Most, if not all, disease surveillance systems are federated in the sense that hospitals, doctors’ offices, pharmacies are the source of most surveillance data. Although a health department may request or mandate that these organizations report data, we are not aware of any requirements about the method of data collection or audits or other measures of quality control.
Because of the heterogeneity and lack of control over the processes by which the data are generated, data sources in a federated disease surveillance system are black boxes the reliability, completeness, and accuracy of which are not fully understood by the recipient.
In this paper, we use the variance-to-mean ratio of daily counts of surveillance events as a metric of data quality. We use thermometer sales data as an example of data from a federated disease surveillance system. We test a hypothesis that removing stores with higher baseline variability from pooled surveillance data will improve the signal-to-noise ratio of thermometer sales for an influenza outbreak.
Objective
We developed a novel method for monitoring the quality of data in a federated disease surveillance system, which we define as ‘a surveillance system in which a set of organizations that are not owned or controlled by public health provide data.’