Happ 2015.

Study characteristics
Methods	Randomised cluster step‐wedged trial. Each if the 6 ICUs were randomised to a 3‐month intervention across 18 months
Participants	1440 participants (814 intervention phase, 626 control phase) Inclusion criteria: aged ≥ 18 years; first ICU admission during the hospital stay; mechanically ventilated for ≥ 2 days via an endotracheal or tracheostomy tube; awake, alert, and responsive to verbal communication from the clinicians. Exclusion criteria: people requiring brief intubation (< 2 days) in which they were extubated shortly after awakening.
Interventions	Intervention SPEACS‐2 consisting of: 6 × 10‐minute online educational modules involving narrated text slides and video exemplars of communication assessment and techniques (60 minutes); reference manual, pocket reference cards, assessment‐intervention algorithm; communication cart in the ICU containing assistive communication tools and materials; communication resource nurses (champions) – minimum of 2 per ICU; weekly teaching posters "communication strategy of the week"; weekly patient case conference with speech language pathologist. Control Communication interventions during the control period comprised usual care of that unit.
Outcomes	Intervention fidelity – training completion; nurse knowledge acquisition; communication supply usage; attendance at SLP bedside teaching; and adherence to training principles(intervention enactment). Nurse satisfaction and comfort with communication. Physical restraint. Heavy sedation. Coma‐free days. Pain documentation ICU‐acquired pressure ulcer, ≥ grade II. Unplanned endotracheal or tracheal tube extubation. Ventilator‐free days. Length of stay (ICU and hospital). Cost‐adjusted charges.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation of the units to intervention period was conducted by the statistician using computer‐generated random ordering. Random selection of electronic charts within quarters and within units used a computer‐generated random number table for chart selection by unit within each quarter until 30 participants meeting criteria were reached.
Allocation concealment (selection bias)	Unclear risk	Allocation concealment of randomisation of ICUs to stepped wedge is challenging due to the need to notify units to prepare for practice change.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Unable to blind participants and personnel given the nature of the intervention.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Trained staff, blinded to intervention assignment abstracted clinical data from EMR.
Incomplete outcome data (attrition bias) All outcomes	Low risk	No evidence of incomplete data.
Selective reporting (reporting bias)	Low risk	No evidence of selective reporting.
Other bias	Unclear risk	Funding: not considered source of bias: Robert Wood Johnson Foundation Interdisciplinary Nursing Quality Research Initiative grant #66633. Author declared potential conflict of interest: the SPEACS‐2 programme is accessible online at go.osu.edu/speacs2. Dr Happ holds the Creative Commons copyright.