Apply RoB 2 to the result for a specific crossover comparison, answer each signalling question with period and carryover effects in view, and record in the justification how the analysis handled them. RoB 2 assesses a specific randomized-trial result across bias domains using signalling questions and written justifications, producing low risk, some concerns, or high risk judgements (Cochrane Handbook, Chapter 8).
Scope the assessment to one result
Name the outcome, the comparison, and the time point before answering anything. A winter-health trial may report several outcomes, and each result gets its own judgement. For a crossover, the relevant result is the within-participant comparison of the treatments, so state whether the analysis used paired or period-adjusted data.
Work through the signalling questions with period and carryover in view
- For the randomization domain, check whether the sequence was generated and concealed in a way that supports the crossover order, and note whether the order of treatments was randomized. Record the answer and the evidence you used.
- For the deviations domain, check whether participants or staff could tell which treatment was active in each period, and whether deviations differed by period. Record how the analysis handled them.
- For the missing-data domain, check whether dropouts occurred after the first period and whether the analysis accounted for participants who completed only one period. Record the result.
- For the measurement domain, check whether outcome assessment could be influenced by knowing the treatment or the period. Record the result.
- For the selection-of-reported-result domain, check whether the reported analysis matches the pre-specified plan for the crossover comparison. Record the result.
Record period and carryover effects explicitly
Period effects arise when outcomes differ between the first and second treatment periods, for example because winter illness risk changes across the season. Carryover effects arise when a treatment's effect persists into the following period. For each, write down whether the analysis tested for the effect, adjusted for it, or left it unaddressed, and let that record inform the domain judgement. A paired analysis that ignores a period effect, or an analysis that pools periods without addressing carryover, changes what the result can support.
Reach and justify the judgement
Assign low risk, some concerns, or high risk for the result, then write a justification that names the domains driving the judgement and the period or carryover handling behind it. Keep the justification specific to the crossover design so a reader can see which feature of the analysis produced the rating.