A TWO-STAGE LINEAR MIXED-EFFECTS/COX MODEL FOR LONGITUDINAL DATA WITH MEASUREMENT ERROR AND SURVIVAL

Size: px
Start display at page:

Download "A TWO-STAGE LINEAR MIXED-EFFECTS/COX MODEL FOR LONGITUDINAL DATA WITH MEASUREMENT ERROR AND SURVIVAL"

Transcription

1 A TWO-STAGE LINEAR MIXED-EFFECTS/COX MODEL FOR LONGITUDINAL DATA WITH MEASUREMENT ERROR AND SURVIVAL Christopher H. Morrell, Loyola College in Maryland, and Larry J. Brant, NIA Christopher H. Morrell, Mathematical Sciences Department, Loyola College in Maryland, 4501 North Charles Street, Baltimore, MD Introduction Recently a number of papers have considered both longitudinal changes in a variable and the associated effect on the length of time to the occurrence of an event (Altman and De Stavola, 1994; Bycott and Taylor, 1998; Dafni and Tsiatis, 1998; De Gruttola and Tu, 1994; Hogan and Laird, 1997; Tsiatis et al., 1995; Wulfsohn and Tsiatis, 1997). Some of these and other papers also address the issue of the effect of measurement errors in the covariate on the estimate of the parameters in a Cox model (Hu et al., 1998; Hughes, 1993; Pepe et al. 1989; Prentice, 1982). It is well known that measurement error in the covariates leads to underestimates of the parameters in the Cox model (Prentice, 1982). This underestimation is sometimes referred to as regression dilution bias (Hughes, 1993). A number of these papers present models that jointly account for the repeated measurements over time and their association with the outcome event. To fit the joint models, the EM algorithm may be used to estimate the parameters from a suitably defined likelihood, or Markov-Chain Monte Carlo methods may be applied. These methods require extensive time to develop the software routines to fit these models. Furthermore, any model reformulation would require additional time to modify the procedure. In this paper, we explore a two-stage procedure similar to those proposed by Bycott and Taylor (1998) and Dafni and Tsiatis (1998) and evaluate its properties using a Monte Carlo study. The procedure first models the repeated measurements over time using a linear mixed-effects (LME) model (Laird and Ware, 1982) and then uses individual predictions from this LME model as either fixed- or timedependent covariates in a Cox model. In the first stage the pattern of change for each individual is estimated by a LME model. This model will smooth out irregularities in the observed data and also account for measurement error or other noise. As an example, we investigate the effects of systolic blood pressure on the time to development of coronary heart disease (CHD) using data from the Baltimore Longitudinal Study of Aging (BLSA). We study the effect of various numbers of repeated measurement times and the number of observations at each measurement time on the Cox model parameter estimate. We consider data for individuals with only one measurement time, individuals with up to three measurement times (or visits), and individuals with up to five visits. In addition, we consider single observations at each visit versus up to four repeated blood pressure measurements per visit. We model these various longitudinal data sets with a LME model using a backward elimination procedure as described in Morrell et al. (1997). Using the estimated random effects from the LME models, empirical Bayes estimates of the beginning systolic blood pressure for each individual are obtained for use in a fixed covariates Cox model relating systolic blood pressure to time to the occurrence of CHD. In addition, the LME models are used to estimate the blood pressure for each individual at each event time in a timedependent or updated covariates Cox model (Altman and De Stavola, 1994). This method may be applied using commercially available software that can fit both the LME model as well as both fixed and timedependent covariate Cox models (for example, proc mixed and proc phreg, (SAS, 1996)). This procedure is similar to the one employed by Tsiatis et al. (1995) in their study of CD4 counts in AIDS patients. However, we also investigate the effects on the Cox model parameters of varying numbers of visits and varying numbers of measurements per visit. This twostage procedure has the advantage of readily allowing for the inclusion of other covariates or other quantities derived from the LME model, for example, rate of change in the covariate, which may influence the time to event. However, caution must be exercised in using the time-dependent Cox model if the survival/censoring time for individuals are far beyond the last follow-up time used in the LME model for that individual since extrapolation of the predicted path well beyond the observed data may be risky. Finally, the properties of the Cox model parameters obtained using these methods are evaluated using a Monte Carlo simulation study with various levels of censoring. 2. Statistical Methods 2.1. Linear Mixed-Effects Model Using the notation in Laird and Ware (1982), the model for the i th cluster or individual is y = Xi + Zi bi + ei, i =1,..., m Reprinted from the 2000 Proceedings of the Biometrics Section of the American Statistical Association i 198

2 where y i is the n i 1 vector of observations for cluster i, X i is the design matrix of independent variables for cluster i, β is the p 1 vector of parameters for the fixed effects in the LME model, Z i is the design matrix of predictor variables for the random effects, b i is the q 1 vector of random effects for cluster i, e i is the error vector, and m represents the number of different clusters. It is assumed that b i ~ N (0, D) where D is a q q variance-covariance matrix of the b i, e i ~ N (0, σ 2 I), and the b i and e i are independent. The fixed effects, β, give the population average intercept and slopes. They model the systematic variation in the data that can be linked to explanatory variables that differ among the clusters and other explanatory variables that vary within the clusters. In contrast, the random effects, b i, account for the heterogeneity among the clusters by allowing their intercepts and partial slopes for each cluster to differ from the overall average. Finally, the random errors, e i, account for the unexplained variation in the data Cox Model The proportional hazards model, as formulated by Cox (1972), studies the relationship of a number of covariates to the time to survival in the presence of censoring. In the Cox model the hazard rate or intensity of failure for the survival time of a subject with covariate vector X C is λ(t X C ) = λ 0 (t)exp(θ X C ), t 0, where θ is a vector of unknown regression coefficients and λ 0 (t) is an arbitrary baseline hazard function. The problem is to estimate the regression parameter, θ, and baseline hazard function, λ 0 (t), in the presence of right censored survival data. This is accomplished by maximizing the partial likelihood (Kalbfleisch and Prentice, 1980). However, if the covariate, X C, is measured with error, the estimate of θ will be biased downward (Prentice, 1982). In addition, if the covariate vector, X C, is repeatedly measured over time, then the timedependent or updated covariates Cox model (Altman and De Stavola,1994) may be used to assess the association of the covariates with survival. To achieve this it is necessary to specify the value of the covariates at each survival time Determining the Covariate Values Follow-up studies that examine the relationship between a set of covariates and the time to the occurrence of an outcome rely on measurements that may often be only a single measurement of each covariate. If these single values are measured with error or have short-term biological variability, the parameter estimate of the association to the outcome event will be biased downward. If multiple measurements are available at each time point, these multiple values may be used to overcome some of the problems associated with measurement error, for example, by using the average value at each time point. A second issue that needs to be addressed is how to efficiently use the repeated values over time to estimate the association of the covariates with the time to the event. Altman and De Stavola (1994) mention the possibility of estimating the time path for the covariate and using these time paths in the Cox model as time-dependent or updated covariates. This paper addresses both issues of measurement errors and updated covariates by using a two-stage procedure. First, the linear mixed-effects model is used to describe the repeated measurements or patterns of change in the data for each subject. This will reduce the problems associated with measurement errors and permit the time path of the covariates for each subject to be estimated. Second, empirical Bayes shrinkage estimates from the mixed-effects model are used either to estimate the baseline value of the covariates, which may be used as fixed covariates in the Cox model, or to estimate the value of the covariate at each event time in an updated covariates analysis. This updated covariates analysis is simple to implement in SAS. The time-dependent covariate is specified in the model statement and then a formula obtained from the final LME model is given to evaluate the covariate. This formula will include, as variables, the estimated random effects so that individual-specific values of the covariate may be calculated at each time point. 3. Data Analysis For this paper, systolic blood pressure measurements that have been collected on 932 male participants of the Baltimore Longitudinal Study of Aging (BLSA) are used for examining the risk of systolic blood pressure on time to an event of coronary heart disease (CHD). CHD is defined as the occurrence of a coronary death, or the diagnosis of a myocardial infarction by history or pathologic Q- waves, or angina pectoris. Out of the 932 participants 185 experienced a coronary event. Table 1 displays some descriptive statistics for the CHD cases and censored events. The participants with CHD tend to be older and have shorter survival times. We used a maximum of 5 visits or repeated examinations per participant chosen to be as uniformly spaced over the interval of study as possible. 199

