An attempt to find out
How many other dogs, cats, and domestic hoofed animals do dogs kill or injure each year in the U.S.?
How many are killed or seriously injured by pit bulls?
Often asked, these questions have eluded answer because no agency systematically tracks dog attacks on other animals, and even if any did, under-reporting would pose two major problems.
First, people tend not to report attacks by their own dogs on other animals in their households.
Second, people also tend not to report attacks by other people’s dogs on animals whose cost of replacement with a similar animal would be less than the deductible on a typical homeowner’s insurance policy.
Problems in data compilation and analysis, however, often have time-tested solutions.
The problems inherent in estimating dog attack fatalities and serious injuries inflicted on other animals are similar to those that were involved more than 20 years ago in estimating U.S. animal shelter killing, when relatively few shelters published their annual statistics and many did not keep any.
In 1994 I responded by developing and publishing an estimate based on proportionally weighting the limited data that was available to get regionally balanced representation.
The then newly formed National Council on Pet Population Study & Policy then spent more than two years doing a much more comprehensive set of surveys which in 1996 demonstrated that my estimate had been right on target.
I have produced and published annually updated shelter killing estimates every year since. The numbers have been used by almost every major U.S. animal protection organization, both national and regional, and continue to be affirmed as accurate by other entities doing relevant studies.
To develop an estimate of the numbers of animals killed or injured by dog attacks, I collected media reports of 424 dogs killing 593 other animals and injuring more than 200 during calendar year 2013. This was done with the understanding that these reports would provide only “tip-of-the-iceberg” data from which to do proportional weighting and projection.
The most essential part of the exercise was to develop a consistent, systematic, and accurate method of estimating the extent of under-reporting. This began with investigating why these cases came to light, but not others.
As it happened, every account resulting in a media report also involved one or more of three elements: at least one human was injured while intervening in the dog attack, and/or police shot the dog or dogs involved, and/or the victim animals were equines, livestock, or poultry of cash value exceeding $1,000.
Only one of the reported attacks involved dogs killing or injuring other animals belonging to their own household.
For each reported attack, there was probably a minimum of one unreported attack involving animals of the same household, since animals of the same household have more exposure to a dangerous dog than any other animals.
Also, for each reported attack, there was probably at least one unreported attack involving animals not of the same household, in which no humans were hurt, no dogs were shot, and the animals killed or injured could be replaced for less than $1,000.
This may be greatly underestimating the numbers of such attacks, but I prefer to err on the side of caution.
Accordingly, before introducing regional proportional weighting, my assumed ratio of reports to attacks was one-out-of-three.
Next I looked at the geographic representation of the reported dog attacks on other animals to ensure that proportional weighting could produce accurate data. Reported attacks occurred in communities including 14% of the human population of the U.S., including 26% of the populations of the Gulf Coast states (Deep South plus Texas), 20% of the populations of the West Coast states, and 12% to 12.7% of the populations of the Rocky Mountains, Midwest, South Atlantic, and Northeastern states.
Reported attacks occurred in communities including only 7% of the human population of the Appalachian states. Paradoxically, the Arkansas-Tennessee-Kentucky-West Virginia nexus leads the U.S. in frequency of fatal and disfiguring dog attacks on humans. Under-reporting of dog attacks on other animals may be more pronounced in Appalachia precisely because of the frequency of dog attacks on humans: the attacks on humans are so frequent as to make dog attacks on other animals seem less newsworthy.
Reported dog attacks on other animals were in all regions broadly distributed enough to suggest that the data from communities with reported attacks could be projected to the whole of each region.
Adjusting for community size
In small communities, of less than 10,000 people, almost every incident appears likely to be reported if a person is injured by a dog, police shoot a dog, or the animals injured in a dog attack have cash value of more than $1,000. Thus the number of reported attacks in communities of less than 10,000 people could simply be multiplied by three to get the probable numbers of total animal victims: only 67% might be unreported.
In communities of up to 100,000 people, however, minor injuries to humans, police shooting dogs, and injuries to animals of cash value exceeding $1,000 are much less likely to be deemed newsworthy by local media. Yet there is no reason to believe these communities have fewer dogs, or fewer other pets, though they probably have far fewer equines and livestock.
Extrapolating from the ratio of reported dog attacks on other animals to human population in communities of less than 10,000 people, the ratio of reported to unreported attacks in communities of up to 100,000 people appears to be one to nine: 90% of dog attacks on other animals do not receive media attention.
In communities of 100,000 people on up, and presuming for statistical purposes that no dog attacks on horses, livestock, or poultry occur in these communities, since such attacks would be relatively rare, 98% of dog attacks on other animals appear to receive no media attention.
