Methodologic challenges in the analysis of count data in radiology health services research

Bahman Sayyar Roudsari, Christopher Mack, Jeffrey G. Jarvik

Research output: Contribution to journalArticle

7 Citations (Scopus)

Abstract

Purpose: The authors explain that negative binomial (NB) and zero-inflated NB (ZINB) distributions are probably the most commonly seen distributions of outcomes in radiology health services research. Using simulation data, the authors demonstrate the potential errors in adopting an inappropriate model in the analysis of count outcomes in this field of research. Methods: A hypothetical database with 5,000 records was generated to evaluate the associations between the number of head CT studies (with Poisson, NB, and ZINB distributions) and age, gender, mechanism of injury, and injury severity. Linear, Poisson, NB, and ZINB regression models were used to analyze these hypothetical data. Results: For analysis of the number of head CT studies with an NB distribution, using linear regression resulted in biased estimates. Poisson regression resulted in artificially narrow confidence intervals. For the analyses of the number of head CT studies with a ZINB distribution, Poisson and NB regression models overestimated the association between the number of head CT studies and the predictors, while linear regression resulted in incorrect point estimates. Conclusions: With substantial increases in health care costs and the upcoming health care overhaul, pressure on radiology health services research will increase. To provide valid estimates of the predictors of utilization pattern, researchers should adopt models that appropriately deal with the skewed count outcomes, or the results might be incorrect.

Original languageEnglish (US)
Pages (from-to)575-582
Number of pages8
JournalJournal of the American College of Radiology
Volume8
Issue number8
DOIs
StatePublished - Jan 1 2011
Externally publishedYes

Fingerprint

Health Services Research
Radiology
Head
Binomial Distribution
Linear Models
Poisson Distribution
Age Distribution
Wounds and Injuries
Statistical Models
Health Care Costs
Research Personnel
Databases
Confidence Intervals
Delivery of Health Care
Pressure
Research

Keywords

  • count data analysis
  • methods
  • negative binomial analysis
  • Poisson analysis
  • Radiology health services research

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging

Cite this

Methodologic challenges in the analysis of count data in radiology health services research. / Sayyar Roudsari, Bahman; Mack, Christopher; Jarvik, Jeffrey G.

In: Journal of the American College of Radiology, Vol. 8, No. 8, 01.01.2011, p. 575-582.

Research output: Contribution to journalArticle

@article{9e0e926c8aa04444a385544928d0aa7d,
title = "Methodologic challenges in the analysis of count data in radiology health services research",
abstract = "Purpose: The authors explain that negative binomial (NB) and zero-inflated NB (ZINB) distributions are probably the most commonly seen distributions of outcomes in radiology health services research. Using simulation data, the authors demonstrate the potential errors in adopting an inappropriate model in the analysis of count outcomes in this field of research. Methods: A hypothetical database with 5,000 records was generated to evaluate the associations between the number of head CT studies (with Poisson, NB, and ZINB distributions) and age, gender, mechanism of injury, and injury severity. Linear, Poisson, NB, and ZINB regression models were used to analyze these hypothetical data. Results: For analysis of the number of head CT studies with an NB distribution, using linear regression resulted in biased estimates. Poisson regression resulted in artificially narrow confidence intervals. For the analyses of the number of head CT studies with a ZINB distribution, Poisson and NB regression models overestimated the association between the number of head CT studies and the predictors, while linear regression resulted in incorrect point estimates. Conclusions: With substantial increases in health care costs and the upcoming health care overhaul, pressure on radiology health services research will increase. To provide valid estimates of the predictors of utilization pattern, researchers should adopt models that appropriately deal with the skewed count outcomes, or the results might be incorrect.",
keywords = "count data analysis, methods, negative binomial analysis, Poisson analysis, Radiology health services research",
author = "{Sayyar Roudsari}, Bahman and Christopher Mack and Jarvik, {Jeffrey G.}",
year = "2011",
month = "1",
day = "1",
doi = "10.1016/j.jacr.2011.02.002",
language = "English (US)",
volume = "8",
pages = "575--582",
journal = "Journal of the American College of Radiology",
issn = "1558-349X",
publisher = "Elsevier BV",
number = "8",

}

TY - JOUR

T1 - Methodologic challenges in the analysis of count data in radiology health services research

AU - Sayyar Roudsari, Bahman

AU - Mack, Christopher

AU - Jarvik, Jeffrey G.

PY - 2011/1/1

Y1 - 2011/1/1

N2 - Purpose: The authors explain that negative binomial (NB) and zero-inflated NB (ZINB) distributions are probably the most commonly seen distributions of outcomes in radiology health services research. Using simulation data, the authors demonstrate the potential errors in adopting an inappropriate model in the analysis of count outcomes in this field of research. Methods: A hypothetical database with 5,000 records was generated to evaluate the associations between the number of head CT studies (with Poisson, NB, and ZINB distributions) and age, gender, mechanism of injury, and injury severity. Linear, Poisson, NB, and ZINB regression models were used to analyze these hypothetical data. Results: For analysis of the number of head CT studies with an NB distribution, using linear regression resulted in biased estimates. Poisson regression resulted in artificially narrow confidence intervals. For the analyses of the number of head CT studies with a ZINB distribution, Poisson and NB regression models overestimated the association between the number of head CT studies and the predictors, while linear regression resulted in incorrect point estimates. Conclusions: With substantial increases in health care costs and the upcoming health care overhaul, pressure on radiology health services research will increase. To provide valid estimates of the predictors of utilization pattern, researchers should adopt models that appropriately deal with the skewed count outcomes, or the results might be incorrect.

AB - Purpose: The authors explain that negative binomial (NB) and zero-inflated NB (ZINB) distributions are probably the most commonly seen distributions of outcomes in radiology health services research. Using simulation data, the authors demonstrate the potential errors in adopting an inappropriate model in the analysis of count outcomes in this field of research. Methods: A hypothetical database with 5,000 records was generated to evaluate the associations between the number of head CT studies (with Poisson, NB, and ZINB distributions) and age, gender, mechanism of injury, and injury severity. Linear, Poisson, NB, and ZINB regression models were used to analyze these hypothetical data. Results: For analysis of the number of head CT studies with an NB distribution, using linear regression resulted in biased estimates. Poisson regression resulted in artificially narrow confidence intervals. For the analyses of the number of head CT studies with a ZINB distribution, Poisson and NB regression models overestimated the association between the number of head CT studies and the predictors, while linear regression resulted in incorrect point estimates. Conclusions: With substantial increases in health care costs and the upcoming health care overhaul, pressure on radiology health services research will increase. To provide valid estimates of the predictors of utilization pattern, researchers should adopt models that appropriately deal with the skewed count outcomes, or the results might be incorrect.

KW - count data analysis

KW - methods

KW - negative binomial analysis

KW - Poisson analysis

KW - Radiology health services research

UR - http://www.scopus.com/inward/record.url?scp=84928096538&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84928096538&partnerID=8YFLogxK

U2 - 10.1016/j.jacr.2011.02.002

DO - 10.1016/j.jacr.2011.02.002

M3 - Article

VL - 8

SP - 575

EP - 582

JO - Journal of the American College of Radiology

JF - Journal of the American College of Radiology

SN - 1558-349X

IS - 8

ER -