3 Table 1. Descriptive statistics (mean standard deviatio n) of BLSA sample data. Systolic Blood Pressure Censored (n = 747) CHD ( n = 185) Age at baseline Survival time There was an average of 3.1 visits per participant. In addition, up to 4 repeated blood pressure readings were recorded at each visit during the 22 days the participants spend at the BLSA. There was an average of 3.0 blood pressure readings per visit. To investigate the association of systolic blood pressure on the time to CHD, the data is analyzed in various ways: 1. The first blood pressure from the first visit is used in a fixed-covariate Cox model. 2. The mean of the repeated values at the first visit is used in a fixed-covariate Cox model. 3. Repeated values at first visit are modeled using a LME model. The beginning blood pressure is estimated from the LME model for use in a fixedcovariate Cox model. 4. The pattern of change based on a single value per visit with a maximum of 3 visits per individual is obtained using a LME model, then a) estimate the beginning blood pressure from the LME model for use in a fixed-covariate Cox model and b) estimate the blood pressure at each event time from the LME model for use in an updated covariate Cox model. 5. Model the pattern of change based on multiple values per visit with a maximum of 3 visits per individual using a LME model then estimate the Cox parameters as in 4 above. 6. Model the pattern of change based on a single value per visit with a maximum of 5 visits per individual using a LME model then estimate the Cox parameters as in 4 above. 7. Model the pattern of change based on multiple values per visit with a maximum of 5 visits per individual using a LME model then estimate the Cox parameters as in 4 above. These possibilities allow us to address how the association of systolic blood pressure on time to CHD depends on the number of repeated visits and the use of single or multiple measurements per visit. The LME model used will depend on the number of visits available in each case. For repeated measurements at first visit, the LME model will only contain fixed and random intercept terms. If y ij is the jth blood pressure for the ith participant, y ij = β 0 + b i 0 + e ij. If we use a maximum of 3 visits, there is only an average of 2.3 visits per participant. In this case we entertain models that contain up to linear terms in follow-up time with both fixed and random effects for time as well as terms in age at first visit (fage), fage 2, fage 3, and interactions of these terms with follow-up time in the LME model. Using a backward elimination procedure (Morrell, Pearson, and Brant, 1997) the final model for the kth observation at visit j for participant i is 2 y ij k = β 0 + b i 0 + (β 1 + b i 1 ) time ij + β 2 fage i + β 3 fage i + β 4 fage 3 i + β 5 time ij fage i + e ij k for both single and multiple observations per visit. With a maximum of 5 visits, follow-up time 2 terms are included as both fixed and random effects as well as interactions with fage. The final model is y ij k = β 0 + b i 0 + (β 1 + b i 1 ) time ij + (β 2 + b i 2 ) time 2 ij + β 3 fage i + β 4 fage 2 i + β 5 fage 3 i + β 6 time ij fage i + e ij k for both single and multiple observations per visit. Figure 1 shows the expected longitudinal trends for two of these data sets. The mean of the baseline systolic blood pressures obtained for the CHD cases and censored observations via the various methods were calculated. The values predicted from the LME models all have smaller standard deviations than for a single observed value or the mean of the observed values. This indicates that some of the noise in the original data has been eliminated by obtaining predictions from the LME model. In addition, using predicted values from LME models based on single values always lead to smaller standard deviations of the predicted values illustrating that these estimates are subject to more shrinkage than when prediction is based on multiple observations per visit. For single observations per visit, the means of the predicted values from the LME model for the CHD cases are smaller than for multiple values per visit whereas the means for the censored cases are similar for the single and multiple values per visit cases. The Cox models are fit with the systolic blood pressure represented as described above. The parameter estimates and the standard errors from the Cox model are given in Table 2. The table shows that using the mean of observation at first visit provides a larger parameter estimate than when only a single value measured at baseline is used. Similarly, estimating the baseline value at first visit from the LME model using multiple measurements at first visit additionally increases the estimate. Interestingly, for this data, the estimates based on repeated visits with single values per visit provide larger estimates than when multiple measurements are used at each visit. 200

4 This may be due to the estimates in the single value cases having more shrinkage towards the mean than in the multiple observations case or due to the fact that, or average, the covariates were smaller for the CHD cases for the single versus multiple observations per visit condition. In addition, increasing from a maximum of 3 visits to a maximum of 5 visits gives a modest increase for single values but almost no change for multiple measurements. In all cases the parameter estimates from the updated covariates Cox model are smaller than their counterparts in the fixed covariates Cox model although a similar pattern holds between 3 and 5 visits and single versus multiple measurements. In this analysis the true value of risk relating systolic blood pressure to time to CHD is unknown. Consequently, it is not clear which of these approaches provides for the best results. Therefore, a simulation study is used to assess the procedure. 4. Simulation Study To assess the bias and precision of these procedures a Monte Carlo study is performed using FORTRAN routines to fit the LME model and estimate the Cox parameters. For simplicity, in the simulation study we assume that each participant has 5 visits and 4 observations at each visit. To generate the data for the Monte Carlo simulation study we adapt the method Table 2. Parameter estimates and standard errors from the Cox model for the various methods of obtaining the SBP covariate values. Systolic Blood Pressure Baseline Cox Model Observed single value ( ) Mean of values at first visit Predicted from LME model: Multiple values at first visit Single value with at most 3 visits Multiple values with at most 3 visits Single values with at most 5 visits Multiple values with at most 5 visits ( ) ( ) ( ) ( ) ( ) ( ) Updated Covariates Cox Model ( ) ( ) ( ) ( ) developed by Dafni (1993) for generating the time to event and the censoring/event variable. The LME model is fit using a FORTRAN routine developed by Mary Lindstrom (Lindstrom and Bates, 1988). The parameters of the Cox model are estimated using the routines developed from Kalbfleisch and Prentice (1980). For a sample data set, the simulation algorithm is checked against the results produced by SAS to ensure that identical results are obtained. In the simulation study we only use linear growth-curve models. The simulation uses the same subsets of data as used in the analysis of the SBP/CHD data. For the case of three repeated measurements, the first, third, and fifth observations are used in the analysis. In addition, the Cox model parameters are estimated using the Atrue@ baseline value of the covariate (before error is added) as a fixed covariate as well as the Atrue@ path of the covariate in a time-dependent analysis. Some of the parameters of the LME and Cox models are chosen to approximately match the BLSA data. In particular, the longitudinal rate of change in the covariate is chosen to be 1, the covariance matrix of the random effects is , where the entry below the main diagonal gives the correlation, the error standard deviation is 8, the Cox parameter is 0.1, and three values of the censoring mean are chosen to obtain mild, moderate, and severe censoring. For the purposes of space, only the censoring pattern closest to the actual data is presented here. Table 3 contains the results of the proportional hazards model regressions. In the simulations, on average 27.1%, 51.0%, and 73.0% of the cases are events. The first row of the table shows that when the true values of the covariate are used in the Cox model the parameter estimates are unbiased with coverage proportions close to the nominal level. As expected, a single value provides estimates that are severely biased towards zero. While the use of the mean of the values at first visit somewhat improves this situation, in these two cases for all the 500 replications the estimates were less than 0.1 for each of the simulations. In all cases the standard errors produced from the Cox model are very similar to the standard deviation of the Monte Carlo estimates. This indicates that the standard errors provided by the Cox model appropriately assess the uncertainty in the estimate. When the LME model is used to predict the covariates in the Cox model, some general observations may be made. As the number of visits 201

