The Centers for Disease Control (CDC) Report on Fatal Dog Attacks...

Overwhelmingly this report is used by the media, council members and legislators in an attempt to prove a case for passing breed specific legislation.  So I feel in necessary to set the record straight on this report for all to see.
 

Here are some quotes from the CDC and Doctors involved in the studies explaining how the report is INACCURATE:

• Procedure: We collected data from The Humane Society of the United States (HSUS) and media accounts related to dog bite attacks and fatalities, using methods from previous studies (CDC Special Report on breeds involved in fatal human attacks in the United States between 1979 and 1998, September 2000).

• Ideally, breed-specific bite rates would be calculated to compare breeds and quantify the relative dangerousness of each breed. For example, 10 fatal attacks by Breed X relative to a population of 10,000 X’s (1/1,000) implies a greater risk than 100 attacks by Breed Y relative
  to a population of 1,000,000 Y’s (0.1/1,000). Without consideration of the population sizes, Breed Y would be perceived to be the more dangerous breed on the basis of the number of fatalities.   (CDC Special Report on breeds involved in fatal human attacks in the United States between 1979 and 1998, September 2000).   NOTE:  The CDC study does NOT use population as a factor.

• Considering only bites that resulted in fatalities, because they are more easily ascertained than nonfatal bites, the numerator of a dog breed-specific human DBRF rate requires a complete accounting of human DBRF as well as an accurate determination of the breeds involved. Numerator data may be biased for 4 reasons. First, the human DBRF reported here are likely underestimated; prior work suggests the approach we used identifies only 74% of actual cases.1,2 Second, to the extent that attacks by 1 breed are more newsworthy
  than those by other breeds, our methods may have resulted in differential ascertainment of fatalities by breed. Third, because identification of a dog’s breed may be subjective (even experts may disagree on the breed of a particular dog), DBRF may be differentially ascribed to breeds with a reputation for aggression.  Fourth, it is not clear how to count attacks by crossbred dogs. Ignoring these data underestimates breed involvement (29% of attacking dogs were crossbred dogs), whereas including them permits a single dog to be counted more than once.   (CDC Special Report on breeds involved in fatal human attacks in the United States between 1979 and 1998, September 2000)

• Finally, it is imperative to keep in mind that even if breed-specific bite rates could be accurately calculated, they do not factor in owner related issues. For example, less responsible owners or owners who want to foster aggression in their dogs may be drawn differentially to certain breeds. (CDC Special Report on breeds involved in fatal human attacks in the United States between 1979 and 1998, September 2000)

• (after 1998, the CDC stopped tracking which breeds of dogs are involved in fatal attacks; according to a CDC spokesperson, that information is no longer considered to be of discernable value)  (Pit Bulls in the City, Indy Tails July 2005)

• "There are enormous difficulties in collecting dog bite data," Dr. Gilchrist said.  She explained that no centralized reporting system for dog bites exists, and incidents are typically relayed to a number of entities, such as the police, veterinarians, animal control, and emergency rooms, making meaningful analysis nearly impossible.  (CDC releases epidemiologic survey of dog bites in 2001, September 2003) 

• When multiple dogs of the same breed were involved in the same fatal episode, that breed was counted only once (eg, if 10 Akitas attacked and killed a person, that breed was counted once rather than 10 times). When crossbred dogs were involved in a fatality, each suspected breed in the dog’s lineage was counted once for that episode. Second, we tallied data by dog. When multiple dogs of the same breed were involved in a single incident, each
  dog was counted individually. We allocated crossbred dogs into separate breeds and counted them similarly (eg, if 3 Great Dane-Rottweiler crossbreeds attacked a person, Great Dane was counted 3 times under crossbred, and Rottweiler was counted 3 times under crossbred). Data are presented separately for dogs identified as pure- and crossbred.   (CDC Special Report on breeds involved in fatal human attacks in the United States between 1979 and 1998, September 2000)