The initial clue that an outbreak is occurring can come in various ways.
It may be when a person realizes that several other people who were all together at an event have become ill and he or she calls the local health department.
It may be when a physician realizes she has seen more than the usual number of patients with the same illness.
It may be when a county health department gets an unusually large number of reports of illness.
The hardest outbreaks to detect are those that are spread over a large geographic area, with only a few cases in each state. These outbreaks can be detected by combining surveillance reports at the regional or national level and looking for increases in infections of a specific type. This is why state public health laboratories determine the serotype of Salmonella bacteria isolated from people.
New "DNA fingerprinting" technologies can make detecting outbreaks easier too. For example, the new molecular subtyping network, PulseNet, allows state laboratories and CDC to compare strains of E coli O157: H7 and an increasing number of other pathogens from all across the United States to detect widespread outbreaks.
After an apparent cluster of cases is detected, it is important to determine whether these cases represent a real increase above the expected number of cases and whether they really might be related.
Sometimes a cluster of reported cases is caused by something other than an actual outbreak of illness. For example, if the person responsible for reporting has just returned from a vacation and is clearing up a backlog of cases by reporting them all at once, the sudden surge of reports is just a false cluster.