Table 1

Alternative design features

Optimal design featureAlternative options ranked by internal validity of evidence
Options 1Options 2
Pre-intervention random allocation of participants or clusters to intervention or control group
  • Pre-intervention matching of control group to intervention group on key confounders, or statistical adjustment for baseline differences in these factors

  • Restrict participation by key confounders

  • Allocation by quantitative measure of need

Concurrent intervention and control group
  • Comparison of changes over time in the intervention group with contemporaneous changes in the general population

  • Before-and-after study with examination of other explanatory factors to increase the plausibility that any changes that are observed over time may be due to the intervention

Prospective follow-up from baseline to outcomes
  • Proxy outcome measures

  • Retrospective measurement of exposure