Background Driving accidents cause 1.2 million deaths each year worldwide, with alcohol a causative factor in at least one fifth. Breath alcohol is a well-established surrogate measure for blood alcohol and the portable breathalyser technology available to police forces has evidential force in some jurisdictions. However, in recent years cheap breathalysers have been marketed directly to the UK public, potentially allowing users to assess their own fitness to drive. We aimed to determine the accuracy of three selected breathalysers marketed to the UK public.

Methods Diagnostic accuracy study of three personal breathalysers (two single-use and one digital multi-use) available to UK consumers, using as reference test an evidential standard breathalyser used by police. We recruited 208 participants aged 18 or over who had consumed alcohol in participating licensed bars in Oxford, United Kingdom. Participants answered a short questionnaire including self-reported recollection of alcohol consumption during the preceding 12 h, and used the breathalysers in a random order. We calculated sensitivity and specificity of each index device for detection of alcohol levels at or over the UK/US driving limit (35 µg/100ml breath alcohol concentration), using the police breathalyser as reference standard. We also calculated diagnostic accuracy of self-reported alcohol consumption.

Results 38/208 (18.3%) of participants were at or over the driving limit according to the police breathalyser. The digital multi-use breathalyser had a sensitivity of 89.5% (95% CI 75.9–95.8%) and a specificity of 64.1% (95% CI 56.6–71.0%). The single-use breathalysers had a sensitivity of 94.7% (95% CI 75.4–99.1%) and 26.3% (95% CI 11.8–48.8%), and a specificity of 50.6% (95% CI 40.4–60.7%) and 97.5% (95% CI 91.4–99.3%) respectively. Self-reported alcohol consumption threshold of 5 UK units or fewer had a higher sensitivity than all personal breathalysers.

Conclusion While two of the breathalysers tested had good sensitivity, one had very poor sensitivity. None were superior to estimation of self-reported alcohol consumption. Limitations of our study include the surrogate reference standard. The value, and safety, of marketing insensitive breathalysers to the public is open to question.