5 increases the parameter estimates become less biased, using multiple observations as opposed to single observations improves the estimates, and the updated covariates estimates provide improvements over the baseline covariates. This latter observation is not surprising since the hazard function was assumed to be a function of the linear trend for each individual. These results agree with Dafni (1993) who showed that the two-stage model gave results that were still slightly biased towards zero, even though they eliminated much of the bias of the uncorrected values. Interestingly, as the proportion of events increases, the Cox model parameter estimates become more biased and the standard errors of the estimates decrease. This leads to lower coverage proportions. Based on the CHD data example, the parameter estimates from the Cox model were larger for the single value at 3 or 5 visits than for the multiple values scenario. In contrast, the simulations show that on average the parameters based on single values will be more biased towards zero than those based on multiple values. 5. Conclusions We have demonstrated the use of a two-stage model for using longitudinal data that may be measured with error or possess short-term biological variability to assess the risk of time to developing an event. Our procedure may be applied with commercially available software. For example, first proc mixed in SAS (SAS Institute Inc., 1996) may be used to fit the LME model and obtain the estimated fixed and random effects from the covariate data and then proc phreg may be used to fit both the fixed-covariate and time-dependent or updated covariate Cox model. Wulfsohn and Tsiatis (1997) developed a unified model to simultaneously estimate the covariate and failure time processes. Based on their example, their model provides less biased estimates than the twostage process. However, software to fit their model is not generally available. In addition, the two-stage process readily allows for the incorporation of additional fixed and time-varying covariates and for the inclusion of functions of the time-varying covariates. For example, the rate of change in the covariate may be evaluated as the derivative of the function that defines the pattern of change over time and this derivative may be used either as a covariate at baseline to evaluate how the initial rate of change impacts time to event or as a time-varying covariate. However, caution must be exercised in using the time-dependent Cox model if survival/censoring time for individuals are far beyond the last follow-up time used in the LME model for that individual since extrapolation of the predicted path well beyond the observed data may be risky. Acknowledgements We also thank Dr. Urania Dafni for her assistance on the method of generating the data for the Monte Carlo simulations. References D.G. Altman and B.L. De Stavola, APractical problems in fitting a proportional hazards model to data with updated measurements of the covariates,@ Stat. Med., vol. 13, pp , P. Bycott and J. Taylor, AA comparison of smoothing techniques for cd4 data measured with error in a timedependent Cox proportional hazards model,@ Stat. Med., vol. 17, pp , D.R. Cox, ARegression models and life tables@ (with discussion), J. Roy. Statist. Soc. B, vol. 43, pp , U.G. Dafni. Evaluating Surrogate markers of clinical outcome when measured with error. Ph.D. dissertation, Biostatistics Library, Harvard School of Public Health, U.G. Dafni and A.A. Tsiatis, AEvaluating surrogate markers of clinical outcome when measured with error,@ Biometrics, vol. 54, pp , V. De Gruttola and X.M. Tu, AModeling progression of CD-4-lymphocyte count and its relationship to survival time,@ Biometrics, vol. 50, pp , J.W. Hogan and N.M. Laird, AModel-based approaches ro analysing incomplete longitudinal and failure time data,@ Stat. Med., vol. 16, pp , P. Hu, A.A. Tsiatis, and M. Davidian, AEstimating the parameters in the Cox model when covariate variables are measured with error,@ Biometrics, vol. 54, pp , M.D. Hughes, ARegression dilution in the proportional hazards model,@ Biometrics, vol. 49, pp , J.D. Kalbfleisch and R.L. Prentice, The Statistical Analysis of Failure Time Data, John Wiley and Sons, New York, N.M. Laird and J.H. Ware, ARandom-effects models for longitudinal data,@ Biometrics, vol. 38, pp , M.J. Lindstrom and D.M. Bates, ANewton-Raphson and EM algorithms for linear mixed-effects models for repeated-measures data,@ JASA, vol. 83, pp ,

6 C.H. Morrell, J.D. Pearson, and L.J. Brant, ALinear transformations of linear mixed-effects The American Statistician, vol. 51, pp , M.S. Pepe, S.G. Self, and R.L. Prentice, AFurther Results on covariate measurement errors in cohort studies with time to response Stat. Med., vol. 8, pp , R.L. Prentice, ACovariate measurement errors and parameter estimation in a failure time regression model,@ Biometrika, vol. 69, pp , SAS Institute Inc. SAS/STAT Software: Changes and Enhancements through Release 6.11, Cary, NC: SAS Institute Inc., A.A. Tsiatis, V. De Gruttola, and M.S. Wulfsohn, AModeling the relationship of survival to longitudinal data measures with error. Applications to survival and CD4 counts in patients with AIDS,@ JASA, vol. 90, pp , M.S. Wulfsohn and A.A. Tsiatis, AA joint model for survival and longitudinal data measured with error,@ Biometrics, vol. 53, pp , Table 3. Mean (standard deviation) and standard errors of parameter estimates, coverage proportions for 95% confidence intervals for the Cox model parameter, θ H, and MSE ( 10-5 ) from the 500 Monte Carlo replications of the LME-Cox procedure for the various methods of obtaining the covariate values. Baseline Covariate Updated Covariates Covariate Parameter SE MSE Coverage Parameter SE MSE Coverage Without error ( ) ( ) ( ) ( ) Observed single value ( ) ( ) Mean of values at first visit ( ) ( ) Predicted from LME model: Multiple values at first visit ( ) ( ) Single value at visits ( ) ( ) ( ) ( ) Multiple values at 3 visits ( ) ( ) ( ) ( ) Single values at visits ( ) ( ) ( ) ( ) Multiple values at 5 visits ( ) ( ) ( ) ( ) H True value of θ = 0.1. On average, 27.1% of the cases are events. Single Observations With at Most 3 Visits Multiple Values at All Visits SBP 120 SBP Age Figure 1. Expected longitudinal trends based on the linear mixed-effects model for two of the data sets. Age 203

FULL LIKELIHOOD INFERENCES IN THE COX MODEL

FULL LIKELIHOOD INFERENCES IN THE COX MODEL October 20, 2007 FULL LIKELIHOOD INFERENCES IN THE COX MODEL BY JIAN-JIAN REN 1 AND MAI ZHOU 2 University of Central Florida and University of Kentucky Abstract We use the empirical likelihood approach

More information

Review Article Analysis of Longitudinal and Survival Data: Joint Modeling, Inference Methods, and Issues

Review Article Analysis of Longitudinal and Survival Data: Joint Modeling, Inference Methods, and Issues Journal of Probability and Statistics Volume 2012, Article ID 640153, 17 pages doi:10.1155/2012/640153 Review Article Analysis of Longitudinal and Survival Data: Joint Modeling, Inference Methods, and

More information

Longitudinal + Reliability = Joint Modeling

Longitudinal + Reliability = Joint Modeling Longitudinal + Reliability = Joint Modeling Carles Serrat Institute of Statistics and Mathematics Applied to Building CYTED-HAROSA International Workshop November 21-22, 2013 Barcelona Mainly from Rizopoulos,

More information

[Part 2] Model Development for the Prediction of Survival Times using Longitudinal Measurements

