Validation of a Dehydration Scoring System

RESEARCH DESIGN AND METHODS

Objectives:

Objective 1: Validity; Objective 2: Reliability; Objective 3: To obtain preliminary data in determining the effectiveness of the Dehydration Scoring System in predicting resource utilization and disposition.

Study Design:

This will be a prospective study of patients presenting to the Emergency Department (ED) with clinical signs of dehydration. All previously healthy children aged 1 month (mo) up to 16 years of (yo) age with vomiting, diarrhea, or poor oral (PO) intake from presumed gastroenteritis will be eligible for inclusion after obtaining informed consent and assent if applicable. Exclusion criteria will include a history of cardiac or renal disease, diabetes mellitus (DM), malnutrition or failure to thrive (FTT), and any malabsorptive disorder or short gut syndrome . Children will further be excluded if they are clinically unstable, requiring immediate life saving interventions or if no parent or legal guardian is available for consent.

Clinical measures:

Upon obtaining informed consent, a data sheet will be started to gather study weights, dehydration scores at specified intervals, patient demographics and key clinical findings, as well as resource utilization (none, oral rehydration therapy (ORT), intravenous (IV) therapy, observation status, admission to the hospital). Information regarding whether the patients have received the rotavirus vaccine and how many doses will also be collected.

As part of standard of care, patients will be weighed using either a single infant or pediatric scale located in triage at presentation to the Pediatric ED (PED) (Time 0) (age appropriate infant scale versus standing stadiometer in Triage). Infants will be weighed naked and older children in a hospital gown. Change in weight at discharge or follow up (f/u), compared to baseline is an adequate measure of dehydration that has been used standardly in studies of this nature.

Emergency department nurses will obtain a dehydration score when the patient is transferred to a treatment room (time 0). Prior to initiation of treatment, and as close to the nurse's Dehydration Score as is possible, a clinician (Faculty, fellow, nurse practitioner or physician assistant) will also score the patient as they arrive into the treatment area prior to initiation of treatment. Clinicians will be masked to the initial Dehydration Score as scored by the nurse. Subjects will continue to receive Dehydration Scores every hour after the initiation of treatment until a disposition has been determined (i.e. discharge home or admit to hospital) (Times T1,2,3…). At discharge, or at time decision to admit has been made, the children will be weighed on the same scale as initially used. Subjects will be asked to return in 1 week (+/- 3 days) for wellness (dehydration) score, repeat weight (same scale), and further continuation of historical and physical examination findings. Research or clinical nurse will obtain the final follow-up Dehydration Score.

STATISTICAL METHODS

Statistical Analysis

Objective 1: The validity of the scale, in comparison to the gold standard of change in weight, will be assessed by comparing the ratio of the change in dehydration score (discharge minus baseline) to the ratio change in weight (discharge minus baseline) for all subjects. A comparison will also be made between the ratio change in score and ratio change in weight using the measurements at follow-up. Lin's concordance correlation coefficient will be estimated for both comparisons. This method of measuring correlation is preferred over Pearson's correlation coefficient, which is limited to estimating the degree of linearity between the two tests and not the accuracy.

Objective 2: Repeatability of the Dehydration Scale by different raters (nurse and clinician at baseline) will be tested by obtaining a concordance correlation coefficient.

Objective 3: Correlation between Dehydration Score and resource utilization/disposition outcomes. We will characterize the mean dehydration scores across treatments (i.e. resource utilization), and roughly identify cut-points (natural or by clinical significance) and estimate predictive values.

Sample size calculation: The primary aim of the proposed study is to evaluate the validity of the dehydration scale compared to the gold standard of change in weight. The hypothesis is that the correlation of the two measures will produce a correlation of at least 0.95, exceeding the 0.75 cut-off to demonstrate an 'excellent' correlation. To detect this difference at 80% power (alpha=0.05), 54 subjects will be sufficient. Assuming a significant loss to follow-up at one week (~40-50%), a total of 100 subjects will be targeted for enrollment to ensure that 54 subjects are available to test the hypothesis of a correlation between change in score and weight at follow-up. This estimated sample size will further be sufficient to evaluate interobserver reliability and obtain a concordance correlation coefficient.