The 131st Annual Meeting (November 15-19, 2003) of APHA

Abstract #69792

David L Buckeridge, MD, MSc¹, Paul Switzer, PhD², and Mark A Musen, MD, PhD¹. (1) Stanford Medical Informatics, Stanford University, MSOB X-215, 251 Campus Drive, Stanford, CA 94305-5479, 650 723 6979, david.buckeridge@stanford.edu, (2) Statistics, Stanford University, Sequoia 136, Stanford, CA 94305

Background: Pre-diagnostic or syndromic surveillance necessitates following multiple illness categories concurrently. A modular approach to surveillance can facilitate the evaluation and implementation of methods for pre-diagnostic surveillance. We present a modular approach to space-time surveillance of multiple illness categories and examine the performance of one implementation for surveillance of San Francisco 911 data.

Methods: We define a set of illness categories, to which individual events, such as 911 calls, are mapped. We then decompose the surveillance task into (1) a local operation and (2) a global operation. The local operation generates expected values, observed values, and aberrancy measures by spatial unit and time step. The global operation uses the results of the local operation to generate overall and illness category aberrancy measures at each time step across the region under surveillance.

Results: We implemented the local operation by using spatial time-series forecasting and simulation to determine individual, and joint, illness category aberrancy measures. For the global method, we examined specific illness categories only when the joint aberrancy measure exceeded a threshold determined by an acceptable false positive rate. When applied to San Francisco 911 data for respiratory, cardiovascular and trauma illness categories, this implementation identified an overall and respiratory aberrancy one week prior to a rise in Northern California hospital admissions for influenza related conditions.

Conclusion: A modular surveillance approach allowed implementation and evaluation of space–time analytic methods. One implementation identified large-scale respiratory illness from 911 data seven days prior to when traditional indicators would have done so.

Learning Objectives:

At the conclusion of the session, the participant (learner) in this session will be able to:
• Articulate the challenges in conducting space-time surveillance of data sources with multiple disease categories.
• Recognize how a modular approach can facilitate development and evaluation of methods for pre-diagnostic surveillance.
• Assess the suitability of analytic methods for pre-diagnostic surveillance.

Keywords: Surveillance, Bioterrorism

Related Web page: smi-web.stanford.edu/projects/biostorm/

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.