1) WHO's definition of diarrhoea is "the passage of three or more loose or liquid stools per day". However, the vast majority of surveys do not enquire whether the child had three (or more) loose stools in one day - most commonly, they just ask: “Has (NAME) had diarrhoea in the last 2 weeks?” (i.e. it is left up to the respondent to decide whether the loose / liquid stool really was diarrhoea or not). You have to decide whether your survey will assess and consider the frequency of loose stool (as suggested above) or whether it will follow the methodology commonly used by other surveys (such DHS or MICS). If you choose the latter option, ensure that the enumerators are very clear about (and can explain) what is understood as diarrhoea.
2) This indicator relies on accurate age assessment. Since people often do not remember the exact dates of their children’s birth, the data collectors should always verify the child’s age. This can be done by reviewing the child’s birth certificate, vaccination card or another document; however, since many caregivers do not have such documents (and since they can include mistakes), it is essential that your data collectors are able to verify the child’s age by using local events calendars. Read FAO’s Guidelines (see below) to learn how to prepare local events calendars and how to train data collectors in their correct use.
3) The stool of children aged up to 8 months is naturally loose and may not indicate diarrhoea. Therefore, include in your survey children aged 8 - 23 / 59 months only.
4) A standard recall period of two weeks is commonly used by most agencies. However, a number of pieces of research showed that such a long recall period leads to imprecise data that underestimates diarrhoea prevalence by up to 45% (for example, 1, 2) . If your data does not need to be comparable with other surveys (e.g. is used only for comparing your project's baseline and endline values), consider using a shorter recall period of 7 days (for maximum precision, you can also use a 3 day recall period though this will require increasing the sample size by approx. 70%). Always ensure that both baseline and endline surveys use the same recall period.
5) The prevalence of diarrhoea is often prone to seasonal differences. Do your best to collect baseline and endline data in the same period, otherwise it is very likely that they will not be comparable (i.e. saying little about the benefits of your project).