November 19, 2017

Efficacy of drug detection by fully-trained police dogs varies by breed, training level, type of drug and search environment

The assessment of how efficient drug detection dogs were studied when detecting various drugs in various testing sites.

The most common uses of detection dogs by law enforcement are for narcotics and explosives. There are at least 30 sets of detection tasks that trained dogs perform to find drugs. Performance of detection dogs can be assessed by both detection speed and detection accuracy. Detection speed is important because too slow of a reaction can reduce the speed of the process, yet too fast of a reaction could result in consequences such as setting off a tripwire. Detection accuracy involves measuring correct hits, false alerts, correct rejections, and misses. The two main parameters that can be calculated in terms of accuracy are sensitivity (proportion of hits to misses), and specificity (proportion of correct rejections to false alerts). Drug detection dogs are generally trained to detect several of the most commonly used and trafficked drugs. The training allows the dogs to alert to odors considered “strong” in terms of the human capability of smell (usually of Cannabis origin).

In this study, the aim was to assess, in actual training and testing environments used by the Polish police, how effective dogs trained by the police were at illicit substance detection. This was dependent on the type of drug, dog breed, dog experience with the searching site, and drug odor residuals. The study compared the performances of 68 Labrador retrievers, 61 German Shepherds, 25 Terriers, and 10 English Cocker spaniels of both sexes. The dogs were tested shortly before the first certification of their detection proficiency, as they were all considered to be fully trained drug detection dogs. A total of 1219 tests were conducted, 440 with German Shepherds, 517 with Labrador retrievers, 203 with Terriers, and 59 with English Cocker spaniels.

The drugs used in the tests were not pharmaceutical grade, but of street material. 10-15 grams of hashish, marijuana, amphetamine, cocaine, and heroin were hidden approximately 1 hour before the tests. Each drug was placed in an unsealed plastic bag in the following locations: a known room to the dog, an unknown room to the dog, inside a car, and outside a car. A dog indicated a site where the drug odor was found either by scratching at the site (called an “active alert”) or laying at the site (called a “passive alert”). When the dog alerted to an odor and was correct, they were rewarded. When the dog alerted to an odor and was incorrect, the dog was simply told “no.” The time from start to correct indication, number of false alerts, and number of passes of the dog closer than 1 m from the sample were recorded. If 10 min had elapsed without a result, the test was considered a miss. In 7% of the trials, a miss was recorded.

On average, the hidden drug samples were found after 64 seconds of the start. The shortest mean detection time was for marijuana, while the longest mean detection time was for heroin. Detection time did not differ significantly during the search in known vs. unknown rooms, or inside vs. outside cars. Searching outside cars often ended in a miss, and searching inside cars was far less accurate than in the known or unknown rooms. 87.7% of indications were correct, while 5.3% were false indications. Dogs were equally efficient at searching in well known vs. unknown rooms with strange (non-target but novelty) odors, with 83.2% correct indications. Dogs were less accurate when searching inside and outside of cars.

The odor of hashish was indicated by dogs in 100% of the trials conducted 24 hours after it was removed from the room, and in 80% of the trials conducted 48 hours after it was removed. Heroin odors were undetected by any of the dogs after 48 hours. The drugs were therefore ranked from easiest to most difficult to find: marijuana, hashish, amphetamine, cocaine, and heroin.

German Shepherds were superior to the other breeds in giving correct indications, while Terriers showed poor detection performance (yet are used because they can fit in smaller spaces). A previous study compared using an “electronic nose”, which can detect and process vapors, and canines for drug detection. The performances of the machine and the dogs were approximately the same, yet the dogs had faster detection times. Ultimately, it was concluded that dogs were still the most effective method of drug detection, and that further research is needed to consider other factors.


Reference: Jezierski, T, Adamkiewicz, E, Walczak, M, 2014. Efficacy of drug detection by fully-trained police dogs varies by breed, training level, type of drug and search environment. Forensic Science International. Volume 237: 112-118.

Reviewed by Sadie Scott.