A new study has found many drivers who tested positive for drug-use didn’t report taking them – underlining the importance of blood or saliva tests.

The research from the Insurance Institute for Highway Safety (IIHS) was based on analysis of the two most recent national roadside surveys.

“Although we expected drug use to be underreported, it was surprising how inconsistent the results were from one survey year to another,” said IIHS Senior Research Scientist Angela Eichelberger.

“These findings suggest that self-reported drug use is not a good measure for monitoring trends in drug use in this population.”

The study found information on drug use from the Fatality Analysis Reporting System – a census of fatal crashes on US roads – is problematic. Many drivers involved in fatal crashes are never tested for drugs, and when they are tested, inconsistencies in testing procedures among different states and different time periods make it difficult to accurately estimate trends, said the report.

The study suggested the national roadside survey of alcohol and drug use by drivers as a better source of data. To gather results, researchers work with local police to safely stop drivers, who are asked to participate anonymously. Drivers who are found to be impaired do not face charges. However, they are provided with alternative transportation.