The effectiveness of U.S. public health surveillance systems for situational awareness during the 2009 H1N1 pandemic: a retrospective analysis

Abstract

Background: The 2009 H1N1 outbreak provides an opportunity to learn about the strengths and weaknesses of current U.S. public health surveillance systems and to identify implications for measuring public health emergency preparedness.

Methodology/principal findings: We adopted a "triangulation" approach in which multiple contemporary data sources, each with different expected biases, are compared to identify time patterns that are likely to reflect biases versus those that are more likely to be indicative of actual infection rates. This approach is grounded in the understanding that surveillance data are the result of a series of decisions made by patients, health care providers, and public health professionals about seeking and providing health care and about reporting cases to health authorities. Although limited by the lack of a gold standard, this analysis suggests that children and young adults are over-represented in many pH1N1 surveillance systems, especially in the spring wave. In addition, the nearly two-month delay between the Northeast and the South in the Fall peak in some surveillance data seems to at least partially reflect regional differences in concerns about pH1N1 rather than real differences in pH1N1 infection rates.

Conclusions/significance: Although the extent of the biases suggested by this analysis cannot be known precisely, the analysis identifies underlying problems with surveillance systems--in particular their dependence on patient and provider behavior, which is influenced by a changing information environment--that could limit situational awareness in future public health emergencies. To improve situational awareness in future health emergencies, population-based surveillance systems such as telephone surveys of representative population samples and seroprevalence surveys in well-defined population cohorts are needed.

Figure 1. Google “Swine Flu” indices for the Northeast, South and entire United States Panel A.

Google Insights “Swine Flu” activity and news indices, U.S., April 12, 2009–January 2, 2010. Panel B. Google Insights “Swine Flu” activity index, U.S., Georgia (GA), and Massachusetts (MA), August 2, 2009–January 2, 2010. Source: Google Insights search for “swine flu” at http://www.google.com/insights/search/#q=SWINE%20FLU&geo=US&date=1%2F2009%2012m&cmpt=q.

Figure 2. pH1N1 infection, hospitalization, and death rates.

Panel A. Aggregate hospitalization and death reporting activity (AHDRA) hospitalization and death rates per 100,000 population by age group, laboratory-confirmed pH1N1 influenza infection—United States, August 2009–February 2010. Source: Jhung, 2011. Panel B, C & D. 2009 H1N1-Related Deaths, Hospitalizations and Cases, U.S. April 2009–January 16, 2010. Author's calculations based on CDC EIP program estimates: Updated CDC estimates of 2009 H1N1 influenza cases, hospitalizations and deaths in the United States, April 2009–April 10, 2010 Available from: http://www.cdc.gov/h1n1flu/estimates_2009_h1n1.htm.

Figure 3. Influenza-related emergency department visits and hospitalizations.

Panel A. Rate of ILI syndrome visits (based on chief complaint) to New York City emergency departments by age group, April 1, 2009–July 6, 2009. Panel B. Laboratory-confirmed H1N1 hospital admissions and emergency department (ED) Visits for ILI in New York City, April 1, 2009–July 6, 2009. Source: New York City Department of Health and Mental Hygiene Health Alert #27: Pandemic (H1N1) 2009 influenza update, revised reporting requirements and testing procedures Available from: http://www.nyc.gov/html/doh/downloads/pdf/cd/2009/09md27.pdf.

Figure 4. Outpatient influenza-like illness surveillance.

Percentage of visits for ILI Reported by the U.S. Outpatient Influenza-like Illness Surveillance Network (ILINet), National Summary 2008–09 and Previous Two Seasons. Source: CDC Flu View, 2008–2009 Influenza Season Week 25 ending June 27, 2009. Available from: http://www.cdc.gov/flu/weekly/weeklyarchives2008-2009/weekly25.htm.

Figure 5. Laboratory-confirmed influenza cases reported to CDC by U.S.

WHO/NREVSS collaborating laboratories, national summary, 2008–09. Source: CDC Flu View, 2008–2009 Influenza Season Week 25 ending June 27, 2009. Available from: http://www.cdc.gov/flu/weekly/weeklyarchives2008-2009/weekly25.htm.

Figure 6. Influenza-like illness surveillance for the Northeast, South and entire United States.

Panel A. Percentage of visits for influenza-like illness (ILINet), U.S. and South, July 31, 2009–January 8, 2010; Google Flu Trends index, U.S. and Georgia (GA), August 2, 2009–January 10, 2010; scaled DiSTRIBuTE ILI trends, Georgia (GA), October 3, 2009–January 9, 2010. Panel B. Percentage of visits for influenza-like illness (ILINet), U.S. and Northeast (NE), July 31, 2009–January 8, 2010; Google Flu Trends index, U.S. and Massachusetts (MA), August 2, 2009–January 10, 2010; scaled DiSTRIBuTE ILI trends, Massachusetts (MA), October 3, 2009–January 9, 2010. Sources: author's calculations based on CDC, Google, and DiSTRIBuTE data available from: http://www.cdc.gov/h1n1flu/cdcresponse.htm, http://www.google.org/flutrends/, and http://isdsdistribute.org/moreinfo.php.