Wednesday, November 02, 2005

Someone to watch over me

Flu season (i.e., "seasonal influenza" as opposed to pandemic influenza) is approaching. When? Where? What? Important questions, since they differ from year to year. A couple of years ago the season started in November but last year it was much later, January. Finding the answer to these questions is an important function of the country's public health infrastructure. It is disease surveillance.

Surveillance is always the poor sister in public health , along with vital records and vital statistics. Essential, supremely important but with no strong constituency and easy pickings for politically appointed bureaucrats looking for low hanging fruit for budget cuts. Go after the weakest, leave the strong guys (drug companies) alone.

OK, enough of that. Flu surveillance. At least how it's practiced now. Allegedly the 300 plus page Pandemic Flu Plan due out today will propose some grand Biosurveillance Program, but it will likely be just a modification of the current one. We'll have to see.

Most of the current system is overseen by CDC. The CDC website describes seven components, overlapping and non of them perfect. Internationally there is a network of 120 designated WHO and National "reference" laboratories used for identifying and confirming viral specimens beyond the resources of more routine facilities. They report the total number of respiratory viruses identified and the number of those that are either influenza A or B.
  • In the US, specifically, there is also a network about 1000 health care providers that report the total number of patients they see and of those, the number that have what is termed "influenza like illness" (ILI). An ILI is an episode of illness where the patient presents with fever over 100o F., a sore throat or a cough, and have no other known cause (such as strep throat). During periods when there is no flu outbreak, most ILIs are not influenza, but in the setting of a seasonal outbreak, over 80% of ILIs are actually influenza (this is a simple consequence of Bayes Theorem in classical probability and will not detain us here). CDC weights the percentage of ILIs by state population and compares it to a baseline of "normal" ILIs (2.5%), plus or minus some known variation.

  • Another component uses death certificates filed in 122 cities throughout the US. The percent of deaths in which pneumonia or infuenza is given as an underlying or contributing cause of death is compared with a baseline and epidemic threshold calculated each week. These data are usually two to three weeks old and there is an additional two weeks from reporting to posting on the CDC website.

  • In addition to the 122 city data, state health departments report each week on a five point scale whether there is no flu activity, sporadic activity, local activity, regional activity or widespread flu activity in the state. Each level has a definition in terms of laboratory confirmed cases.

  • Three newer components are also used specifically for pediatric influenza. A Nationally Notifiable Disease Surveillance System reports on influenza-associated mortality in laboratory confirmed cases of influenza in children less than 18 years old. For non-fatal pediatric cases, the EIP Influenza Project conducts surveillance for laboratory-confirmed influenza related hospitalizations in persons less than 18 years of age in 57 counties covering 11 metropolitan areas of 10 states (San Francisco CA, Denver CO, New Haven CT, Atlanta GA, Baltimore MD, Minneapolis/St. Paul MN, Albuquerque NM, Albany NY, Rochester NY, Portland OR, and Nashville TN). Finally, there is the new Vaccine Surveillance Network, a "population-based estimates of laboratory-confirmed influenza hospitalization rates for children less than 5 years old residing in 3 counties: Hamilton County OH, Davidson County TN, and Monroe County NY. Children admitted to NVSN hospitals with fever or respiratory symptoms are prospectively enrolled and respiratory samples are collected and tested by viral culture and RT-PCR. NVSN estimated rates are reported every 2 weeks."
CDC puts these seven sources together to get a picture of influenza activity. Some data are only national (pneumonia and influenza deaths) while some are reported both nationally and for designated regions. Reporting is voluntary.

Data for the preceding week is posted at the CDC Flu Activity website on Fridays. Here is the summary for this week. The site has more details on the reports for each individual component:
Synopsis: During week 42 (October 16 - October 22, 2005)*, influenza activity was low in the United States. Nine (1.1%) specimens tested by U.S. World Health Organization (WHO) and National Respiratory and Enteric Virus Surveillance System (NREVSS) collaborating laboratories were positive for influenza virus. The proportion of patient visits to sentinel providers for influenza-like illness (ILI) and the proportion of deaths attributed to pneumonia and influenza were below baseline levels. One state reported regional influenza activity; 12 states, New York City, and Puerto Rico reported sporadic influenza activity; and 37 states and the District of Columbia reported no influenza activity.