Colombia DHS 2015 Treatment-seeking behavior [message #14922] |
Tue, 15 May 2018 16:11 ![Go to next message Go to previous message](/theme/default/images/down.png) |
award
Messages: 1 Registered: May 2018
|
Member |
|
|
Hello, I'm trying to estimate treatment-seeking behavior for illness using the Colombia 2015 DHS. However, the responses in the published report are very different from what is in the household member recoded dataset I downloaded for Stata.
The results in question are on pages 130-131 of the final report: https://www.dhsprogram.com/pubs/pdf/FR334/FR334.pdf
These responses reflect what is in the annexed questionnaire on page 319, questions 88 and 91
Question 88 asks what type of problem the person has had in the past 30 days, with illness as the answer choice of interest.
Question 91 asks about treatment sought with these choices:
• Médico, terapeuta, institución de salud
• Farmacia o droguería
• Remedios caseros
• Auto medicación
• Nada
• Otro
In the Stata dataset, these appear to correspond to sh88a and sh91 (shperwgt is the weight var).
Var sh88a is a yes/no indicator of answering illness as a health problem
Var sh91 is about treatment for the health problem, but the response categories are:
• general doctor, specialist, dentist, therapist health institute
• alternative therapy
• field worker/nurse
• drugstore/pharmacy
• traditional medicine
• homemade remedies
The weighted percentages for sh91 match the table on page 130 exactly, in the same order, but clearly the headings are quite different.
Can this discrepancy be explained? Additionally, the Stata dataset does not have a choice for no treatment sought, and there are many entries with both a treatment choice in sh91 and a reason given for not seeking treatment in sh92. Finally, the tables in the report do not seem to limit to the response being an illness in question 88, but include answers to 88 of both yes and no.
Thanks very much in advance for any assistance you could give to help me better understand these data and ensure I'm using the correct estimates.
|
|
|