Re: st: Import data with complex structure

[Nick Cox <njcoxstata@gmail.com>]

This looks to me like a job for

either (1) customised input using -file-.

or (2) reading it in as a -str244- if possible and then processing
within Stata.

(1) requires at least moderate fluency with Stata programming and (2)
requires at least moderate fluency with Stata data management commands
and functions.

Your example implies missing fields in many cases.

Nick

On Wed, Mar 13, 2013 at 11:19 AM, Pietro Manuel Ferraro
<manuel.ferraro@hotmail.com> wrote:

> I have a .txt file I would need to input in Stata and I can't figure out how
> to do it. Basically, the file has data from about 1500 patients who
> undergone ambulatory blood pressure monitoring (ABPM). Some of these
> patients have more than one ABPM exam.
>
> The file is structured so that:
> - the first line has headers for the second line (name, surname, date of
> birth, sex, etc.)
> - the second line has patient-level data (see above)
> - the third line is empty
> - the fourth line has headers for the fifth line (date of measurement,
> instrument type, etc.)
> - the fifth line has ABPM-level data (see above)
> - the sixth line is empty
> - the seventh line has headers for the following lines (hour, systolic blood
> pressure, diastolic blood pressure, heart rate etc.)
>
> Then a variable number of lines follow, depending on the number of
> measurements for that particular ABPM exam.
>
> After the lines with measurement-level data, the file starts again with
> headers for the next patient. I found here on the list a similar topic in
> which someone suggested to treat each level separately and then merge the
> files together, but I'm unable to implement such strategy since: 1) the
> lower-level data (e.g., ABPM exams for each patient and measurements for
> each ABPM) don't have any identifier for the higher levels 2) the headers
> are repeated along the file 3) each patient may have a variable number of
> ABPM exams and each ABPM exam may have a variable number of measurements, so
> I can't instruct Stata to jump to pre-specified lines.
>
> The file appears to be like this (with fictitious data):
>
> =====start file=====
> Surname          Name  Address           Phone              Dob     Sex
> Weight             Height
> DOE   JOHN                         01/12/1925    M         67        170
>
> ABPM Date    Wake up(hr)    Rest(hr)           Therapy(hr)     Type
> Instrument
> 19/02/2003    7          23                    24        TM2430
>
> Date    Hour    Sys       Dia      HR
> 19/02/2003    09:22   120      80        60
> 19/02/2003    09:31   130      80        58
> 19/02/2003    09:45   120      80        60
> .
>
> Surname          Name  Address           Phone              Dob     Sex
> Weight             Height
> WRIGHT        JACK                          01/10/1974    M         88
> 180
> .
> =====end file=====
>
> Any idea how to proceed?
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/