A case-control study to assess Utah's shaken baby prevention program

Background: Evaluation of unproven injury interventions with uneven penetrance where it is considered unethical to withhold the intervention is challenging. We performed a case-control study of an educational intervention to prevent shaken baby syndrome (SBS). Simultaneously, we performed a bias detection study using sham cases. If the intervention were effective, one would expect a lower odds of injury in cases who received the intervention and no effect in sham cases.

Methods: Cases were mothers' of Utah resident children < 2 years old born between January 2001 and July 2007 who were admitted to hospital for SBS. Sham cases differed on injury: injured in a motor vehicle crash. Controls matched on infant birth date were selected randomly from the Utah birth certificate database for cases and sham cases. Cases, sham cases, and controls were interviewed to ascertain receipt of teaching in their birth hospital about shaken baby syndrome. Birth records about teaching delivered were obtained for cases, sham cases, and 10% of controls.

Results: 56 cases and 25 sham cases were identified. 39% of cases, 40% of sham cases, and 41% of controls received video teaching about SBS. The odds of injury given teaching were 0.9 (95% CI: 0.5, 1.8) for cases and 0.9 (95% CI: 0.4, 2.2) for sham cases. Intervention effectiveness was similar for cases and sham cases.

Conclusions: A case-control study with a simultaneous bias control study is a stronger method to measure the effectiveness of an educational intervention with uneven penetrance than a simple case-control study.