Examples of study designs commonly used in the evaluation of injury interventions: quasiexperimental designs
| Adapted from Shortell and Richardson, 1978.23 | |
| (1) Single time series design | |
| Group 1 | Multiple observations→intervene→multiple observations |
| Examples: monitored national trend in poisonings before and after child resistant packaging required for oral prescription drugs46; monitored trend in homicides in Washington, DC, before and after gun control law enacted47 | |
| (2) Multiple time series design | |
| Group 1 | Multiple observations→intervene→multiple observations |
| Group 2 | Multiple observations→no action→multiple observations |
| Examples: monitored trends in fire related injury rates in intervention and comparison areas before and after smoke detector giveaway program in Oklahoma City16; monitored trends in alcohol related fatal crashes in intervention and comparison areas before and after community intervention in Massachusetts29 | |
| (3) Non-equivalent control group design | |
| Group 1 | Observe→intervene→observe |
| Group 2 | Observe→no action→observe |
| Examples: assessed bicycle helmet use in intervention and comparison areas before and after helmet law went into effect in Maryland7; assessed child restraint use in Tennessee and Kentucky in intervention and comparison areas before and after child restraint law went into effect in Tennessee48; assessed motor vehicle crashes involving teenaged drivers in intervention and comparison areas in Connecticut before and after funding for driver education eliminated49 | |
| (4) Sequential cohort designs | |
| Group 1 | Intervene→observe |
| Group 2 | No action→observe |
| Example: assessed child poisonings after mothers of one birth cohort in New Zealand received poison prevention instructions and Mr Yuk labels while mothers of next birth cohort received neither50 | |
| Group 1 | Observe→intervene→observe→no action→observe |
| Group 2 | Observe→no action→observe→intervene→observe |
| Example: assessed before and after knowledge and attitudes of sequential cohorts of children attending a safety village in Maryland19 | |
| (5) Case-control design | |
| Group 1 | Select persons with injury of interest |
| Group 2 | Select matched control group without that injury |
| Then observe retrospectively presence or absence of intervention in the two groups | |
| Example: presence or absence of bicycle helmet use recorded in persons treated in Seattle emergency departments for head or non-head bicycle related injuries51 | |