Biostat Longitudinal Research WORKSHOP IN BIOSTATISTICS DATE: Thursday, October 3, 1996 TIME: 1:15 p.m. - 3:00 p.m. PLACE: Medical School Office Building, Room 303 TITLE: Models for Growth Curve Data SPEAKER:Alan M. Garber, MD, PhD Department of Medicine, Stanford VA Medical Center, Palo Alto National Bureau of Economic Research This workshop will be a preliminary report on research that is collaborative with Thomas MaCurdy (of the Department of Economics, Stanford; the Hoover Institution; and the National Bureau of Economic Research) and Mark McClellan (of the Departments of Economics, Health Research and Policy, and Medicine, Stanford and the National Bureau of Economic Research). The subject matter is medical expenditures for major illnesses in the US elderly population. In any given year, expenditures for Medicare are concentrated in a minority of the elderly, many of whom die. We show that the pattern of medical expenditures near the end of life varies substantially across different demographic and clinical groups and is changing over time as a result of changes in the prevalence of terminal diseases and their treatment. Our identification of high-cost end-of-life illnesses, and their predictability, is guiding our further research on the distribution of medical expenditures among the elderly. Our goal is to describe the risk of persistently high medical expenditures associated with a history of particular health problems and patterns of medical care utilization. The results of this analysis have implications for proposed Medicare reforms, such as allowing greater choice of health plans, medical savings accounts, and "risk-adjusted" health plan payments. --------------------------------------------------------------- WORKSHOP IN BIOSTATISTICS DATE: Thursday, October 10, 1996 TIME: 1:15 p.m. - 3:00 p.m. PLACE: Medical School Office Building, Room 303 TITLE: Calibration with Repeated Measures SPEAKER:Professor Samuel D. Oman Department of Statistics Hebrew University We consider linear calibration in the context of repeated measures. Suppose two scalar quantities x and Y are related by Y = a + bx + error, and that at the calibration step repeated measurements on both x and Y are available for a number of sampling units. At the prediction step a new Y measurement (possibly together with previous measurements for the new unit) is available, and it is desired to estimate the corresponding unknown x value. Examples are x = week of pregnancy and Y = fetal ultrasound bone measurement for a number of women during pregnancy monitoring; and two indices of renal function (x precise yet expensive, Y inexact yet cheaper) for cancer patients at different times during the course of chemotherapy. We allow for both the intercept a and the slope b to vary between units, resulting in a random regression coefficient model at the calibration step. This complicates the prediction step, since the unknown x affects the covariance structure of Y. We discuss point estimation and obtain a confidence region for x similar to Fieller's. Our methods are illustrated on a set of ultrasound bladder-size measurements. --------------------------------------------------------------- WORKSHOP IN BIOSTATISTICS DATE: Thursday, October 17, 1996 TIME: 1:15 p.m. - 3:00 p.m. PLACE: Medical School Office Building, Room 303 TITLE: Models for Growth Curve Data SPEAKER:Trevor Hastie Departments of Statistics and of Health Research and Policy Stanford University We develop some models for repeated measurements of Bone Mineral Density of adolescent children. Some of the goals of the study are to compare the average growth curves for different groups, and to try to make inferences about individuals (despite the fact that only 2-5 measurements are available per individual). We review some new software in Splus which allows one to fit these models in a very natural way. This is a biostatistics consulting project with Laura Bachrach (PI) and May-Choo Wang. To: WORKSHOP(Personal Dist. List) cc: PAN@OSIRIS