EVALUATING THE VALIDITY OF PARENTAL REPORTS OF HEALTHCARE UTILIZATION FOR CHILDREN WITH SUSPECTED GENETIC DISORDERS OF UNKNOWN CAUSE
Method: Parents of children enrolled in the CAUSES Research Clinic (a genome-wide sequencing program at British Columbia Children’s Hospital (BCCH)) were invited to complete an online survey on their child’s utilization of healthcare resources. Parents were asked to report the number and nature of appointments the child attended at BCCH in the 6 months prior to survey completion, and comparable data were extracted from patients’ charts using the electronic medical record (EMR) system at the hospital.
Two coders rated whether each entry in the EMR was reported in the survey or not and whether there was evidence in the EMR for each appointment reported in the survey. Inter-coder agreement was calculated using Cohen’s kappa, and entries were coded as matched only if both coders agreed.
Result: Forty-four (44) respondents completed the relevant items, reporting 111 appointments at BCCH in the 6 months prior to the survey, and providing short descriptions for 99 of those appointments. EMR data included a total of 188 appointments in the same time period. Inter-coder agreement was high for both the EMR data (Cohen’s κ = 0.80, 95% CI = [0.71 to 0.88]) and the survey data (κ = 0.83, CI = [0.70 to 0.96]).
74% (73/99) of the appointments described in the survey also appeared in the EMR, but only 41% (77/188) of appointments recorded in the EMR were described in the survey. The correlation between the total number of appointments reported for each child in the survey and the number of appointments recorded in the EMR was 0.59 (95% CI = [0.35 to 0.75]).
Conclusion: Parental surveys appear to be a useful method for measuring relative levels of healthcare utilization for pediatric patients with chronic disorders. The reported resource use is generally supported by patients’ EMRs, and the individual-level utilization volume reported in the survey is highly correlated to that recorded in the EMR. The total level of utilization appears to be underreported, but this result may reflect redundant entries in the EMR and is sensitive to outliers with very high resource use for whom only a subset of relevant appointments was reported.