When doing routine reporting we never know in advance if there are some observations for a given table in analysis.

Therefore: read rawfile_from_patientsystem.csv /close means y sex

would stop if no persons had any data in variable y The technique in analysis to control this is:

read rawfile_from_patientsystem.csv /close * first count the number (with /q the result is not shown) count if y <> . /q imif $count > 0 then means y sex else type "No data in y or sex" endif

Another useful technique is to find the period in the data. By sorting on the date variable and save the result as a constant this can be used in a "report".

read rawfile_from_patientsystem.csv * sort on a given variable containing date: sort entrydate define first ________________ global define last ________________ global first = string(vmeb001[1]) count last = string(vmeb001[$count]) type "test - @first - @last" /h3

If one adds then "cls" to clear the screen or "set echo = off", then only the result of the type command is shown. But only do this when it is all working. I am writing on a manual of "how to write menusystems" for release later this year. The techniques above will be part of that.

Jens Lauritsen EpiData Association