Language: English Français

# Before and an after holiday effect-on nutritional indices

This question was posted the Assessment and Surveillance forum area and has 6 replies.

### Mark Myatt

Frequent user

8 Oct 2015, 08:44

It is possible to assess the difference between prevalence estimates returned by two surveys. A simple approach is to estimate the standard error in each survey:

```    SE = (UCL - LCL) / (2 * 1.96)
```

then poole the two SEs:

```    PooledSE = sqrt(SE1^2 + SE2^2)
```

and calculate a z-test:

```    z = abs(prevalence1 - prevalence2) / PooledSE
```

The p-value is taken from the normal distribution (if abs(z) > 1.96 the p < 0.05).

You use an estimation approach:

```    difference = abs(prevalence1 - prevalence2)
95% CI = difference +/- 1.96 * PooledSE
```

I think you may have an issue with the sample size in the second "survey". The SE for the second "survey" will be wide. This means that you will not see a significant difference unless the difference is very large.

I hope this is of some use.

### Mark Myatt

Frequent user

8 Oct 2015, 09:13

Just remembered ...

You will be able to get a bit more power if you use the mean MUAC (or mean WHZ, WAZ, HAZ) and test for differences between means rather than prevalences.

I still think you will have sample size issues.

### Mark Myatt

Frequent user

8 Oct 2015, 15:20

Happy to help.

I think you mean the Composite Index of Anthropometric Failure (CIAF).

You can find the original article here.

### Mark Myatt

Frequent user

8 Oct 2015, 15:22

Also, see here.

### Mark Myatt

Frequent user

13 Oct 2015, 13:27

Thanks to Chris Hillbruner (Decision Support Advisor - FEWS NET) who writes ...

This is another interesting example of using CIAF, in this case to explore the relationship between mortality risk and the presence of multiple anthropometric deficits.