dosresmeta (version 1.0)

dosresmeta: performing dose-response meta-analysis

Description

The function estimates a dose-response relation for each study included in the analysis, taking into account the correlation among observation. It estimates the covariance matrix according to one of the methods described by Greeland S. and Longnecker M., Hamling J., and Easton D. and use it to efficiently estimate the dose-response relation, as described by Orsini N. The estimates are then combined according to principle of multivariate random effect meta-analysis to provide an overall pooled functional relation.

Usage

dosresmeta(formula, study, se, cov, data, fcov, 
	method = "reml", psmethod = "h", alpha = 0.05)

Arguments

formula

an object of class "formula" (or one that can be coerced to that class): a symbolic description of the dose-response relation to be estimated for each study included in the analysis. The details of model specification are given under "Detai

study

vector of characters (2 or 1-dimensional) containing the names of id variable and type (of study) variable.

se

vector of characters (2 or 1-dimensional) containg the name of the variable reporting standar errors, or alternatively the names of the variables for the upper and lower bounds of log relative risks.

cov

2 dimensional character vector containg the names of the variables needed to approximate the coviariance matrix of log relative risks (n,cases): n is the size of the study for case-control and cumulative incidence data, the Person-Times for incidence ra

data

the dataset (or object coercible by "as.data.frame" to a data frame) containing the variables in the model formula and the variables specified by the previous statements.

fcov

an optional list of matrices that to specifies the covariance matrix of log relative risks within the same study. It must be the same length of the number of studies included in the analyses.

method

method to estimate a pooled dose response relation: "fixed" for fixed-effects models, "ml" or "reml" for random-effects models fitted through (restricted) maximum likelihood, and "mm" for random- effects models fitted through method of moments. The defa

psmethod

method to approximate the covariance matrix of log relative risks: "gl" for the method described by Greenland S. and Longnecker M., "h" for the method describe by Hamling J., "f" for the one presented by Easton D. In the last case a correct list of matr

alpha

the scalar inticating the alpha nominal values used in the pubblished data, by defaul equal to .05.

Value

• The function returns an object of class "mvmeta" in case of multiple studies, a list if estimating a dose-response relation for a single study.

Details

The function estimates efficientlly the dose-response relation specified in the formula by estimating the covariance matrix according to the method indicated in the psmethod. This method varies according to the type of the studies included in the analysis. The variable reporting the type of study must be specified in the study. The values for the variable must be "cc", "ir", "ci" for case-control, incidence-rate and cumulative-incidence data respectively.

References

Greenland, S., Longnecker, M. P. (1992). Methods for trend estimation from summarized dose-response data, with applications to meta-analysis. American journal of epidemiology, 135(11), 1301-1309. Orsini, N., Bellocco, R., Greenland, S. (2006). Generalized least squares for trend estimation of summarized dose-response data. Stata Journal, 6(1), 40. Gasparrini, A., Armstrong, B., Kenward, M. G. (2012). Multivariate meta-analysis for non-linear and other multi-parameter associations. Statistics in Medicine, 31(29), 3821-3839.

See Also

mvmeta,Hmisc

Examples

### FIRST EXAMPLE: Single case-control study
### linear trend using Hamling's method

# Inspect data
data(cc_ex)

# Fitting the model
(mod1<-dosresmeta(formula=logrr~dose,study="cc",
	cov=c("n","case"),se=c("loglb","logub"),data=cc_ex))

# Parameter Estimate
mod1$Param
# Approximated Covariance Matrix
mod1$Sigma


### SECOND EXAMPLE: Multiple studies
### linear trend using Greenland and Longnecker method
### method of moments

# Inspect data
data(ex_alcohol_crc)

# Fitting the model
(mod2<-dosresmeta(formula=logrr~dose,study=c("study","type"),
	cov=c("peryears","cases"),se="se",data=ex_alcohol_crc,
	psmethod="gl",method="mm"))

# Summarize the results
summary(mod2)
# residual and fitted values
residuals(mod2)
fitted(mod2)
# Pseudo counts and estimated covariance matrices
mod2$psnumber
mod2$Ccov