How well does a telephone questionnaire measure drinking water intake?

Background: Few studies have assessed the validity or reliability of drinki ng water intake estimations. In studies investigating potential waterborne disease, this is very important. Methods: We carried out a validity and reliability study to evaluate water intake estimations. The validity part of the study compared a telephone que stionnaire to a diary. The reliability part of the study compared two telep hone questionnaires. Results: 253 participants were recruited. The most commonly consumed water intake categories were plain and boiled tap water For persons aged less tha n 12 years, the median volume of plain water consumption was 2.9 glasses/da y according to the diary. For persons aged 12 years and over, the median vo lume of plain water consumption was 3.0 glasses/day and for boiled water 3. 0 glasses/day according to the diary. There was negligible systematic over or under reporting by participants for the telephone questionnaire in compa rison to the diary and tor the comparison of both telephone questionnaires. Water intake estimates were converted to categorical variables and the per cent agreement and weighted kappa statistics were calculated. For both the validity and reliability parts of the study the per cent agreement was 57- 78% and the weighted kappa was 0.57-0.81. Conclusion: Water intake estimates show only moderate levels of validity an d reliability when using a telephone questionnaire in comparison to a diary . Implications: Investigators of potential waterborne disease or nutritional research should exercise some caution in drawing conclusions from minor dif ferences in drinking water intake that are obtained from a telephone questi onnaire.