[Part 2] Model Development for the Prediction of Survival Times using Longitudinal Measurements [Part 2] Model Development for the Prediction of Survival Times using Longitudinal Measurements Aasthaa Bansal PhD Pharmaceutical Outcomes Research & Policy Program University of Washington 69 Biomarkers

More information

BIAS OF MAXIMUM-LIKELIHOOD ESTIMATES IN LOGISTIC AND COX REGRESSION MODELS: A COMPARATIVE SIMULATION STUDY

BIAS OF MAXIMUM-LIKELIHOOD ESTIMATES IN LOGISTIC AND COX REGRESSION MODELS: A COMPARATIVE SIMULATION STUDY BIAS OF MAXIMUM-LIKELIHOOD ESTIMATES IN LOGISTIC AND COX REGRESSION MODELS: A COMPARATIVE SIMULATION STUDY Ingo Langner 1, Ralf Bender 2, Rebecca Lenz-Tönjes 1, Helmut Küchenhoff 2, Maria Blettner 2 1

More information

GROUPED SURVIVAL DATA. Florida State University and Medical College of Wisconsin

GROUPED SURVIVAL DATA. Florida State University and Medical College of Wisconsin FITTING COX'S PROPORTIONAL HAZARDS MODEL USING GROUPED SURVIVAL DATA Ian W. McKeague and Mei-Jie Zhang Florida State University and Medical College of Wisconsin Cox's proportional hazard model is often

More information

Lecture 5 Models and methods for recurrent event data

Lecture 5 Models and methods for recurrent event data Lecture 5 Models and methods for recurrent event data Recurrent and multiple events are commonly encountered in longitudinal studies. In this chapter we consider ordered recurrent and multiple events.

More information

Semiparametric Mixed Effects Models with Flexible Random Effects Distribution

Semiparametric Mixed Effects Models with Flexible Random Effects Distribution Semiparametric Mixed Effects Models with Flexible Random Effects Distribution Marie Davidian North Carolina State University davidian@stat.ncsu.edu www.stat.ncsu.edu/ davidian Joint work with A. Tsiatis,

More information

Multivariate Survival Analysis

Multivariate Survival Analysis Multivariate Survival Analysis Previously we have assumed that either (X i, δ i ) or (X i, δ i, Z i ), i = 1,..., n, are i.i.d.. This may not always be the case. Multivariate survival data can arise in

More information

Continuous Time Survival in Latent Variable Models

Continuous Time Survival in Latent Variable Models Continuous Time Survival in Latent Variable Models Tihomir Asparouhov 1, Katherine Masyn 2, Bengt Muthen 3 Muthen & Muthen 1 University of California, Davis 2 University of California, Los Angeles 3 Abstract

More information

Estimation in Generalized Linear Models with Heterogeneous Random Effects. Woncheol Jang Johan Lim. May 19, 2004

Estimation in Generalized Linear Models with Heterogeneous Random Effects. Woncheol Jang Johan Lim. May 19, 2004 Estimation in Generalized Linear Models with Heterogeneous Random Effects Woncheol Jang Johan Lim May 19, 2004 Abstract The penalized quasi-likelihood (PQL) approach is the most common estimation procedure

More information

STAT331. Cox s Proportional Hazards Model

STAT331. Cox s Proportional Hazards Model STAT331 Cox s Proportional Hazards Model In this unit we introduce Cox s proportional hazards (Cox s PH) model, give a heuristic development of the partial likelihood function, and discuss adaptations

More information

Power and Sample Size Calculations with the Additive Hazards Model

Power and Sample Size Calculations with the Additive Hazards Model Journal of Data Science 10(2012), 143-155 Power and Sample Size Calculations with the Additive Hazards Model Ling Chen, Chengjie Xiong, J. Philip Miller and Feng Gao Washington University School of Medicine

More information

On Estimating the Relationship between Longitudinal Measurements and Time-to-Event Data Using a Simple Two-Stage Procedure

On Estimating the Relationship between Longitudinal Measurements and Time-to-Event Data Using a Simple Two-Stage Procedure Biometrics DOI: 10.1111/j.1541-0420.2009.01324.x On Estimating the Relationship between Longitudinal Measurements and Time-to-Event Data Using a Simple Two-Stage Procedure Paul S. Albert 1, and Joanna

More information

Estimation of Conditional Kendall s Tau for Bivariate Interval Censored Data

Estimation of Conditional Kendall s Tau for Bivariate Interval Censored Data Communications for Statistical Applications and Methods 2015, Vol. 22, No. 6, 599 604 DOI: http://dx.doi.org/10.5351/csam.2015.22.6.599 Print ISSN 2287-7843 / Online ISSN 2383-4757 Estimation of Conditional

More information

Approximate Median Regression via the Box-Cox Transformation

Approximate Median Regression via the Box-Cox Transformation Approximate Median Regression via the Box-Cox Transformation Garrett M. Fitzmaurice,StuartR.Lipsitz, and Michael Parzen Median regression is used increasingly in many different areas of applications. The

More information

Tutorial 6: Tutorial on Translating between GLIMMPSE Power Analysis and Data Analysis. Acknowledgements:

Tutorial 6: Tutorial on Translating between GLIMMPSE Power Analysis and Data Analysis. Acknowledgements: Tutorial 6: Tutorial on Translating between GLIMMPSE Power Analysis and Data Analysis Anna E. Barón, Keith E. Muller, Sarah M. Kreidler, and Deborah H. Glueck Acknowledgements: The project was supported

More information

Part IV Extensions: Competing Risks Endpoints and Non-Parametric AUC(t) Estimation

Part IV Extensions: Competing Risks Endpoints and Non-Parametric AUC(t) Estimation Part IV Extensions: Competing Risks Endpoints and Non-Parametric AUC(t) Estimation Patrick J. Heagerty PhD Department of Biostatistics University of Washington 166 ISCB 2010 Session Four Outline Examples

More information

Restricted Maximum Likelihood in Linear Regression and Linear Mixed-Effects Model

Restricted Maximum Likelihood in Linear Regression and Linear Mixed-Effects Model Restricted Maximum Likelihood in Linear Regression and Linear Mixed-Effects Model Xiuming Zhang zhangxiuming@u.nus.edu A*STAR-NUS Clinical Imaging Research Center October, 015 Summary This report derives

More information

Survival Analysis for Case-Cohort Studies

Survival Analysis for Case-Cohort Studies Survival Analysis for ase-ohort Studies Petr Klášterecký Dept. of Probability and Mathematical Statistics, Faculty of Mathematics and Physics, harles University, Prague, zech Republic e-mail: petr.klasterecky@matfyz.cz

More information

PQL Estimation Biases in Generalized Linear Mixed Models

PQL Estimation Biases in Generalized Linear Mixed Models PQL Estimation Biases in Generalized Linear Mixed Models Woncheol Jang Johan Lim March 18, 2006 Abstract The penalized quasi-likelihood (PQL) approach is the most common estimation procedure for the generalized

More information

Improving Efficiency of Inferences in Randomized Clinical Trials Using Auxiliary Covariates

Improving Efficiency of Inferences in Randomized Clinical Trials Using Auxiliary Covariates Improving Efficiency of Inferences in Randomized Clinical Trials Using Auxiliary Covariates Anastasios (Butch) Tsiatis Department of Statistics North Carolina State University http://www.stat.ncsu.edu/

