Purpose: The purpose of this study was to retrospectively apply early event detection (EED) techniques based on the field of statistical process control to line-entry import data maintained by U.S. Customs and Border Protection (CBP) to determine if these methods could have detected a signal that would have indicated an anomaly in the supply chain for wheat gluten. An EED signal could have alerted regulatory authorities prior to illness reports.
Methods: Line entry data of wheat gluten for animal feed imported from China from 2003 – 2007 were obtained from CBP. The CUSUM control chart methodology was applied to daily quantities of wheat gluten imports after adjusting for systematic trends in the data. A threshold was applied to optimize the balance between sensitivity of signal detection and minimization of false signals.
Results: The analysis covered 543 shipments imported on 374 of 1,631 total days of data. The average quantity per shipment was 69,000 kg (std. dev., 40,000 kg). A threshold was chosen to achieve a minimum time between false signals of approximately one month. Multiple signals were detected in 2006: The first was detected seven months prior to the first known animal illnesses and the last was detected two months prior to the first known animal illnesses.
Significance: Monitoring of line-entry import data has the potential to alert us to anomalies in the supply chains for food ingredients. This approach can be used by the private sector and regulatory authorities to monitor imports in real time to target limited inspection and testing resources to the highest risk food ingredients. Moving forward, integrated EMA scoring models that incorporate trade data, QA methods, supply chain structure, and pricing can predict EMA potential.