Short Courses: Statistical Considerations in Health Economic Evaluations (Last updated 07/21/15)
ISPOR 18th Annual Meeting
8:00  5:00, May 19, 2013
Course Description:
Adoption and diffusion of new medical treatments depend increasingly on
analysis of costs and costeffectiveness. During this course, we
discuss design issues and applications of the collection of primary
economic data as well as statistical considerations,
including the effect of distributional assumptions, univariate and
multivariable analyses of data, sample size and power calculations, and
estimation of sampling uncertainty. Participants will be guided through
applications of statistical considerations in economic analysis using
specific handson exercises.Statistical analysis will
be done by use of STATA. A trial version of the software will be
distributed for those who do not have the software. Participants who wish to have handson experience must bring their laptops.ading for this course.
Lectures
WorkshopSpecific Computer Programs and Data
Download EITHER the zip file (contains all of the following individual Stata files) OR the individual files themselves
Zip file of programs and data subdirectory
Analysis of cost
 rchapter5.dta, Stata dataset
 glmdiagnostic.do, .do file that loads the program glmdiag: glmdiag evaluates the link and family for generalized linear models
 bsmultiv.do, bootstrap program custom written for rchapter5.dta. Users can modify the program for their own datasets
 bsmvpred.ispor.dta, bootstrapped dataset constructed by use of bsmultiv.do (contains 3000 bootstrapped replications)
Sampling uncertainty (revised 05/03/13)
 iprogs.do, (revised 07/14/16) immediate form programs for calculating confidence intervals for costeffectiveness ratios (fielleri), confidenceintervals for NMB (nmbi and nmb1i), the acceptability curve (accepti) and the fraction acceptable for a specified WTP (accept1i), the VOI curve (voii), and the voi for a specified WTP (voi1i). The iprogsdoc program provides a brief overview of the programs and the syntax for running them
 ceagraphs.do, (revised 07/14/16, works with revised iprogs.do only) programs that when used in conjunction with fielleri, accepti, accept1i, and nmbi, nmb1i, and voii graphically display the point estimate and confidence limits for the costeffectiveness ratio (fiellergraph), the acceptability graph (accgraph), a line through the origin that demarcates regions of the costeffectiveness plane that represent good value for a specified WTP (accept1i), the point estimate and confidence limits for NMB defined for varying levels of WTP (nmbgraph), the point estimate and confidence limits for NMB for a specified level of WTP (nmb1graph), and the VOI curve (voigraph).. The ceagraphsdoc program provides a brief overview of the programs and the syntax for running them
 bsceaprogs.do, (last revised 04/30/13), programs for using a bootstrapped dataset to estimate confidence intervals for costeffectiveness ratios, confidence intervals for NMB , the acceptability curve, and value of information. The bsceaprogsdoc program provides a brief overview of the programs and the syntax for running them
 bsceagraphs.do: (last revised 05/03/2013), Stata .do file (works with bsceaprogs.do only). programs that when used in conjunction with bsciceri, bsaccept, bsnmb, and bsvoi graphically display the point estimate and confidence limits for the costeffectiveness ratio (bscicergraph), the acceptability graph (bsaccgraph), the point estimate and confidence limits for NMB for the NMB graph (bsnmbgraph), and the VOI graph. The bsceagraphsdoc program provides a brief overview of the programs and the syntax for running them
 bsceaprogslog.log.pdf Log file
 dataforslides.dta Stata data used in uncertainty lecture
 Bootstrap ordering problems
Sample size
 ssizeprg.do, programs for estimating sample size for costeffectiveness analysis (updated 4/1/10)
Faculty:
 Jalpa Doshi, PhD, University of Pennsylvania
 Henry Glick, PhD, University of Pennsylvania