More information

SEMIPARAMETRIC APPROACHES TO INFERENCE IN JOINT MODELS FOR LONGITUDINAL AND TIME-TO-EVENT DATA

SEMIPARAMETRIC APPROACHES TO INFERENCE IN JOINT MODELS FOR LONGITUDINAL AND TIME-TO-EVENT DATA SEMIPARAMETRIC APPROACHES TO INFERENCE IN JOINT MODELS FOR LONGITUDINAL AND TIME-TO-EVENT DATA Marie Davidian and Anastasios A. Tsiatis http://www.stat.ncsu.edu/ davidian/ http://www.stat.ncsu.edu/ tsiatis/

More information

Stock Sampling with Interval-Censored Elapsed Duration: A Monte Carlo Analysis

Stock Sampling with Interval-Censored Elapsed Duration: A Monte Carlo Analysis Stock Sampling with Interval-Censored Elapsed Duration: A Monte Carlo Analysis Michael P. Babington and Javier Cano-Urbina August 31, 2018 Abstract Duration data obtained from a given stock of individuals

More information

Modelling Survival Events with Longitudinal Data Measured with Error

Modelling Survival Events with Longitudinal Data Measured with Error Modelling Survival Events with Longitudinal Data Measured with Error Hongsheng Dai, Jianxin Pan & Yanchun Bao First version: 14 December 29 Research Report No. 16, 29, Probability and Statistics Group

More information

Multi-state Models: An Overview

Multi-state Models: An Overview Multi-state Models: An Overview Andrew Titman Lancaster University 14 April 2016 Overview Introduction to multi-state modelling Examples of applications Continuously observed processes Intermittently observed

More information

Approximation of Survival Function by Taylor Series for General Partly Interval Censored Data

Approximation of Survival Function by Taylor Series for General Partly Interval Censored Data Malaysian Journal of Mathematical Sciences 11(3): 33 315 (217) MALAYSIAN JOURNAL OF MATHEMATICAL SCIENCES Journal homepage: http://einspem.upm.edu.my/journal Approximation of Survival Function by Taylor

More information

Optimal Treatment Regimes for Survival Endpoints from a Classification Perspective. Anastasios (Butch) Tsiatis and Xiaofei Bai

Optimal Treatment Regimes for Survival Endpoints from a Classification Perspective. Anastasios (Butch) Tsiatis and Xiaofei Bai Optimal Treatment Regimes for Survival Endpoints from a Classification Perspective Anastasios (Butch) Tsiatis and Xiaofei Bai Department of Statistics North Carolina State University 1/35 Optimal Treatment

More information

UNIVERSITY OF CALIFORNIA, SAN DIEGO

UNIVERSITY OF CALIFORNIA, SAN DIEGO UNIVERSITY OF CALIFORNIA, SAN DIEGO Estimation of the primary hazard ratio in the presence of a secondary covariate with non-proportional hazards An undergraduate honors thesis submitted to the Department

More information

Correction for classical covariate measurement error and extensions to life-course studies

Correction for classical covariate measurement error and extensions to life-course studies Correction for classical covariate measurement error and extensions to life-course studies Jonathan William Bartlett A thesis submitted to the University of London for the degree of Doctor of Philosophy

More information

Beyond GLM and likelihood

Beyond GLM and likelihood Stat 6620: Applied Linear Models Department of Statistics Western Michigan University Statistics curriculum Core knowledge (modeling and estimation) Math stat 1 (probability, distributions, convergence

More information

Modelling geoadditive survival data

Modelling geoadditive survival data Modelling geoadditive survival data Thomas Kneib & Ludwig Fahrmeir Department of Statistics, Ludwig-Maximilians-University Munich 1. Leukemia survival data 2. Structured hazard regression 3. Mixed model

More information

On the relation between initial value and slope

On the relation between initial value and slope Biostatistics (2005), 6, 3, pp. 395 403 doi:10.1093/biostatistics/kxi017 Advance Access publication on April 14, 2005 On the relation between initial value and slope K. BYTH NHMRC Clinical Trials Centre,

More information

Best Linear Unbiased Prediction: an Illustration Based on, but Not Limited to, Shelf Life Estimation

Best Linear Unbiased Prediction: an Illustration Based on, but Not Limited to, Shelf Life Estimation Libraries Conference on Applied Statistics in Agriculture 015-7th Annual Conference Proceedings Best Linear Unbiased Prediction: an Illustration Based on, but Not Limited to, Shelf Life Estimation Maryna

More information

Frailty Modeling for clustered survival data: a simulation study

Frailty Modeling for clustered survival data: a simulation study Frailty Modeling for clustered survival data: a simulation study IAA Oslo 2015 Souad ROMDHANE LaREMFiQ - IHEC University of Sousse (Tunisia) souad_romdhane@yahoo.fr Lotfi BELKACEM LaREMFiQ - IHEC University

More information

PENALIZED LIKELIHOOD PARAMETER ESTIMATION FOR ADDITIVE HAZARD MODELS WITH INTERVAL CENSORED DATA

PENALIZED LIKELIHOOD PARAMETER ESTIMATION FOR ADDITIVE HAZARD MODELS WITH INTERVAL CENSORED DATA PENALIZED LIKELIHOOD PARAMETER ESTIMATION FOR ADDITIVE HAZARD MODELS WITH INTERVAL CENSORED DATA Kasun Rathnayake ; A/Prof Jun Ma Department of Statistics Faculty of Science and Engineering Macquarie University

More information

Faculty of Health Sciences. Regression models. Counts, Poisson regression, Lene Theil Skovgaard. Dept. of Biostatistics

Faculty of Health Sciences. Regression models. Counts, Poisson regression, Lene Theil Skovgaard. Dept. of Biostatistics Faculty of Health Sciences Regression models Counts, Poisson regression, 27-5-2013 Lene Theil Skovgaard Dept. of Biostatistics 1 / 36 Count outcome PKA & LTS, Sect. 7.2 Poisson regression The Binomial

More information

Pairwise rank based likelihood for estimating the relationship between two homogeneous populations and their mixture proportion

Pairwise rank based likelihood for estimating the relationship between two homogeneous populations and their mixture proportion Pairwise rank based likelihood for estimating the relationship between two homogeneous populations and their mixture proportion Glenn Heller and Jing Qin Department of Epidemiology and Biostatistics Memorial

More information

Tests of independence for censored bivariate failure time data

Tests of independence for censored bivariate failure time data Tests of independence for censored bivariate failure time data Abstract Bivariate failure time data is widely used in survival analysis, for example, in twins study. This article presents a class of χ

More information

A penalized likelihood approach to joint modeling of longitudinal measurements and time-to-event data

A penalized likelihood approach to joint modeling of longitudinal measurements and time-to-event data Statistics and Its Interface Volume 1 2008 33 45 A penalized likelihood approach to joint modeling of longitudinal measurements and time-to-event data Wen Ye, Xihong Lin and Jeremy M. G. Taylor Recently

More information

REGRESSION ANALYSIS FOR TIME-TO-EVENT DATA THE PROPORTIONAL HAZARDS (COX) MODEL ST520

REGRESSION ANALYSIS FOR TIME-TO-EVENT DATA THE PROPORTIONAL HAZARDS (COX) MODEL ST520 REGRESSION ANALYSIS FOR TIME-TO-EVENT DATA THE PROPORTIONAL HAZARDS (COX) MODEL ST520 Department of Statistics North Carolina State University Presented by: Butch Tsiatis, Department of Statistics, NCSU

More information

Lecture 7 Time-dependent Covariates in Cox Regression

Lecture 7 Time-dependent Covariates in Cox Regression Lecture 7 Time-dependent Covariates in Cox Regression So far, we ve been considering the following Cox PH model: λ(t Z) = λ 0 (t) exp(β Z) = λ 0 (t) exp( β j Z j ) where β j is the parameter for the the

More information

A comparison of inverse transform and composition methods of data simulation from the Lindley distribution

