Applied longitudinal data analysis for epidemiology : a practical guide

The most important techniques available for longitudinal data analysis are discussed in this book. The discussion includes simple techniques such as the paired t-test and summary statistics, but also more sophisticated techniques such as generalized estimating equations and random coefficient analysis. A distinction is made between longitudinal analysis with continuous, dichotomous, and categorical outcome variables. This practical guide is especially suitable for non-statisticians and all those undertaking medical research or epidemiological studies. Cover......Page 1 Half-title......Page 3 Title......Page 5 Copyright......Page 6 Dedication......Page 7 Contents......Page 9 Preface......Page 17 Acknowledgements......Page 18 1.1 Introduction......Page 19 1.3 Prior knowledge......Page 20 1.4 Example......Page 21 1.7 Statistical notation......Page 23 2.1 Introduction......Page 25 2.2.1 Period and cohort effects......Page 27 2.2.2 Other confounding effects......Page 31 2.2.3 Example......Page 32 2.3 Experimental (longitudinal) studies......Page 33 3.1 Two measurements......Page 36 3.1.1 Example......Page 38 3.2 Non-parametric equivalent of the paired t-test......Page 39 3.2.1 Example......Page 40 3.3 More than two measurements......Page 41 3.3.1 The ‘univariate’ approach: numerical example......Page 44 3.3.2 The shape of the relationship between an outcome variable and time......Page 47 3.3.3 A numerical example......Page 48 3.3.4 Example......Page 50 3.4 The ‘univariate’ or the ‘multivariate’ approach?......Page 55 3.5 Comparing groups......Page 56 3.5.1 The ‘univariate’ approach: numerical example......Page 57 3.5.2 Example......Page 59 3.6 Comments......Page 63 3.7 Post-hoc procedures......Page 64 3.7.1 Example......Page 65 3.8 Different contrasts......Page 66 3.8.1 Example......Page 67 3.9 Non-parametric equivalent of MANOVA for repeated measurements......Page 70 3.9.1 Example......Page 71 4.2 ‘Traditional’ methods......Page 73 4.3 Example......Page 75 4.4 Longitudinal methods......Page 78 4.5.2 Working correlation structures......Page 80 4.5.3 Interpretation of the regression coefficients derived from GEE analysis......Page 84 4.5.4.1 Introduction......Page 86 4.5.4.2 Results of a GEE analysis......Page 87 4.5.4.3 Different correlation structures......Page 90 4.5.4.4 Unequally spaced time intervals......Page 93 4.6.2 Random coefficient analysis in longitudinal studies......Page 95 4.6.3.1 Results of a random coefficient analysis......Page 98 4.6.4 Comments......Page 106 4.7 Comparison between GEE analysis and random coefficient analysis......Page 109 4.7.2 Equal variances over time......Page 110 4.7.3 The correction for covariance......Page 111 4.8 The modelling of time......Page 113 4.8.1 Example......Page 116 5.2.1 Time-lag model......Page 120 5.2.2 Modelling of changes......Page 123 5.2.3 Autoregressive model......Page 125 5.2.5.1 Introduction......Page 126 5.2.5.3 GEE analysis......Page 127 5.2.5.4 Random coefficient analysis......Page 130 5.3 Comments......Page 132 5.4 Another example......Page 136 6.1.1 Two measurements......Page 138 6.1.3 Comparing groups......Page 140 6.1.4.2 Development over time......Page 141 6.1.4.3 Comparing groups......Page 144 6.2.2 Example......Page 146 6.2.3 Sophisticated methods......Page 147 6.2.4.1 Generalized estimating equations......Page 149 6.2.4.2 Random coefficient analysis......Page 155 6.2.5 Comparison between GEE analysis and random coefficient analysis......Page 158 6.2.6 Alternative models......Page 161 6.2.7 Comments......Page 162 7.1.1 Two measurements......Page 163 7.1.2 More than two measurements......Page 164 7.1.4 Example......Page 165 7.1.5.2 Example......Page 169 7.1.5.3 Sophisticated methods......Page 170 7.1.5.4 Example......Page 171 7.2 ‘Count’ outcome variables......Page 174 7.2.1.1 Introduction......Page 175 7.2.1.2 GEE analysis......Page 176 7.2.1.3 Random coefficient analysis......Page 181 7.2.2 Comparison between GEE analysis and random coefficient analysis......Page 183 8.2 Continuous outcome variables......Page 185 8.2.1 A numerical example......Page 189 8.2.2 Example......Page 191 8.3.1 Example......Page 193 8.4 Comments......Page 195 8.6 Conclusions......Page 196 9.1 Introduction......Page 197 9.2.1 Introduction......Page 199 9.2.2 Simple analysis......Page 200 9.2.3 Summary statistics......Page 202 9.2.4 MANOVA for repeated measurements......Page 203 9.2.4.1 MANOVA for repeated measurements corrected for the baseline value......Page 204 9.2.5 Sophisticated analysis......Page 206 9.3.2 Simple analysis......Page 213 9.3.3 Sophisticated analysis......Page 214 9.4 Comments......Page 218 10.1 Introduction......Page 220 10.2 Ignorable or informative missing data?......Page 222 10.3.1 Generating datasets with missing data......Page 223 10.3.2 Analysis of determinants for missing data......Page 224 10.4 Analysis performed on datasets with missing data......Page 225 10.4.1 Example......Page 226 10.5 Comments......Page 230 10.6.1.2 Longitudinal imputation methods......Page 231 10.6.1.3 Multiple imputation method......Page 232 10.6.3.1 Continuous outcome variables......Page 234 10.6.3.2 Dichotomous outcome variables......Page 237 10.6.4 Comments......Page 239 10.8 Conclusions......Page 241 11.2 Continuous outcome variables......Page 243 11.3 Dichotomous and categorical outcome variables......Page 248 11.4 Example......Page 252 11.4.1 Two measurements......Page 253 11.4.2 More than two measurements......Page 255 11.5.1 Interpretation of tracking coefficients......Page 256 11.5.3 Grouping of continuous outcome variables......Page 257 