Clinical Question: Can a history obtained from parents of a sick child confirm or rule out clinically significant dehydration?

Setting: Emergency department

Study Design: Cohort (prospective)

Synopsis: The researchers recruited 135 parent-child pairs presenting to an emergency department with the parent reporting vomiting, decreased intake of fluids, or diarrhea in the child. The otherwise normal children were between one month and five years of age. Most (93 percent) of the parents were high school graduates, and all spoke and read English.

Before being evaluated, parents completed a computer-based interview asking about specific signs of dehydration. Nurses and emergency physicians were not given the results of this survey but completed their own history. The ultimate fluid deficit in each child was calculated by comparing weight on presentation with a second weight obtained two to three days after the initial visit. Clinically significant dehydration was defined as a fluid deficit of at least 5 percent and was present in 15.5 percent of the children for whom they had complete weight data (54 percent).

In general, the history obtained from parents was sensitive (range: 73 to 100 percent) but not specific (range: 0 to 49 percent). In other words, the absence of the sign was highly predictive of the child not having clinically important dehydration, whereas the presence of the sign did not confirm dehydration. Parent reports of no decreased oral intake, no decrease in urine output, or no history of vomiting were highly predictive of clinically insignificant dehydration (sensitivity for each: 100 percent). Characterization of a child as not “ill appearing”by either the parent or the nurse was highly predictive of clinically insignificant dehydration (sensitivity: 91 and 90 percent). Conversely, reports of the presence of a sunken fontanelle by either the parent or nurse were predictive of dehydration (specificity: 82 and 90 percent). The question remains as to whether a face-to-face interview would be as reliable.

Bottom Line: Parents of children with suspected dehydration because of acute illness can be useful in ruling out, but not ruling in, clinically significant dehydration. Children did not have clinical dehydration if parents did not report a decrease in oral intake, a decrease in urine output, or an episode of vomiting. Similarly, characterization of the child as “not ill-appearing” also ruled out dehydration. The report of a sunken fontanelle predicted dehydration. (Level of Evidence: 1c)