STATISTICAL METHODS FOR OPHTHALMOLOGIC AND CLUSTER DATA

Summary

Principal Investigator: B Rosner
Affiliation: Harvard University
Country: USA
Abstract: DESCRIPTION: (Applicant's Abstract) The analysis of ophthalmological data pose unique challenges for the data analyst because the fundamental unit of analysis is not well defined. If the eye is used as the unit of analysis, then one must account for the correlation between fellow eyes in performing the analysis; otherwise one will underestimate p- values and widths of confidence intervals. There has been much interest in he past 15 years in the development of techniques for addressing this issue including GEE methods (Liang and Zeger, 1986) and generalized linear models (Rosner, 1984). Most of these methods are concerned with parametric statistical models such as linear or logistic regression. One gap in the literature that we intend to fill in this proposal is the incorporation of clustering effects for standard nonparametric tests. Many types of ophthalmologic data are not normally distributed (e.g. Humphrey visual field data), and require nonparametric methods for their analysis. If one wishes to use the eye as the unit of analysis, then it is important to incorporate clustering effects into these methods that take the correlation between fellow eyes into account. Many ophthalmologists prefer to use the person as the unit of analysis (e.g. using visual function in the better eye) rather than the eye. However, Olkin and Viana (1995) showed that the distribution of visual function in the better eye as a function of a person-specific covariate (e.g. age) is not normally distributed, and propose special methods for this type of analysis. We propose to extend this work to the case of eye specific covariates and the case of several covariates in the same model. Another issue is that some ophthalmic conditions (e.g. cataract) have subclassifications with possibly different etiologies (e.g. nuclear, cortical, PSC cataract). However, the important public health question is: what is the overall probability of any type of cataract as a function of risk factors (e.g., antioxidant use). We intent to generalize the standard logistic model, to be compound logistic model which accounts for the possibly differential risk factor profiles. Furthermore, there has been much work on censored survival data where the failure times occur only at specific visits. We intend to incorporate clustering effects in interval censored data. A final goal of our proposal is to perform an analysis of the natural course of retinitis pigmentosa over a 12 year period. This will allow estimation of the rates of decline of the ERG and visual field over a long period of time as well as provide estimates for an individual patient of the number of years of useful vision they can expect to have.
Funding Period: 1998-08-01 - 2002-07-31
more information: NIH RePORT

Top Publications

  1. pmc Assessing discrimination of risk prediction rules in a clustered data setting
    Bernard Rosner
    Channing Division of Network Medicine, Brigham and Women s Hospital Harvard Medical School, 181 Longwood Avenue, Boston, MA 02115, USA
    Lifetime Data Anal 19:242-56. 2013
  2. pmc Regression methods when the eye is the unit of analysis
    Robert J Glynn
    Division of Preventive Medicine and Channing Lab, Department of Medicine, Brigham and Women s Hospital, Boston, MA 02215, USA
    Ophthalmic Epidemiol 19:159-65. 2012
  3. pmc Joint analysis of current status and marker data: an extension of a bivariate threshold model
    Xingwei Tong
    Beijing Normal University
    Int J Biostat 4:Article 21. 2008
  4. pmc Power and sample size estimation for the clustered wilcoxon test
    Bernard Rosner
    Channing Laboratory, Harvard Medical School, Boston, Massachusetts, 02115 USA
    Biometrics 67:646-53. 2011
  5. pmc Threshold regression for survival data with time-varying covariates
    Mei Ling Ting Lee
    Department of Epidemiology and Biostatistics, University of Maryland, College Park, MD, USA
    Stat Med 29:896-905. 2010
  6. pmc Evaluation of risk factors for cataract types in a competing risks framework
    Robert J Glynn
    Division of Preventive Medicine, Brigham and Women s Hospital, Harvard Medical School, Boston, Massachusetts 02215, USA
    Ophthalmic Epidemiol 16:98-106. 2009
  7. doi Power and sample size estimation for the Wilcoxon rank sum test with application to comparisons of C statistics from alternative prediction models
    B Rosner
    Channing Laboratory, Harvard Medical School, Boston, Massachusetts 02115, USA
    Biometrics 65:188-97. 2009
  8. ncbi A nonparametric test for observational non-normally distributed ophthalmic data with eye-specific exposures and outcomes
    Bernard Rosner
    Channing Laboratory, Harvard Medical School, Boston, Massachusetts 02115, USA
    Ophthalmic Epidemiol 14:243-50. 2007
  9. ncbi Extension of the rank sum test for clustered data: two-group comparisons with group membership defined at the subunit level
    Bernard Rosner
    Channing Laboratory, Harvard Medical School, Boston, Massachusetts 02115, USA
    Biometrics 62:1251-9. 2006
  10. ncbi The Wilcoxon signed rank test for paired comparisons of clustered data
    Bernard Rosner
    Channing Laboratory, Harvard Medical School, Boston, Massachusetts 02115, USA
    Biometrics 62:185-92. 2006

