The modeling of medical expenditure data from a longitudinal survey using the generalized method of moments (GMM) approach

Zachary Hass, Michael Levine, Laura P. Sands, Jeffrey Ting, Huiping Xu

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Medical expenditure data analysis has recently become an important problem in biostatistics. These data typically have a number of features making their analysis rather difficult. Commonly, they are heavily right-skewed, contain a large percentage of zeros, and often exhibit large numbers of missing observations because of death and/or the lack of follow-up. They are also commonly obtained from records that are linked to large longitudinal data surveys. In this manuscript, we suggest a novel approach to modeling these data through the use of generalized method of moments estimation procedure combined with appropriate weights that account for both dropout due to death and the probability of being sampled from among the National Long Term Care Survey (NLTCS) subjects. This approach seems particularly appropriate because of the large number of subjects relative to the length of observation period (in months). We also use a simulation study to compare our proposed approach with and without the use of weights. The proposed model is applied to medical expenditure data obtained from the 2004-2005 NLTCS-linked Medicare database. The results suggest that the amount of medical expenditures incurred is strongly associated with higher number of activities of daily living (ADL) disabilities and self-reports of unmet need for help with ADL disabilities.

Original languageEnglish (US)
JournalStatistics in Medicine
DOIs
StateAccepted/In press - 2016

Fingerprint

Generalized Method of Moments
Health Expenditures
Longitudinal Studies
Long-Term Care
Activities of Daily Living
Disability
Modeling
Biostatistics
Weights and Measures
Medicare
Moment Estimation
Missing Observations
Self Report
Data Modeling
Drop out
Large Data
Longitudinal Data
Term
Observation
Databases

Keywords

  • Generalized method of moments (GMM)
  • Inverse probability weighting-generalized estimating equations (IPW-GEE)
  • Longitudinal data survey
  • Medical expenditure data
  • Modified sandwich estimator

ASJC Scopus subject areas

  • Epidemiology
  • Statistics and Probability

Cite this

The modeling of medical expenditure data from a longitudinal survey using the generalized method of moments (GMM) approach. / Hass, Zachary; Levine, Michael; Sands, Laura P.; Ting, Jeffrey; Xu, Huiping.

In: Statistics in Medicine, 2016.

Research output: Contribution to journalArticle

@article{b504aebed6944e19a1ba795ae5399fc9,
title = "The modeling of medical expenditure data from a longitudinal survey using the generalized method of moments (GMM) approach",
abstract = "Medical expenditure data analysis has recently become an important problem in biostatistics. These data typically have a number of features making their analysis rather difficult. Commonly, they are heavily right-skewed, contain a large percentage of zeros, and often exhibit large numbers of missing observations because of death and/or the lack of follow-up. They are also commonly obtained from records that are linked to large longitudinal data surveys. In this manuscript, we suggest a novel approach to modeling these data through the use of generalized method of moments estimation procedure combined with appropriate weights that account for both dropout due to death and the probability of being sampled from among the National Long Term Care Survey (NLTCS) subjects. This approach seems particularly appropriate because of the large number of subjects relative to the length of observation period (in months). We also use a simulation study to compare our proposed approach with and without the use of weights. The proposed model is applied to medical expenditure data obtained from the 2004-2005 NLTCS-linked Medicare database. The results suggest that the amount of medical expenditures incurred is strongly associated with higher number of activities of daily living (ADL) disabilities and self-reports of unmet need for help with ADL disabilities.",
keywords = "Generalized method of moments (GMM), Inverse probability weighting-generalized estimating equations (IPW-GEE), Longitudinal data survey, Medical expenditure data, Modified sandwich estimator",
author = "Zachary Hass and Michael Levine and Sands, {Laura P.} and Jeffrey Ting and Huiping Xu",
year = "2016",
doi = "10.1002/sim.6878",
language = "English (US)",
journal = "Statistics in Medicine",
issn = "0277-6715",
publisher = "John Wiley and Sons Ltd",

}

TY - JOUR

T1 - The modeling of medical expenditure data from a longitudinal survey using the generalized method of moments (GMM) approach

AU - Hass, Zachary

AU - Levine, Michael

AU - Sands, Laura P.

AU - Ting, Jeffrey

AU - Xu, Huiping

PY - 2016

Y1 - 2016

N2 - Medical expenditure data analysis has recently become an important problem in biostatistics. These data typically have a number of features making their analysis rather difficult. Commonly, they are heavily right-skewed, contain a large percentage of zeros, and often exhibit large numbers of missing observations because of death and/or the lack of follow-up. They are also commonly obtained from records that are linked to large longitudinal data surveys. In this manuscript, we suggest a novel approach to modeling these data through the use of generalized method of moments estimation procedure combined with appropriate weights that account for both dropout due to death and the probability of being sampled from among the National Long Term Care Survey (NLTCS) subjects. This approach seems particularly appropriate because of the large number of subjects relative to the length of observation period (in months). We also use a simulation study to compare our proposed approach with and without the use of weights. The proposed model is applied to medical expenditure data obtained from the 2004-2005 NLTCS-linked Medicare database. The results suggest that the amount of medical expenditures incurred is strongly associated with higher number of activities of daily living (ADL) disabilities and self-reports of unmet need for help with ADL disabilities.

AB - Medical expenditure data analysis has recently become an important problem in biostatistics. These data typically have a number of features making their analysis rather difficult. Commonly, they are heavily right-skewed, contain a large percentage of zeros, and often exhibit large numbers of missing observations because of death and/or the lack of follow-up. They are also commonly obtained from records that are linked to large longitudinal data surveys. In this manuscript, we suggest a novel approach to modeling these data through the use of generalized method of moments estimation procedure combined with appropriate weights that account for both dropout due to death and the probability of being sampled from among the National Long Term Care Survey (NLTCS) subjects. This approach seems particularly appropriate because of the large number of subjects relative to the length of observation period (in months). We also use a simulation study to compare our proposed approach with and without the use of weights. The proposed model is applied to medical expenditure data obtained from the 2004-2005 NLTCS-linked Medicare database. The results suggest that the amount of medical expenditures incurred is strongly associated with higher number of activities of daily living (ADL) disabilities and self-reports of unmet need for help with ADL disabilities.

KW - Generalized method of moments (GMM)

KW - Inverse probability weighting-generalized estimating equations (IPW-GEE)

KW - Longitudinal data survey

KW - Medical expenditure data

KW - Modified sandwich estimator

UR - http://www.scopus.com/inward/record.url?scp=84957991094&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84957991094&partnerID=8YFLogxK

U2 - 10.1002/sim.6878

DO - 10.1002/sim.6878

M3 - Article

C2 - 26823052

AN - SCOPUS:84957991094

JO - Statistics in Medicine

JF - Statistics in Medicine

SN - 0277-6715

ER -