics (AIC and BIC weights, Random Forest feature classification, and Bayesian model averaging) on 320 episode treatment groups. Using the data from a major health insurer, which consists of more than 33 million observations from 9 million claimants, we compare the various methods on both speed and precision, and also examine the wide range of selected models for the different ETGs. Several case studies are provided for illustration. It is found that Random Forest feature selection is computationally efficient and sufficiently accurate, hence being preferred in this large data set. When feasible (on smaller data sets), Bayesian model averaging is preferred because of the posterior model probabilities.

[1]  J. Kuha.  AIC and BIC , 2004 .

[2]  M. C. Jones,et al.  A skew extension of the t‐distribution, with applications , 2003 .

[3]  Stuart A. Klugman,et al.  Loss Models: From Data to Decisions , 1998 .

[4]  Ian Duncan,et al.  A prediction model for targeting low-cost, high-risk members of managed care organizations. , 2003, The American journal of managed care.

[5]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[6]  Trevor Hastie,et al.  The Elements of Statistical Learning , 2001 .

[7]  M. Steel,et al.  A new class of skewed multivariate distributions with applications to regression analysis , 2007 .

[8]  Andrew Gelman,et al.  Handbook of Markov Chain Monte Carlo , 2011 .

[9]  Irina Miroshnik,et al.  HOW TO BE A BAYESIAN IN SAS : MODEL SELECTION UNCERTAINTY IN PROC LOGISTIC AND PROC GENMOD , 2000 .

[10]  Edward W. Frees,et al.  Predicting the Frequency and Amount of Health Care Expenditures , 2011 .

[11]  Radford M. Neal.  MCMC Using Hamiltonian Dynamics , 2011, 1206.1901.

[12]  Robert Tibshirani,et al.  The Elements of Statistical Learning: Data Mining, Inference, and Prediction, 2nd Edition , 2001, Springer Series in Statistics.

[13]  Dove Hg,et al.  Episode Treatment Groups (ETGs): a patient classification system for measuring outcomes performance by episode of illness. , 2000 .

[14]  Brian Hartman,et al.  Model Selection and Averaging in Financial Risk Management , 2013 .

[15]  David R. Anderson,et al.  Model selection and multimodel inference : a practical information-theoretic approach , 2003 .

[16]  Henry G. Dove,et al.  Beyond Severity of Illness: Evaluating Differences in Patient Intensity and Complexity for Valid Assessment of Medical Practice Pattern Variation , 2005 .

[17]  B. Carlin,et al.  Bayesian Model Choice Via Markov Chain Monte Carlo Methods , 1995 .

[18]  Robert S Gold,et al.  Risk‐Adjusted Indices for Measuring the Quality of Inpatient Care , 2010, Quality management in health care.

[19]  Matthew P. Wand,et al.  Fully simplified multivariate normal updates in non-conjugate variational message passing , 2014, J. Mach. Learn. Res..

[20]  Peter S. Fader,et al.  Model Selection Using Database Characteristics: Developing a Classification Tree for Longitudinal Incidence Data , 2013, Mark. Sci..

[21]  Feifei Zhang.  Loss Models: From Data to Decisions, 4th Edition, by Stuart A. Klugman, Harry H. Panjer and Gordon E. Willmot: Wiley Series in Probability and Statistics, 2012, 512pp. ISBN: 978-1-118-31532-3 , 2013, Annals of Actuarial Science.

[22]  R. Leary,et al.  All-payer severity-adjusted diagnosis-related groups: a uniform method to severity-adjust discharge data. , 1997, Topics in health information management.

[23]  Peter Congdon,et al.  Bayesian model choice based on Monte Carlo estimates of posterior model probabilities , 2006, Comput. Stat. Data Anal..

[24]  S. Kotz,et al.  Statistical Size Distributions in Economics and Actuarial Sciences , 2003 .

[25]  Andrew Gelman,et al.  The No-U-turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo , 2011, J. Mach. Learn. Res..

[26]  B. Schmeiser,et al.  Performance of the Gibbs, Hit-and-Run, and Metropolis Samplers , 1993 .

[27]  Martin Eling,et al.  Fitting insurance claims to skewed distributions: Are the skew-normal and skew-student good models? , 2012 .

[28]  Andy Liaw,et al.  Classification and Regression by randomForest , 2007 .

[29]  Hirotugu Akaike,et al.  On the Likelihood of a Time Series Model , 1978 .