Scientific Experts

Detail Information

Publications12

  1. pmc Assessing discrimination of risk prediction rules in a clustered data setting
    Bernard Rosner
    Channing Division of Network Medicine, Brigham and Women s Hospital Harvard Medical School, 181 Longwood Avenue, Boston, MA 02115, USA
    Lifetime Data Anal 19:242-56. 2013
    ..Both data analyses based on progression of AMD and simulation studies show reasonable accuracy of this extended Mann-Whitney U test to assess discrimination of eye-specific risk prediction rules...
  2. pmc Regression methods when the eye is the unit of analysis
    Robert J Glynn
    Division of Preventive Medicine and Channing Lab, Department of Medicine, Brigham and Women s Hospital, Boston, MA 02215, USA
    Ophthalmic Epidemiol 19:159-65. 2012
    ..Advances have occurred in the development and accessibility of analytic approaches to evaluate determinants of eye-specific outcomes including information from both eyes of some subjects...
  3. pmc Joint analysis of current status and marker data: an extension of a bivariate threshold model
    Xingwei Tong
    Beijing Normal University
    Int J Biostat 4:Article 21. 2008
    ..1998) to the case when only current status data are available. We develop maximum likelihood estimation procedures and provide simulation studies. We apply our methods to a motivating example involving liver tumors in mice...
  4. pmc Power and sample size estimation for the clustered wilcoxon test
    Bernard Rosner
    Channing Laboratory, Harvard Medical School, Boston, Massachusetts, 02115 USA
    Biometrics 67:646-53. 2011
    ..These methods are illustrated with examples from randomized trials in ophthalmology. Enhanced power is achieved with use of the subunit as the unit of analysis instead of the cluster using the ordinary Wilcoxon rank sum test...
  5. pmc Threshold regression for survival data with time-varying covariates
    Mei Ling Ting Lee
    Department of Epidemiology and Biostatistics, University of Maryland, College Park, MD, USA
    Stat Med 29:896-905. 2010
    ..The procedure is also shown to be consistent with the use of an alternative time scale. Finally, we present the connection of the procedure to the concept of a collapsible survival model...
  6. pmc Evaluation of risk factors for cataract types in a competing risks framework
    Robert J Glynn
    Division of Preventive Medicine, Brigham and Women s Hospital, Harvard Medical School, Boston, Massachusetts 02215, USA
    Ophthalmic Epidemiol 16:98-106. 2009
    ..One can compare the impact of risk factors on different types of cataract with methods of competing risk survival analysis that account for tied events...
  7. doi Power and sample size estimation for the Wilcoxon rank sum test with application to comparisons of C statistics from alternative prediction models
    B Rosner
    Channing Laboratory, Harvard Medical School, Boston, Massachusetts 02115, USA
    Biometrics 65:188-97. 2009
    ..These results are based on SAS-callable functions to evaluate the bivariate normal integral and are thus easily implemented with standard software...
  8. ncbi A nonparametric test for observational non-normally distributed ophthalmic data with eye-specific exposures and outcomes
    Bernard Rosner
    Channing Laboratory, Harvard Medical School, Boston, Massachusetts 02115, USA
    Ophthalmic Epidemiol 14:243-50. 2007
    ....
  9. ncbi Extension of the rank sum test for clustered data: two-group comparisons with group membership defined at the subunit level
    Bernard Rosner
    Channing Laboratory, Harvard Medical School, Boston, Massachusetts 02115, USA
    Biometrics 62:1251-9. 2006
    ..We also present comparisons between the clustered Wilcoxon test and each of the signed rank tests and mixed model approaches and show dramatic differences in power in favor of the clustered Wilcoxon test for some designs...
  10. ncbi The Wilcoxon signed rank test for paired comparisons of clustered data
    Bernard Rosner
    Channing Laboratory, Harvard Medical School, Boston, Massachusetts 02115, USA
    Biometrics 62:185-92. 2006
    ....
  11. ncbi Statistical inference for correlated data in ophthalmologic studies
    Man Lai Tang
    Department of Mathematics, Hong Kong Baptist University, Kowloon, Hong Kong
    Stat Med 25:2771-83. 2006
    ..8). On the other hand, the approximate unconditional procedures usually yield empirical type I error rates close to the pre-chosen nominal level. We illustrate our methodologies with a data set from a retinal detachment study...
  12. ncbi Comparison of risk factors for the competing risks of coronary heart disease, stroke, and venous thromboembolism
    Robert J Glynn
    Division of Preventive Medicine and the Channing Laboratory, Brigham and Women s Hospital, Harvard Medical School, Boston, MA 02215, USA
    Am J Epidemiol 162:975-82. 2005
    ..CHD and stroke have broadly comparable risk factor profiles that differ widely from the profile for VTE...