11.6 Conclusions......Page 258 12.2.1 STATA......Page 259 12.2.2 SAS......Page 261 12.2.3 S-PLUS......Page 262 12.2.4 Overview......Page 264 12.3.1 STATA......Page 265 12.3.2 SAS......Page 266 12.3.3 S-PLUS......Page 267 12.4.1 STATA......Page 268 12.4.2 SAS......Page 269 12.4.3 S-PLUS......Page 273 12.4.4 SPSS......Page 275 12.4.5 MLwiN......Page 277 12.4.6 Overview......Page 280 12.5.1 Introduction......Page 281 12.5.2 STATA......Page 282 12.5.3 SAS......Page 283 12.5.4 MLwiN......Page 287 12.5.5 Overview......Page 288 12.6 Categorical and ‘count’ outcome variables......Page 289 12.7 Alternative approach using covariance structures......Page 290 12.7.1 Example......Page 292 13.1 Introduction......Page 298 13.2 Example......Page 301 References......Page 304 Index......Page 313 "This book discusses the most important techniques available for longitudinal data analysis, from simple techniques such as the paired t-test and summary statistics, to more sophisticated ones such as generalized estimating of equations and mixed model analysis. A distinction is made between longitudinal analysis with continuous, dichotomous and categorical outcome variables. The emphasis of the discussion lies in the interpretation and comparison of the results of the different techniques. The second edition includes new chapters on the role of the time variable and presents new features of longitudinal data analysis. Explanations have been clarified where necessary and several chapters have been completely rewritten. The analysis of data from experimental studies and the problem of missing data in longitudinal studies are discussed. Finally, an extensive overview and comparison of different software packages is provided. This practical guide is essential for non-statisticians and researchers working with longitudinal data from epidemiological and clinical studies"-- "The emphasis of this book lies more on the application of statistical techniques for longitudinal data analysis and not so much on the mathematical background. In most other books on the topic of longitudinal data analysis, the mathematical background is the major issue, which may not be surprising since (nearly) all the books on this topic have been written by statisticians. Although statisticians fully understand the difficult mathematical material underlying longitudinal data analysis, they often have difficulty in explaining this complex material in a way that is understandable for the researchers who have to use the technique or interpret the results. Therefore, this book is not written by a statistician, but by an epidemiologist. In fact, an epidemiologist is not primarily interested in the basic (difficult) mathematical background of the statistical methods, but in finding the answer to a specific research question; the epidemiologist wants to know how to apply a statistical technique and how to interpret the results. Owing to their different basic interests and different level of thinking, communication problems between statisticians and epidemiologists are quite common. This, in addition to the growing interest in longitudinal studies, initiated the writing of this book: a book on longitudinal data analysis, which is especially suitable for the "non-statistical" researcher (e.g. the epidemiologist). The aim of this book is to provide a practical guide on how to handle epidemiological data from longitudinal studies"-- In This Book The Most Important Techniques Available For Longitudinal Data Analysis Are Discussed. A Distinction Is Made Between Longitudinal Analysis With Continuous, Dichotomous, And Categorical Outcome Variables. This Practical Guide Is Suitable For Non-statisticians And Researchers Working With Longitudinal Data From Epidemiological And Clinical Studies.--book Jacket. Study Design -- Continuous Outcome Variables -- Continuous Outcome Variables: Relationships With Other Variables -- Other Possibilities For Modelling Longitudinal Data -- Dichotomous Outcome Variables -- Categorical And 'count' Outcome Variables -- Longitudinal Studies With Two Measurements: The Definition And Analysis Of Change -- Analysis Of Experimental Studies -- Missing Data In Longitudinal Studies -- Tracking -- Software For Longitudinal Data Analysis -- Sample Size Calculations. Jos W.r. Twisk. Includes Bibliographical References (p. 286-294) And Index. This book discusses the most important techniques available for longitudinal data analysis, from simple techniques such as the paired t-test and summary statistics, to more sophisticated ones such as generalized estimating of equations and mixed model analysis. A distinction is made between longitudinal analysis with continuous, dichotomous and categorical outcome variables. The emphasis of the discussion lies in the interpretation and comparison of the results of the different techniques. Furthermore, special chapters deal with the analysis of two measurements, experimental studies and the problem of missing data in longitudinal studies. Finally, an extensive overview of different software packages is provided In this book the most important techniques available for longitudinal data analysis are discussed, including simple techniques such as the paired t-test and summary statistics, and more sophisticated techniques such as generalised estimating equations and random coefficient analysis. This practical guide is suitable for non-statisticians involved in medical research and epidemiology Longitudinal studies are defined as studies in which the outcome variable is repeatedly measured; i.e. the outcome variable is measured in the same individual on several different occasions.