help oeratio


oeratio -- ratio of observed to expected outcomes


oeratio [{depvar predictvar}] [if] [in] [, options]

options Description ------------------------------------------------------------------------- Main seonly suppress the display; calculate only the standard error; programmer's option level(#) specify confidence level; if not specified system default is used. ------------------------------------------------------------------------- by is allowed; see [D] by.


oeratio calculates the ratio of the number of observed and expected outcomes of note. If depvar and predictvar (the variable specifying the individual probability of a positive outcome for each subject) are not specified, the command must follow a logit or logistic command.


+------+ ----+ Main +-------------------------------------------------------------

seonly restricts the calculation to only make calculations required to generate the standard error and ratio of observed to expected outcomes.

level set the confidence level. If not specified, the default confidence level is used; see [R] level.


The variance is calculated using the Bernoulli distribution: that is if p is the prediction variable: variance = sum(p*(1-p)). In its current form the standard error, confidence intervals and z-score calculated do not take into account the error of the model.

This command was developed in order to calculate the standardized mortality ratio (SMR) for intensive care units (ICU) when applying logistic mortality prediction models such as MPM, APACHE, PRISM or PIM models to unit data as part of benchmarking procedures.


--------------------------------------------------------------------------- Setup . webuse lbw2 Logistic regression on even records . logit low age lwt race2 race3 smoke ptl ht ui if mod(id,2)==0 Ratio of observed to expected outcomes when model applied to odd records . oeratio if mod(id,2)==1 --------------------------------------------------------------------------- externally validated logistic regression model covariates applied to lbw data . generate p = invlogit(-2 + (0.05* age)) . oeratio low p ---------------------------------------------------------------------------

Saved Results

logit saves the following in r():

Scalars r(N) number of observations r(predicted) number of outcomes predicted e(obs) number of outcomes observed e(ratio) ratio of observed to predicted outcomes e(se) standard error e(z) z score of the difference between observed and predicted outcome

Also see

Calculating standardised mortality rate (as opposed to ratio) [ST] stptime;