The DHS Program User Forum
Discussions regarding The DHS Program data and results
Home » DHS-7 Questionnaire Review - Archived » Maternal and Child Health » Sick child - avoid missing some who go to pharmacy/drug shop
Sick child - avoid missing some who go to pharmacy/drug shop [message #1671] Wed, 26 March 2014 16:47
jquinley is currently offline  jquinley
Messages: 3
Registered: March 2014
Location: USA
1.Public health programs have an increasing interest in knowing where and how children are treated for common childhood illness. It is important to know source of care, but it appears that the current core questionnaire can lead to substantial numbers of children being misclassified as "not seeking advice or treatment".

2a. In the Nigeria DHS 2008 it was found that over half of the time when it was reported that the case did not go for "advice or treatment" the child got a modern medicine. In the Nigeria Malaria Indicator Survey 2011 medicines were asked first, and then place of treatment after and the total percentage of children receiving treatment for fever from a particular location was much higher than the DHS. Most of the increase was in children getting treatment from pharmacy/drug shops. This suggests that many people do not consider going to buy medicines from a pharmacy/drug shop to constitute "going for advice or treatment" so this place is incompletely reported under the current questionnaire. A way to correct this without changing the current order of questions would be:

If 537 (drug given) is Yes, but 534 (went for advice or treatment) was No then add two questions,
537a. Did you go out to get this treatment or did you have it at home already?
(choices: Went out/At home/Don't know)
(If went out ask 537b, otherwise skip)
537b. Where did you get the treatment?
(same choices as for 534).

2b. The DHS asks about diarrhea separately from cough/fever. The same problem appears to occur (many children getting ORS or drugs when respondents answer that they did not go for advice and treatment). In order to preserve the current order of questions what would need to be added would be:

If 524 is any answer a-h or 522a is yes, but 518 is no, then ask:
524a Did you go out to get this treatment or did you have them at home already. (Went out/At home/Don't know) (If went out ask 524b, otherwise skip)
524b. Where did you get the treatment?
(same choices as for 519).

There are other ways to deal with this problem but they would require changing the overall sequence of questions in the survey.

3. Programs to improve management of childhood illness need to know where people go when children have an illness. If they base their analysis on the current survey of locations of care, the program may overestimate that proportion of children who "receive no advice or treatment" and under-estimate the role of providers, especially pharmacies/drug shops. The solution to these two situations may be different (e.g. an iCCM program vs. a program to train pharmacies/drug sellers to provide better advice and care), so the distinction is important. In the Nigerian Malaria Indicator Survey 2011 only about 10% of children with fever stay home with no treatment, but 50% go to pharmacies/drug shops. If you did the same analysis with the DHS 2008 you get more like an equal split between these two, which appears to be due to the way the question is asked.

Indicator in the DHS report. The current standard DHS report does not do a good job of providing data on where sick children get care. It would make sense to have tables showing place of treatment for children with various symptoms, instead of the current "children who go to a provider". If the above questions are added their results should be combined with the original questions to determine the distribution of places sick children get services.

4&5. Priority of questions. These questions would only come into play when there is a discrepancy between "seeking care" and "getting modern medicines". So it would only apply to a minority of sick children, who are themselves a minority of children in the survey. As such it should not add greatly to average survey time. If the survey is done electronically the skip pattern can be programmed so that it does not require time of the interviewer to check for the discrepancy before asking the question. As such, it would make a minor addition to the survey for a substantial improvement in source of care classification.

6. Although the specific experience was with the Nigerian DHS this problem may be common in many countries. A way to check would be to calculate for each country among children with a given symptom (fever or diarrhea would work better than ARI since they are more common) what percentage of sick children do not report going "for advice or treatment" and, of these, what proportion received a modern medicine. If the second proportion is substantial in many countries then this may be a useful addition to the core questionnaire.

John Quinley
A Promise Renewed Secretariat
Next Topic: Sick child - clinical providers working in pharmacies/drug shops
Goto Forum:

Current Time: Sun May 16 20:57:33 Coordinated Universal Time 2021