Case-control (government databases): 5 studies involving 11,479 cases; 10,618 cases were obtained from 2 studies. The number of controls was unclear.
Case-control (family practice): 1 study of 205 case-control pairs, with 18 accidents in the BZD group and 10 accidents in the controls.
Emergency ward: 7 studies involving 6,342 cases.
Police: 20 case series involving approximately 113,737 participants.
Driving test: 9 studies involving 352 participants.
A meta-analysis was not performed for the following reasons: the various types of studies did not yield outcomes that could be pooled; case-control studies did not study the same BZD; and the police and emergency studies were often incomplete.
Government databases: there were limitations in the reporting of data. Methodological limitations included: only incorporating severe accidents, limiting data collection to certain age groups, poor selection of controls, and high non-response rates from eligible drivers. The adjusted OR for traffic accidents involving patients who had filled prescriptions for BZD ranged from 0.9 to 2.4 for 4 studies (Quebec, Seattle, Massachusetts, Tennessee) and from 5.6 to 6.5 for the remaining county (Saskatchewan). There was no explanation for the higher OR.
Emergency room and police data: problems with the studies were encountered. These included the following: few participants received toxicological work-up; few studies used controls or assessed medical conditions; there was no assessment of the distance driven under the influence of BZD, medications, alcohol and drugs, which did not result in accidents; only 2 studies examined drivers with low blood alcohol concentrations to assess which drugs were involved in impairment or accidents; and only 3 studies cited either a lower limit for detecting BZD or actual concentrations.
Police data: the rates in which BZDs and other medications were found in travellers involved in accidents varied substantially from country to country; in the 1970s, these ranged from 65% in Denmark to 1% in California.
Emergency room data: only one emergency study both assessed health conditions and used controls. This study (201 cases, 325 controls) found diazepam in 5% of BZD patients and 2.5% of controls.
Driving experiments: drivers receiving BZD were less able to circle roundabouts, adjust for passengers obstructing their view, avoid cutting corners, anticipate problems, hold lane position, brake, go round curves, steer, park and accomplish emergency manoeuvres. Sixteen of the 100 patients receiving BZDs withdrew from tests in the Netherlands, compared with 1 in 100 of those receiving placebo.