Table 3.

Categorisation of the suitability of different study designs (coupled with different analyses) to draw conclusions regarding a causal association between PAct and outcomes of diabetes and related metabolic disorders. PAct = Patient Activation

Possible study designs+analyses	Suitability of study design and analyses	Rationale
RCTs with causal mediation analysis to assess whether PAct mediates intervention effects	Strong	RCTs are the only study design that allow causal mediation analysis to identify the mechanisms by which interventions exert their effects51
Cohort studies/RCTs or other intervention studies that assess the association between PAct and subsequent outcomes	Moderate	RCTs and longitudinal observational studies can provide temporal insights into the association between PAct and outcomes, which gives some indication of causality.52 If an RCT examines the association between PAct and outcomes independent of study group allocation, randomisation has no bearing; analyses and findings are therefore akin to cohort studies.
RCTs that do not report on the association between PAct and outcomes but that show intervention effects on outcomes AND intervention effects on PAct, AND the intervention explicitly, mainly addresses PAct	Moderate	RCTs provide insight into causal effects of interventions on outcomes. If an intervention explicitly addresses PAct and there is evidence that the intervention influenced both PAct and outcomes, this provides indication for a causal mechanism of PAct on outcomes (though not definitive).
Observational cross-sectional studies	Weak	In cross-sectional designs, the time order of effects cannot be determined and therefore causality cannot be inferred.53
Intervention studies that are not RCTs (eg, pre-post studies) and that do not report on the association between PAct and outcomes but that show changes in outcomes AND changes in PAct.	Weak	Pre-post designs have the strength of temporality to indicate outcomes might be impacted by an intervention, but due to lack of randomisation causality cannot be inferred.54

PAct, patient activation; RCT, randomised controlled trial.