Presumably you are not measuring dead children. In that case you only have living children in your cross-sectional survey sample. You may have a nutrition survey and a mortality survey done at the same time. One problem here is that the mortality data will be in the past and the malnutrition will be in the present. So your analysis will assume that death causes malnutrition. Simple correlation without causality will also be tricky and death may have carried away the malnourished giving high mortality with low prevalence of wasting. Another problem is that you only have summary data. This will be the mortality rate and the prevalence of wasting. Your sample size is one! Am I misunderstanding your question?

I was thinking that he was referring to analyzing correlation between,eg. diarrhoea and wasting.I think you may have mistakenly read morbidity as mortality,Mark.

Yes. You're right! Oops! OK. This time for real! Yes you can do a "correlation" analysis. For (e.g.) you can do a two-by-two table if you have a binary variable like diarrhoea and a binary outcome such as "wasted". The effect size summary would be the risk ratio (relative risk) and a chi-square test for statistical significance.. An analysis of a continuous variable (e.g. MUAC) and a binary variable (e.g. diarrhoea) might be more powerful. The effect size might be Cohen's d and a t-test (or non-parametric equivalent) could be used for statistical significance.

Thanks Mark for the prompt response. I meant correlation between morbidity or IYCN outcome or even WASH against nutrition outcome. This is for a survey done at the same time in the same households

Thanks Mark