jointmeta1.object: Fitted 'jointmeta1' object
In mesudell/joineRmeta: Joint Modelling for Meta-Analytic (Multi-Study) Data

Description Usage Format Value Author(s) See Also

An object returned by the jointmeta1 function, inheriting from class jointmeta1 and representing a fitted joint model for a single longitudinal and a single time-to-event outcome for data from multiple studies. Objects of this class have methods for the generic functions confint, fixef, formula and ranef. Additionally rancov allows the user to extract the estimated covariance matrices for the zero mean random effects.

1	jointmeta1.object

An object of class NULL of length 0.

A list with the following components.

coefficients

a list with the estimated coefficients. The components of this list are:

fixed

the list of fixed effects for sub-models contained in the joint model. The components of this list are:

longitudinal: a data frame containing the estimated fixed effect coefficients from the longitudinal sub-model
survival: a numeric vector containing the estimated fixed effect coefficients from the longitudinal sub-model

random

the list of estimates random effects estimated by the joint model. The components of this list are:

random_ind: a list of matrices of length equal to the number of studies in the dataset. Each matrix has number of columns equal to the number of individual level random effects, and number of rows equal to the number of individuals in the study. As jointmeta1 insists on the presence of random effects at the individual level, this item will always be present.
random_stud: a matrix with number of columns equal to the number of study level random effects, number of rows equal to the number of studies in the dataset. This item is only present if study level random effects are specified in the model fit.

latent

a numeric containing the estimates of the latent association parameters for each level of the random effects. The association parameter for the individual level random effects is labelled gamma_ind_0, and for the study level random effects is labelled gamma_stud_0.

sigma.e

a numeric holding the estimate of the variance of the measurement error variance

rand_cov

a list containing the covariance matrices for the random effects included in the model. The covariance matrix for the individual level random effects is labelled D. If study level random effects are included in the model, the covariance matrix for the study level random effects is also included in the list, labelled A.

hazard

if strat = FALSE in the function call for jointmeta1 then this is a numeric vector containing the common baseline across all studies. If strat = TRuE then this is a list of numeric vectors, each of which is the baseline hazard for each study in the dataset.

loglik

a list containing the overall likelihood for the joint model (labelled jointlhood), and the portions of the likelihood attributable to each sub-model (jointy for the longitudinal component and jointn for the survival component).

numIter

the number of EM algorithm iterations completed during the fitting of the joint model

convergence

a logical value, takes a value of TRUE if convergence was achieved within the set maximum number of iterations, FALSE otherwise.

sharingstrct

a character string denoting the specified sharing structure used in the joint model. Currently only 'randprop' is supported, denoting zero mean random effects sharing structure (see Wulfsohn and Tsiatis (1997)).

sepests

A list containing estimates from the separate longitudinal and survival analyses. If separate results are not requested, the elements of the list are set to 'No separate results requested'. However, if separate analyses are requested in the jointmeta1 function call, the components of this list are:

longests

a list containing estimates from the initial longitudinal fit. The components of this list are:

beta1: a data frame of the estimates of the fixed effects from the longitudinal sub-model
sigma.e: the value of the variance of the measurement error from the longitudinal sub-model
D: the estimate of the covariance matrix for the individual level random effects. Individual level random effects are always included in the joint model
A: the estimate of the covariance matrix for the study level random effects. This is only present if study level random effects are specified in the jointmeta1 function call.
log.like.long: the numeric value of the log likelihood for the initial longitudinal model.
randstart.ind: a list of the conditional modes of the individual level random effects in each study given the data and the estimates of the separate longitudinal model parameters
randstart.ind.cov: a list of the conditional covariance matrices for each individual for the individual level random effects given the data and the estimates of the separate longitudinal model parameters
randstart.stud: a data frame containing the conditional modes of the study level random effects given the data and the estimates of the separate longitudinal model parameters. This is only present if study level random effects were specified in the jointmeta1 function call.
randstart.stud.cov: a list of conditional covariance matrices for each study for the study level random effects given the data and the estimates of the separate longitudinal model parameters. This is only present if study level random effects were specified in the jointmeta1 function call.
modelfit: the initial longitudinal model fit. The model has the same specification as the longitudinal sub-model for the joint model, fitted using the lmer function from package lme4

survests

a list containing estimates from the initial survival fit. The components of this list are:

beta2: vector of the estimates of the fixed effects included in the survival model.
haz: if strat = TRUE then this is a list of numeric vectors of length equal to the number of studies in the dataset, giving the study specific baseline hazard. If strat = FALSE then the baseline is not stratified by study, and this is one numeric vector giving the common baseline across studies.
rs: a counter to indicate the last how many unique event times had occured by the individual's survival time - this is for use during further calculation in the joint model EM algorithm. If a stratified baseline this is a list of numerical vectors, whereas if the baseline is not stratified this is a single numeric vector.
sf: the unique event times observed in the dataset. If a stratified baseline this is a list of numerical vectors, whereas if the baseline is not stratified this is a single numeric vector.
nev: a counter of the number of events that occur at each event time.If a stratified baseline this is a list of numerical vectors, whereas if the baseline is not stratified this is a single numeric vector.
log.like.surv: a numeric containing two values, the log-likelihood with the initial values and the log-likelihood with the final values, see coxph.object
modelfit: the initial survival model fit. The model has the same specification as the survival sub-model for the joint model, fitted using the coxph function from package survival

sep.loglik

a list containing the log-likelihoods estimated from the separate analyses. It contains three elements, namely seplhood - the sum of the log-likelihoods from the separate longitudinal and the separate survival analyses, sepy - the log-likelihood from the separate longitudinal analysis, sepn - the log-likelihood from the separate survival analysis.

data

the jointdata object containing the data the joint model was fitted to

call

the function call supplied to the jointmeta1 function.

numstudies

an integer containing the number of studies present in the data used to fit the joint model

n.bystudy

a numeric vector containing the number of individuals present in each study in the data used to fit the joint model. This will be less than the number of individuals in the supplied dataset, if missing data is present in variables included in the model.

missingids

the ids of any individuals excluded from the analysis due to missing data

nobs

a table containing the number of longitudinal measurements supplied by each study in the data used to fit the model. This will be less than the number of longitudinal measurements in the dataset supplied to the function call, if missing data is present in variables included in the model

Maria Sudell (mesudell@liverpool.ac.uk)

jointmeta1.

mesudell/joineRmeta documentation built on Jan. 24, 2020, 6:06 p.m.