This function returns a two-segment linear dose-response model, by
imposing a zero slope for the initial (left) segment,
detecting one breakpoint where the dose-response relation changes
to a positive slope (if such a breakpoint exists), then reporting
the breakpoint dose, its standard error and p-value, and plotting the model.
Usage
segment(dosecolumn = "", targetcolumn = "", data = NA)
Arguments
dosecolumn
Name of dose column of interest in dataframe.
targetcolumn
Name of response column of interest in dataframe.
data
Input dataframe.
Value
Returns the estimated breakpoint, standard error, and 90 percent confidence limits on the breakpoint,
as well as a plot of the estimated two-segment dose-response relationship.
Details
This function:
1) Attempts to identify one breakpoint using the 'segmented' function,
starting the search at the median of the input dose variable,
and imposing a zero-slope left-hand segment before any identified breakpoint.
2) If a breakpoint has been identified using this iterative approach, the p-value
is returned and model plotted; otherwise the function returns no breakpoint.
This function is currently only intended for use on continuous outcome data.
# Prints the breakpoint, its standard error, and 95% confidence limits# along with a plot of the estimated two-segment linear relationship:segment("dose", "MF_Log", data=DRdata)