Menu ENN Search
Language: English Français

analysis

This question was posted the Assessment forum area and has 1 replies. You can also reply via email – be sure to leave the subject unchanged.

» Post a reply

Anonymous 1727

student of MPH in reproductive health

Normal user

30 May 2013, 21:28

I am going to perform a cross-sectional survey on time of initiation of complementary feeding among children aged six - 23 months of age. The department prefers to employ survival analysis methods. could you recommend how to employ this method for a cross-sectional (one time) survey?
thanks

Mark Myatt

Consultant Epideomiologist

Frequent user

31 May 2013, 09:36

The survival analysis / cohort life-table approach is a very useful approach to problems such as this. It is commonly applied to breastfeeding but there is no reason why it cannot be applied to complementary feeding. These methods make very efficient use of data (i.e. all subjects contribute useful data) so you do not need to use a large sample size. A sample size of about n = 150 should be sufficient for a basic analysis. There are simple and robust methods to test differences between survival curves so questions such as "Are boys weaned earlier than girls?" of "Have practices changed over time?" can be addressed. Useful summary statistics (e.g. averages, percentiles) can be produced from the cohort life-table.

It will probably be easier to think of your problem as cessation (non-survival, death) of exclusive breastfeeding when using a survival analysis / life-table approach. If a child is not currently exclusively breastfed then the survival time for exclusive breastfeeding is the child's age at which the mother reports introducing complementary foods (see below). If a child is currently exclusively breastfed then survival time for exclusive breastfeeding is the child's age and the child is treated as lost to follow-up (i.e. the standard practice in survival analysis of right-censoring losses to follow-up during a study and survivors at the close of a study).

You will have two types of children. These will be (1) children who have not initiated complementary feeding, and (2) children who have initiated complementary feeding. It is easy enough to identify the children in each group using simple questions asked of the mother. You will have current age in both groups. This is useful for group (1) but not group (2). For group (2) you need to be able to ascertain when complementary food were introduced. This is not straightforward because it is a recall question and you may end up with a lot of age-heaping at (e.g.) six months. Survival analysis is robust to modest levels of heaping but I think you may have problems with this application. Probably the only way to counteract this is to restrict the sample to children currently aged 0 - 11 months. This will shorten the recall period. You will still need to pay attention to avoiding age-heaping.

Another reason for shortening the recall period is to avoid having a "lagging indicator". If you use a wide age range (e.g. zero to 5 years) you will have a sample dominated by children who will have initiated complimentary feeding a long time ago and your result will be a sort of (slow to change) historic average. This is not what you want for M&E of programs.

I think there is a major flaw with your "among children aged six - 23 months of age" as this mean you only have recall data for children who initiated complementary feeding between 0 and six months. I think I would go for a 0 to 11 month group for this particular application (0-23 months for general IYCF).

Data analysis is straightforward and can be done by hand, in a spreadsheet, or in any mainstream stats package.

I would not recommend this method as the only method to use. I think you should collect continued breastfeeding, meal frequency, and dietary diversity data so you can make some assessment of the quality of feeding.

I hope this is of some help.

Back to top

» Post a reply