relpois_ag {popEpi}

Excess hazard Poisson model

Description

Estimate a Poisson Piecewise Constant Excess Hazards Model

Usage

relpois_ag(
  formula,
  data,
  d.exp,
  offset = NULL,
  breaks = NULL,
  subset = NULL,
  piecewise = TRUE,
  check = TRUE,
  ...
)

Arguments

formula

a formula with the counts of events as the response. Passed on to glm. May contain usage of the offset() function instead of supplying the offset for the Poisson model via the argument offset.

data

an aggre object (an aggregated data set; see as.aggre and aggre)

d.exp

the counts of expected cases. Mandatory. E.g. d.exp = EXC_CASES, where EXC_CASES is a column in data.

offset

the offset for the Poisson model, supplied as e.g. offset = log(PTIME), where PTIME is a subject-time variable in data. Not mandatory, but almost always should be supplied.

breaks

optional; a numeric vector of [a,b) breaks to specify survival intervals over the follow-up time; if NULL, the existing breaks along the mandatory time scale mentioned in formula are used

subset

a logical vector or condition; e.g. subset = sex == 1; limits the data before estimation

piecewise

logical; if TRUE, and if any time scale from data is used (mentioned) in the formula, the time scale is transformed into a factor variable indicating intervals on the time scale. Otherwise the time scale left as it is, usually a numeric variable. E.g. if formula = counts ~ TS1*VAR1, TS1 is transformed into a factor before fitting model.

check

logical; if TRUE, performs check on the negativity excess cases by factor-like covariates in formula - negative excess cases will very likely lead to non-converging model

...

any other argument passed on to glm such as control or weights

Value

A relpois object created using a custom Poisson family construct.

Author(s)

Joonas Miettinen, Karri Seppa

See Also

lexpand, poisson, glm

Other main functions: Surv(), rate(), relpois(), sir(), sirspline(), survmean(), survtab(), survtab_ag()

Other relpois functions: RPL, relpois(), rpcurve()

Examples

## use the simulated rectal cancer cohort
data(sire, package = "popEpi")
sire$agegr <- cut(sire$dg_age, c(0,45,60,Inf), right=FALSE)

## create aggregated example data
fb <- c(0,3/12,6/12,1,2,3,4,5)
x <- lexpand(sire, birth = bi_date, entry = dg_date,
             exit = ex_date, status=status %in% 1:2,
             breaks = list(fot=fb), 
             pophaz=popmort, pp = FALSE,
             aggre = list(agegr, fot))
             
## fit model using aggregated data
rpm <- relpois_ag(formula = from0to1 ~ fot + agegr,  data = x,
                  d.exp = d.exp, offset = log(pyrs))
summary(rpm)
 
## the usual functions for handling glm models work
rpm2 <- update(rpm, . ~ fot*agegr)
anova(rpm, rpm2, test="LRT")
AIC(rpm, rpm2)

## other features such as residuals or predicting are not guaranteed
## to work as intended.