Table 1.
Main characteristics of the selected studies.
| - Author - (Year of Publication) - Study Area |
- Study Design - Study Period - Length of Follow-Up - Data Sources |
- Study Populations - Intervention and Comparison |
- Outcome - Measure of Effect |
Main Result (p Value or 95% Confidence Interval) |
Quality Assessment Sources of Bias |
|---|---|---|---|---|---|
| Schroer, B. J. & Peiton, W. P. (1978) Huntsville (Alabama, USA) |
Retrospective cohort study April 1975 to December 1976 21 months Linkage of pre-existing databases: - Vehicles registered - Vehicle inspected - Road crashes |
8494 inspected vehicles 30,089 non-inspected vehicles |
Road crash rates (per vehicles) Percent change |
9.1% reduction in inspected vehicles (one-tailed test: p < 0.05) |
NOS 1: 7 Selection: 4 Comparability: 1 Outcome: 2 Secondary sources of information Voluntary assignment of exposure High risk of confounding |
| White, W. T. (1985) Levin township (New Zealand) |
Retrospective cohort study 1980–1984 26 weeks Linkage of pre-existing databases: - Vehicles inspected - Road crashes |
9714 cars inspected; 5898 cars inspected at least twice |
Road crash rates (per inspected cars) Coefficient of the variable “weeks after the last inspection” in a Poisson regression model |
0.028 (one-tailed test: p = 0.0002) |
NOS 1: 6 Selection: 3 Comparability: 1 Outcome: 2 Secondary sources of information High risk of confounding |
| Fosser, S. (1992) Norway |
Randomized controlled trial 1986–1990 Three 1-year periods Primary data about inspections Data accidents (insurance companies) |
204,000 vehicles, including vans and passenger cars Three comparison groups: (1) 4600 cars inspected annually between 1986 and 1988 (2) 4600 vehicles inspected once in 1986 (3) 112,000 non- inspected cars (Control group) |
Road crash rates (per car-days) Not estimated |
No statistically significant differences between rates in the three groups | Jadad scale 2: 2 Dropouts from scheduled interventions No blinding |
| Blows, S. et al. (2003) Region of Auckland (New Zealand) |
Prospective case–control study March 1998 to July 1999 Face-to-face or telephone interviews |
Cases: 571 hospitalized drivers of passenger cars, vans, and light industrial vehicles Controls: 588 drivers of non-accident vehicles, obtained by random cluster sampling |
Frequency of being in possession of a Warrant of Fitness or vehicle inspection certificate Odds ratio |
2.67 (1.46, 4.86) |
NOS 1: 6 Selection: 4 Comparability: 2 Outcome: 0 Type of design (case–control) Information (recall) bias Residual confounding |
| Christensen, P. &Elvik, R. (2007) Norway |
Retrospective pre–post cohort study 1998–2005 Up to 5 years Linkage of pre-existing databases: -Inspections (public road administration) - Crashes (insurance company) |
253,098 passenger cars observed before and after one, two, or three inspections | Road crash rates (per car) Percent change |
Vehicles with one inspection: +2.6% (−0.7%, 6.0%) Vehicles with two inspections: +8.4% (3.9%, 13.2%) Vehicles with three inspections: +4.0% (−23.6%, 41.5%) |
NOS 1: 5 Selection: 3 Comparability: 0 Outcome: 2 Type of design (pre–post) Secondary sources of information High risk of confounding |
| Keall, M.D. & Newstead, S. (2013) New Zealand |
Retrospective cohort study 2003–2009 Up to 6 years Linkage of pre-existing databases: - Crash data - Licensing data - Inspection data |
2,710,797 vehicle-years Compare vehicles subject to annual inspections (aged 6 years or less) versus vehicles subject to 6-month inspections (aged 7 years or more) |
Road crash rates (per vehicle-year) Percent change |
8% reduction in vehicles inspected every 6 months (0.4%, 15%) |
NOS 1: 6 Selection: 3 Comparability: 1 Outcome: 2 Secondary sources of information Risk of confounding |