A comparison of inverse transform and composition methods of data simulation from the Lindley distribution Communications for Statistical Applications and Methods 2016, Vol. 23, No. 6, 517 529 http://dx.doi.org/10.5351/csam.2016.23.6.517 Print ISSN 2287-7843 / Online ISSN 2383-4757 A comparison of inverse transform

More information

ECLT 5810 Linear Regression and Logistic Regression for Classification. Prof. Wai Lam

ECLT 5810 Linear Regression and Logistic Regression for Classification. Prof. Wai Lam ECLT 5810 Linear Regression and Logistic Regression for Classification Prof. Wai Lam Linear Regression Models Least Squares Input vectors is an attribute / feature / predictor (independent variable) The

More information

University of California, Berkeley

University of California, Berkeley University of California, Berkeley U.C. Berkeley Division of Biostatistics Working Paper Series Year 2008 Paper 241 A Note on Risk Prediction for Case-Control Studies Sherri Rose Mark J. van der Laan Division

More information

FULL LIKELIHOOD INFERENCES IN THE COX MODEL: AN EMPIRICAL LIKELIHOOD APPROACH

FULL LIKELIHOOD INFERENCES IN THE COX MODEL: AN EMPIRICAL LIKELIHOOD APPROACH FULL LIKELIHOOD INFERENCES IN THE COX MODEL: AN EMPIRICAL LIKELIHOOD APPROACH Jian-Jian Ren 1 and Mai Zhou 2 University of Central Florida and University of Kentucky Abstract: For the regression parameter

More information

A Sampling of IMPACT Research:

A Sampling of IMPACT Research: A Sampling of IMPACT Research: Methods for Analysis with Dropout and Identifying Optimal Treatment Regimes Marie Davidian Department of Statistics North Carolina State University http://www.stat.ncsu.edu/

More information

SAMPLE SIZE ESTIMATION FOR SURVIVAL OUTCOMES IN CLUSTER-RANDOMIZED STUDIES WITH SMALL CLUSTER SIZES BIOMETRICS (JUNE 2000)

SAMPLE SIZE ESTIMATION FOR SURVIVAL OUTCOMES IN CLUSTER-RANDOMIZED STUDIES WITH SMALL CLUSTER SIZES BIOMETRICS (JUNE 2000) SAMPLE SIZE ESTIMATION FOR SURVIVAL OUTCOMES IN CLUSTER-RANDOMIZED STUDIES WITH SMALL CLUSTER SIZES BIOMETRICS (JUNE 2000) AMITA K. MANATUNGA THE ROLLINS SCHOOL OF PUBLIC HEALTH OF EMORY UNIVERSITY SHANDE

More information

Stat 642, Lecture notes for 04/12/05 96

Stat 642, Lecture notes for 04/12/05 96 Stat 642, Lecture notes for 04/12/05 96 Hosmer-Lemeshow Statistic The Hosmer-Lemeshow Statistic is another measure of lack of fit. Hosmer and Lemeshow recommend partitioning the observations into 10 equal

More information

On the Breslow estimator

On the Breslow estimator Lifetime Data Anal (27) 13:471 48 DOI 1.17/s1985-7-948-y On the Breslow estimator D. Y. Lin Received: 5 April 27 / Accepted: 16 July 27 / Published online: 2 September 27 Springer Science+Business Media,

More information

Joint modelling of longitudinal measurements and event time data

Joint modelling of longitudinal measurements and event time data Biostatistics (2000), 1, 4,pp. 465 480 Printed in Great Britain Joint modelling of longitudinal measurements and event time data ROBIN HENDERSON, PETER DIGGLE, ANGELA DOBSON Medical Statistics Unit, Lancaster

More information

Joint Modeling of Event Time and Nonignorable Missing Longitudinal Data

Joint Modeling of Event Time and Nonignorable Missing Longitudinal Data Lifetime Data Analysis, 8, 99 115, 2002 # 2002 Kluwer Academic Publishers. Printed in The Netherlands. Joint Modeling of Event Time and Nonignorable Missing Longitudinal Data JEAN-FRANÇOIS DUPUY jean-francois.dupuy@univ-ubs.fr

More information

ECLT 5810 Linear Regression and Logistic Regression for Classification. Prof. Wai Lam

ECLT 5810 Linear Regression and Logistic Regression for Classification. Prof. Wai Lam ECLT 5810 Linear Regression and Logistic Regression for Classification Prof. Wai Lam Linear Regression Models Least Squares Input vectors is an attribute / feature / predictor (independent variable) The

More information

Measurement Error in Spatial Modeling of Environmental Exposures

Measurement Error in Spatial Modeling of Environmental Exposures Measurement Error in Spatial Modeling of Environmental Exposures Chris Paciorek, Alexandros Gryparis, and Brent Coull August 9, 2005 Department of Biostatistics Harvard School of Public Health www.biostat.harvard.edu/~paciorek

More information

Mantel-Haenszel Test Statistics. for Correlated Binary Data. Department of Statistics, North Carolina State University. Raleigh, NC

Mantel-Haenszel Test Statistics. for Correlated Binary Data. Department of Statistics, North Carolina State University. Raleigh, NC Mantel-Haenszel Test Statistics for Correlated Binary Data by Jie Zhang and Dennis D. Boos Department of Statistics, North Carolina State University Raleigh, NC 27695-8203 tel: (919) 515-1918 fax: (919)

More information

Sample size re-estimation in clinical trials. Dealing with those unknowns. Chris Jennison. University of Kyoto, January 2018

Sample size re-estimation in clinical trials. Dealing with those unknowns. Chris Jennison. University of Kyoto, January 2018 Sample Size Re-estimation in Clinical Trials: Dealing with those unknowns Christopher Jennison Department of Mathematical Sciences, University of Bath, UK http://people.bath.ac.uk/mascj University of Kyoto,

More information

Separate and Joint Modeling of Longitudinal and Event Time Data Using Standard Computer Packages

Separate and Joint Modeling of Longitudinal and Event Time Data Using Standard Computer Packages Separate and Joint Modeling of Longitudinal and Event Time Data Using Standard Computer Packages Xu GUO and Bradley P. CARLIN Many clinical trials and other medical and reliability studies generate both

More information

Mixture modelling of recurrent event times with long-term survivors: Analysis of Hutterite birth intervals. John W. Mac McDonald & Alessandro Rosina

Mixture modelling of recurrent event times with long-term survivors: Analysis of Hutterite birth intervals. John W. Mac McDonald & Alessandro Rosina Mixture modelling of recurrent event times with long-term survivors: Analysis of Hutterite birth intervals John W. Mac McDonald & Alessandro Rosina Quantitative Methods in the Social Sciences Seminar -

More information

Statistical Issues in the Use of Composite Endpoints in Clinical Trials

Statistical Issues in the Use of Composite Endpoints in Clinical Trials Statistical Issues in the Use of Composite Endpoints in Clinical Trials Longyang Wu and Richard Cook Statistics and Actuarial Science, University of Waterloo May 14, 2010 Outline Introduction: A brief

More information

Impact of approximating or ignoring within-study covariances in multivariate meta-analyses

Impact of approximating or ignoring within-study covariances in multivariate meta-analyses STATISTICS IN MEDICINE Statist. Med. (in press) Published online in Wiley InterScience (www.interscience.wiley.com).2913 Impact of approximating or ignoring within-study covariances in multivariate meta-analyses

More information

The STS Surgeon Composite Technical Appendix

The STS Surgeon Composite Technical Appendix The STS Surgeon Composite Technical Appendix Overview Surgeon-specific risk-adjusted operative operative mortality and major complication rates were estimated using a bivariate random-effects logistic

