Average annual per cent change in segmented trend analysis
Computes the average annual per cent change to summarize piecewise linear relationships in segmented regression models.
aapc(ogg, parm, exp.it = FALSE, conf.level = 0.95, wrong.se = TRUE,
.vcov=NULL, .coef=NULL, ...)ogg |
the fitted model returned by |
parm |
the single segmented variable of interest. It can be missing if the model includes a single segmented covariate. If missing and |
exp.it |
logical. If |
conf.level |
the confidence level desidered. |
wrong.se |
logical, if |
.vcov |
The full covariance matrix of estimates. If unspecified (i.e. |
.coef |
The regression parameter estimates. If unspecified (i.e. |
... |
further arguments to be passed on to |
To summarize the fitted piecewise linear relationship, Clegg et al. (2009) proposed the 'average annual per cent change' (AAPC)
computed as the sum of the slopes (beta_j) weighted by corresponding covariate sub-interval width (w_j), namely
mu=sum_j beta_j w_j. Since the weights are the breakpoint differences, the standard error of the AAPC should account
for uncertainty in the breakpoint estimate, as discussed in Muggeo (2010) and implemented by aapc().
aapc returns a numeric vector including point estimate, standard error and confidence interval for the AAPC relevant to variable specified in parm.
exp.it=TRUE would be appropriate only if the response variable is the log of (any) counts.
Vito M. R. Muggeo, vito.muggeo@unipa.it
Clegg LX, Hankey BF, Tiwari R, Feuer EJ, Edwards BK (2009) Estimating average annual per cent change in trend analysis. Statistics in Medicine, 28; 3670-3682
Muggeo, V.M.R. (2010) Comment on ‘Estimating average annual per cent change in trend analysis’ by Clegg et al., Statistics in Medicine; 28, 3670-3682. Statistics in Medicine, 29, 1958–1960.
set.seed(12) x<-1:20 y<-2-.5*x+.7*pmax(x-9,0)-.8*pmax(x-15,0)+rnorm(20)*.3 o<-lm(y~x) os<-segmented(o, psi=c(5,12)) aapc(os)
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.