Attack reporting compared to police shootings
The ratios of reported to unreported dog attacks on other animals relative to community size compare reasonably well to the ratios of media-reported to unreported dog shootings by law enforcement claimed by proponents of laws to require police to be trained in dog behavior. Of 636 shootings of dogs by law enforcement claimed on one online petition, I was able to confirm 173 from media accounts (27%). Of 210 shootings of dogs by law enforcement claimed on another online petition, I was able to confirm 33 from media reports (16%).
Several police agencies serving communities of more than a million people have within the past 10 years released reports on use of firearms which show shooting dogs to stop attacks as by far the most frequent police use of guns. Only 2.5% of these shootings have been confirmed by media in Riverside County, California; 3.4% in New York City.
43,500 animals killed by dogs in 2013
Overall, the 416 dogs who attacked other animals in 2013, the 576 animals they killed, and the 200 animals they seriously injured, according to media reports, appear to represent about 76 times more attacking dogs and more animal victims than the reported totals. The tip of the iceberg, in other words, is about 1/76 of the mass beneath the waves.
According to these calculations, rounding off the numbers to the nearest 100, about 31,400 dogs attacked about 61,500 other animals in the U.S. in 2013, killing 43,500 and seriously injuring 18,100.
The animals killed included about 12,000 dogs, 8,000 cats, 6,000 hooved animals, and 17,000 other small domestic animals, primarily poultry.
The seriously injured included about 12,400 dogs, 4,000 cats, and 1,700 hooved animals. Few small mammals and poultry survived reported dog attacks.
Pit bulls killed 99%
Pit bulls appear to have inflicted not less than 95% of the total fatal attacks on other animals (43,000). Altogether, pit bulls inflicted 96% of the fatal attacks on other dogs (11,520); 95% of the fatal attacks on livestock (5,700); 95% of the fatal attacks on small mammals and poultry (16,150); and 94% of the fatal attacks on cats (11,280).
About 30,000 pit bulls were involved in attacks on other animals. There are about 3.2 million pit bulls in the U.S. at any given time, according to the my annual surveys of dogs offered for sale or adoption via online classified ads. Thus in 2013 about one pit bull in 107 killed or seriously injured another animal, compared with about one dog in 50,000 of other breeds.
USDA Wildlife Services estimated in 2006 that domestic dogs had killed about 21,900 livestock and poultry in 2004, a number rounded off to 22,000 and recycled in several subsequent reports. USDA Wildlife Services also estimated that domestic dogs killed about 2,700 sheep in 2009. USDA Wildlife Services has not tried to quantify dog attacks on other animals, however, and has not published estimates of the numbers of dogs involved by breed.
Adjusting “save” rates proportionate to attacks
What if animal shelters’ claimed “save rates” were proportionately reduced each time they rehome a dog who goes on to kill other animals?
After all, if a dog is not killed by shelter personnel, but the dog kills another animal, there is no net gain in animal lives saved. Further, if the shelter that rehomed the dog believes rehoming dogs who kill other dogs is acceptable, the shelter is saying in effect that harmless animals being dismembered alive is preferable to dispatching a dangerous dog by lethal injection.
Because of under-reporting, it is not possible to know exactly how many other animals are killed by dogs rehomed from any particular shelter. But human fatalities from dog attacks are almost always reported, and human disfigurements from dog attacks appear to be reported much more often than not.
Based on the approximate ratios of fatal and disfiguring dog attacks on humans to the estimated numbers of other animals killed by dogs, shelters’ “save rates” should be reduced by 2,000 animals every time they rehome a dog who goes on to kill a person, and by 400 every time they rehome a dog who goes on to disfigure a person.
One pit bull fan liked the idea
Responded Ed Boks, executive director of Humane Society of Yavapai County, Arizona, when I posted this idea to Facebook: “I would agree with that formula Merritt! There is no excuse for allowing, much less encouraging, risky adoptions just to maintain numbers.” An avowed pit bull advocate, Boks while heading the animal control agencies serving Phoenix, New York City, and Los Angeles presided over more pit bull intakes, killing, and adoptions than any other shelter director in the U.S., with no known instances of a pit bull that his agencies rehomed going on to kill or disfigure a human.
Nationally, fatal and disfiguring attacks by dogs from shelters and rescues have exploded from zero in the first 90 years of the 20th century to 80 since 2010, including 58 by pit bulls, along with 22 fatal & disfiguring attacks by other shelter dogs, mostly Rottweilers & bull mastiffs.
Altogether, 33 U.S. shelter dogs have participated in killing people since 2010, including 24 pit bulls, seven bull mastiffs, and two Rottweilers.
The only dogs rehomed from U.S. shelters to kill anyone before 2000 were two wolf hybrids, rehomed in 1988 and 1989, respectively.
(See also The science of how behavior is inherited in dangerous dogs, by Alexandra Semyonova,