More information

Correlation and Simple Linear Regression

Correlation and Simple Linear Regression Correlation and Simple Linear Regression Sasivimol Rattanasiri, Ph.D Section for Clinical Epidemiology and Biostatistics Ramathibodi Hospital, Mahidol University E-mail: sasivimol.rat@mahidol.ac.th 1 Outline

More information

Partly Conditional Survival Models for Longitudinal Data

Partly Conditional Survival Models for Longitudinal Data UW Biostatistics Working Paper Series 12-19-23 Partly Conditional Survival Models for Longitudinal Data Yingye Zheng Fred Hutchinson Cancer Research Center, yzheng@fhcrc.org Patrick Heagerty University

More information

Support Vector Hazard Regression (SVHR) for Predicting Survival Outcomes. Donglin Zeng, Department of Biostatistics, University of North Carolina

Support Vector Hazard Regression (SVHR) for Predicting Survival Outcomes. Donglin Zeng, Department of Biostatistics, University of North Carolina Support Vector Hazard Regression (SVHR) for Predicting Survival Outcomes Introduction Method Theoretical Results Simulation Studies Application Conclusions Introduction Introduction For survival data,

More information

LARGE NUMBERS OF EXPLANATORY VARIABLES. H.S. Battey. WHAO-PSI, St Louis, 9 September 2018

LARGE NUMBERS OF EXPLANATORY VARIABLES. H.S. Battey. WHAO-PSI, St Louis, 9 September 2018 LARGE NUMBERS OF EXPLANATORY VARIABLES HS Battey Department of Mathematics, Imperial College London WHAO-PSI, St Louis, 9 September 2018 Regression, broadly defined Response variable Y i, eg, blood pressure,

More information

Likelihood ratio testing for zero variance components in linear mixed models

Likelihood ratio testing for zero variance components in linear mixed models Likelihood ratio testing for zero variance components in linear mixed models Sonja Greven 1,3, Ciprian Crainiceanu 2, Annette Peters 3 and Helmut Küchenhoff 1 1 Department of Statistics, LMU Munich University,

More information

MISSING or INCOMPLETE DATA

MISSING or INCOMPLETE DATA MISSING or INCOMPLETE DATA A (fairly) complete review of basic practice Don McLeish and Cyntha Struthers University of Waterloo Dec 5, 2015 Structure of the Workshop Session 1 Common methods for dealing

More information

On Fitting Generalized Linear Mixed Effects Models for Longitudinal Binary Data Using Different Correlation

On Fitting Generalized Linear Mixed Effects Models for Longitudinal Binary Data Using Different Correlation On Fitting Generalized Linear Mixed Effects Models for Longitudinal Binary Data Using Different Correlation Structures Authors: M. Salomé Cabral CEAUL and Departamento de Estatística e Investigação Operacional,

More information

Group Sequential Tests for Delayed Responses. Christopher Jennison. Lisa Hampson. Workshop on Special Topics on Sequential Methodology

Group Sequential Tests for Delayed Responses. Christopher Jennison. Lisa Hampson. Workshop on Special Topics on Sequential Methodology Group Sequential Tests for Delayed Responses Christopher Jennison Department of Mathematical Sciences, University of Bath, UK http://people.bath.ac.uk/mascj Lisa Hampson Department of Mathematics and Statistics,

More information

The equivalence of the Maximum Likelihood and a modified Least Squares for a case of Generalized Linear Model

The equivalence of the Maximum Likelihood and a modified Least Squares for a case of Generalized Linear Model Applied and Computational Mathematics 2014; 3(5): 268-272 Published online November 10, 2014 (http://www.sciencepublishinggroup.com/j/acm) doi: 10.11648/j.acm.20140305.22 ISSN: 2328-5605 (Print); ISSN:

More information

Group Sequential Designs: Theory, Computation and Optimisation

Group Sequential Designs: Theory, Computation and Optimisation Group Sequential Designs: Theory, Computation and Optimisation Christopher Jennison Department of Mathematical Sciences, University of Bath, UK http://people.bath.ac.uk/mascj 8th International Conference

More information

Analysis of Longitudinal Data. Patrick J. Heagerty PhD Department of Biostatistics University of Washington

Analysis of Longitudinal Data. Patrick J. Heagerty PhD Department of Biostatistics University of Washington Analysis of Longitudinal Data Patrick J Heagerty PhD Department of Biostatistics University of Washington Auckland 8 Session One Outline Examples of longitudinal data Scientific motivation Opportunities

More information

Hypothesis Testing Based on the Maximum of Two Statistics from Weighted and Unweighted Estimating Equations

Hypothesis Testing Based on the Maximum of Two Statistics from Weighted and Unweighted Estimating Equations Hypothesis Testing Based on the Maximum of Two Statistics from Weighted and Unweighted Estimating Equations Takeshi Emura and Hisayuki Tsukuma Abstract For testing the regression parameter in multivariate

More information

Latent Variable Model for Weight Gain Prevention Data with Informative Intermittent Missingness

Latent Variable Model for Weight Gain Prevention Data with Informative Intermittent Missingness Journal of Modern Applied Statistical Methods Volume 15 Issue 2 Article 36 11-1-2016 Latent Variable Model for Weight Gain Prevention Data with Informative Intermittent Missingness Li Qin Yale University,

More information

USING TRAJECTORIES FROM A BIVARIATE GROWTH CURVE OF COVARIATES IN A COX MODEL ANALYSIS

USING TRAJECTORIES FROM A BIVARIATE GROWTH CURVE OF COVARIATES IN A COX MODEL ANALYSIS USING TRAJECTORIES FROM A BIVARIATE GROWTH CURVE OF COVARIATES IN A COX MODEL ANALYSIS by Qianyu Dang B.S., Fudan University, 1995 M.S., Kansas State University, 2000 Submitted to the Graduate Faculty

More information

A new strategy for meta-analysis of continuous covariates in observational studies with IPD. Willi Sauerbrei & Patrick Royston

A new strategy for meta-analysis of continuous covariates in observational studies with IPD. Willi Sauerbrei & Patrick Royston A new strategy for meta-analysis of continuous covariates in observational studies with IPD Willi Sauerbrei & Patrick Royston Overview Motivation Continuous variables functional form Fractional polynomials

More information

University of California, Berkeley

University of California, Berkeley University of California, Berkeley U.C. Berkeley Division of Biostatistics Working Paper Series Year 24 Paper 153 A Note on Empirical Likelihood Inference of Residual Life Regression Ying Qing Chen Yichuan

More information

ADVANCED STATISTICAL ANALYSIS OF EPIDEMIOLOGICAL STUDIES. Cox s regression analysis Time dependent explanatory variables

ADVANCED STATISTICAL ANALYSIS OF EPIDEMIOLOGICAL STUDIES. Cox s regression analysis Time dependent explanatory variables ADVANCED STATISTICAL ANALYSIS OF EPIDEMIOLOGICAL STUDIES Cox s regression analysis Time dependent explanatory variables Henrik Ravn Bandim Health Project, Statens Serum Institut 4 November 2011 1 / 53

More information

BIOSTATISTICAL METHODS

BIOSTATISTICAL METHODS BIOSTATISTICAL METHODS FOR TRANSLATIONAL & CLINICAL RESEARCH Cross-over Designs #: DESIGNING CLINICAL RESEARCH The subtraction of measurements from the same subject will mostly cancel or minimize effects

More information

A Comparison of Multiple Imputation Methods for Missing Covariate Values in Recurrent Event Data

A Comparison of Multiple Imputation Methods for Missing Covariate Values in Recurrent Event Data A Comparison of Multiple Imputation Methods for Missing Covariate Values in Recurrent Event Data By Zhao Huo Department of Statistics Uppsala University Supervisor: Ronnie Pingel 2015 Abstract Multiple

