METHODS: Caregivers were asked – at baseline and six months later at program conclusion – to quantify their child’s ED visits and IP stays for treatment of asthma for the previous six months. Medicaid claims data were available from which ED visits and IP stays with a diagnosis of asthma could be quantified. Level of agreement between proxy-report and claims data was measured for ED and IP at baseline and at six months.
RESULTS: One-hundred-thirty-four (134) caregivers completed the program and had continuous enrollment in Medicaid during both six-month recall periods. Overall, caregivers reported a total of 183 ED visits and IP stays combined. Per Medicaid claims data, there were a total of 49 events with asthma as the primary diagnosis. At baseline, 44.8% of caregivers reported more ED visits for asthma than the number of ED visits indicated by Medicaid claims while for 53.0% there was agreement between the two sources. At six months, 12.7% of caregivers reported more ED visits than were indicated by claims data and for 82.8% there was agreement. For IP stays, 9.0% of caregivers reported more than claims data at baseline and there was 91.0% agreement. At six months, only 2.2% of caregivers reported more than was indicated by Medicaid and there was 97.0% agreement. Results were similar when the definition of an ED or IP event was expanded to include any diagnosis of asthma.
CONCLUSIONS: Overall, caregivers reported more than three times as many ED visits and IP stays as were indicated by Medicaid claims. This potential over-reporting was greater at baseline than at six months, leading to a bias away from the null, which could lead to the conclusion that the intervention was more effective than it was. In addition, potential over-reporting was greater for ED visits than for IP stays. This could be due to IP stays being rarer and more memorable events than ED visits. Researchers need to be cognizant of the potential for reporting bias and design studies to minimize its likelihood.