Hi Roger,
thanks for your question. The lower and upper bounds are almost the
same, it depends on how many decimals you consider
. qui eclpci 10 1000
. ret list
scalars:
r(ub) = .0183903560420175
r(lb) = .0047953886961324
r(obs) = 10
macros:
r(cmd) : "eclpci"
. qui cii 1000 10, poisson
. ret list
scalars:
r(se) = .0031622776601684
r(mean) = .01
r(N) = 1000
r(ub) = .0183903558869087
r(lb) = .0047953887792498
The lower and upper bounds are not exactly the same because -eclpci- and
-cii- use different ways to get the results.
The command -cii- calculates the exact confidence interval using an
iterative algorithm (viewsource _crccip.ado), Newton's method, see [R]
-ci- pag 213 (Stata 9 manual). It uses the probability function
-dgammapdx- and -gammap-.
The command -eclpci- calculates an exact confidence interval without an
iterative algorithm, just two lines of code (viewsource eclpci.ado),
using the probability function -invgammap-, as described by
Daly, L., Simple SAS macros for the calculation of exact binomial and
Poisson confidence limits, Comput Biol Med 22:351-361, 1992.
Furthemore, the command -eclpci- has several options to calculate
approximate confidence intervals, as described by
Breslow NE, Day NE., Statistical Methods in Cancer Research: Volume II,
The Design and Analysis of Cohort Studies. Lyon: International Agency
for Research on Cancer, 1987, p.69.
It's quite simple to use -eclpci- to reproduce Table 1 pag 7 of
http://www.doh.wa.gov/Data/Guidelines/WordDocs/CI_guidelines.pdf
forv i = 0(1)99 {
eclpci `i'
}
I wrote -eclpci- with epidemiologists in mind and I hope it can be useful.
Best,
Nicola
Hi Nicola,
Could you clarify how this differs from official Stata's -cii #exposure
#events , poisson- ?
. eclpci 10 1000, format(%9.0g)
SMR: .01 95% CI[ .0047954, .0183904]
. cii 1000 10, poisson
-- Poisson Exact --
Variable | Exposure Mean Std. Err. [95% Conf. Interval]
---------+------------------------------------------------------
| 1000 .01 .0031623 .0047954 .0183904
Thanks,
Roger.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