More information

Proportional hazards regression

Proportional hazards regression Proportional hazards regression Patrick Breheny October 8 Patrick Breheny Survival Data Analysis (BIOS 7210) 1/28 Introduction The model Solving for the MLE Inference Today we will begin discussing regression

More information

Biostatistics Workshop Longitudinal Data Analysis. Session 4 GARRETT FITZMAURICE

Biostatistics Workshop Longitudinal Data Analysis. Session 4 GARRETT FITZMAURICE Biostatistics Workshop 2008 Longitudinal Data Analysis Session 4 GARRETT FITZMAURICE Harvard University 1 LINEAR MIXED EFFECTS MODELS Motivating Example: Influence of Menarche on Changes in Body Fat Prospective

More information

Prerequisite: STATS 7 or STATS 8 or AP90 or (STATS 120A and STATS 120B and STATS 120C). AP90 with a minimum score of 3

Prerequisite: STATS 7 or STATS 8 or AP90 or (STATS 120A and STATS 120B and STATS 120C). AP90 with a minimum score of 3 University of California, Irvine 2017-2018 1 Statistics (STATS) Courses STATS 5. Seminar in Data Science. 1 Unit. An introduction to the field of Data Science; intended for entering freshman and transfers.

More information

Statistical Methods for Alzheimer s Disease Studies

Statistical Methods for Alzheimer s Disease Studies Statistical Methods for Alzheimer s Disease Studies Rebecca A. Betensky, Ph.D. Department of Biostatistics, Harvard T.H. Chan School of Public Health July 19, 2016 1/37 OUTLINE 1 Statistical collaborations

More information

CASE STUDY: Bayesian Incidence Analyses from Cross-Sectional Data with Multiple Markers of Disease Severity. Outline:

CASE STUDY: Bayesian Incidence Analyses from Cross-Sectional Data with Multiple Markers of Disease Severity. Outline: CASE STUDY: Bayesian Incidence Analyses from Cross-Sectional Data with Multiple Markers of Disease Severity Outline: 1. NIEHS Uterine Fibroid Study Design of Study Scientific Questions Difficulties 2.

More information

arxiv: v1 [stat.ap] 6 Apr 2018

arxiv: v1 [stat.ap] 6 Apr 2018 Individualized Dynamic Prediction of Survival under Time-Varying Treatment Strategies Grigorios Papageorgiou 1, 2, Mostafa M. Mokhles 2, Johanna J. M. Takkenberg 2, arxiv:1804.02334v1 [stat.ap] 6 Apr 2018

More information

On Measurement Error Problems with Predictors Derived from Stationary Stochastic Processes and Application to Cocaine Dependence Treatment Data

On Measurement Error Problems with Predictors Derived from Stationary Stochastic Processes and Application to Cocaine Dependence Treatment Data On Measurement Error Problems with Predictors Derived from Stationary Stochastic Processes and Application to Cocaine Dependence Treatment Data Yehua Li Department of Statistics University of Georgia Yongtao

More information

Causal Hazard Ratio Estimation By Instrumental Variables or Principal Stratification. Todd MacKenzie, PhD

Causal Hazard Ratio Estimation By Instrumental Variables or Principal Stratification. Todd MacKenzie, PhD Causal Hazard Ratio Estimation By Instrumental Variables or Principal Stratification Todd MacKenzie, PhD Collaborators A. James O Malley Tor Tosteson Therese Stukel 2 Overview 1. Instrumental variable

More information

Estimating the Mean Response of Treatment Duration Regimes in an Observational Study. Anastasios A. Tsiatis.

Estimating the Mean Response of Treatment Duration Regimes in an Observational Study. Anastasios A. Tsiatis. Estimating the Mean Response of Treatment Duration Regimes in an Observational Study Anastasios A. Tsiatis http://www.stat.ncsu.edu/ tsiatis/ Introduction to Dynamic Treatment Regimes 1 Outline Description

More information

Dynamic Prediction of Disease Progression Using Longitudinal Biomarker Data

Dynamic Prediction of Disease Progression Using Longitudinal Biomarker Data Dynamic Prediction of Disease Progression Using Longitudinal Biomarker Data Xuelin Huang Department of Biostatistics M. D. Anderson Cancer Center The University of Texas Joint Work with Jing Ning, Sangbum

More information

Statistics in medicine

Statistics in medicine Statistics in medicine Lecture 4: and multivariable regression Fatma Shebl, MD, MS, MPH, PhD Assistant Professor Chronic Disease Epidemiology Department Yale School of Public Health Fatma.shebl@yale.edu

More information

An R # Statistic for Fixed Effects in the Linear Mixed Model and Extension to the GLMM

An R # Statistic for Fixed Effects in the Linear Mixed Model and Extension to the GLMM An R Statistic for Fixed Effects in the Linear Mixed Model and Extension to the GLMM Lloyd J. Edwards, Ph.D. UNC-CH Department of Biostatistics email: Lloyd_Edwards@unc.edu Presented to the Department

More information

with the usual assumptions about the error term. The two values of X 1 X 2 0 1

with the usual assumptions about the error term. The two values of X 1 X 2 0 1 Sample questions 1. A researcher is investigating the effects of two factors, X 1 and X 2, each at 2 levels, on a response variable Y. A balanced two-factor factorial design is used with 1 replicate. The

More information

Analysis of Longitudinal Data: Comparison between PROC GLM and PROC MIXED.

Analysis of Longitudinal Data: Comparison between PROC GLM and PROC MIXED. Analysis of Longitudinal Data: Comparison between PROC GLM and PROC MIXED. Maribeth Johnson, Medical College of Georgia, Augusta, GA ABSTRACT Longitudinal data refers to datasets with multiple measurements

More information

Introduction to Statistical Analysis

Introduction to Statistical Analysis Introduction to Statistical Analysis Changyu Shen Richard A. and Susan F. Smith Center for Outcomes Research in Cardiology Beth Israel Deaconess Medical Center Harvard Medical School Objectives Descriptive

More information

Full likelihood inferences in the Cox model: an empirical likelihood approach

Full likelihood inferences in the Cox model: an empirical likelihood approach Ann Inst Stat Math 2011) 63:1005 1018 DOI 10.1007/s10463-010-0272-y Full likelihood inferences in the Cox model: an empirical likelihood approach Jian-Jian Ren Mai Zhou Received: 22 September 2008 / Revised:

More information

Analysing geoadditive regression data: a mixed model approach

Analysing geoadditive regression data: a mixed model approach Analysing geoadditive regression data: a mixed model approach Institut für Statistik, Ludwig-Maximilians-Universität München Joint work with Ludwig Fahrmeir & Stefan Lang 25.11.2005 Spatio-temporal regression

More information

Longitudinal Modeling with Logistic Regression

Longitudinal Modeling with Logistic Regression Newsom 1 Longitudinal Modeling with Logistic Regression Longitudinal designs involve repeated measurements of the same individuals over time There are two general classes of analyses that correspond to

More information

Multicollinearity and A Ridge Parameter Estimation Approach

Multicollinearity and A Ridge Parameter Estimation Approach Journal of Modern Applied Statistical Methods Volume 15 Issue Article 5 11-1-016 Multicollinearity and A Ridge Parameter Estimation Approach Ghadban Khalaf King Khalid University, albadran50@yahoo.com

More information

Sample size calculations for logistic and Poisson regression models

Sample size calculations for logistic and Poisson regression models Biometrika (2), 88, 4, pp. 93 99 2 Biometrika Trust Printed in Great Britain Sample size calculations for logistic and Poisson regression models BY GWOWEN SHIEH Department of Management Science, National

More information