Juthani‐Mehta 2015.

Study characteristics
Methods	Cluster‐RCT randomised at nursing home level, USA Study recruitment and setting details: see Table 4
Participants	Inclusion criteria: long‐term care residents aged > 65 years, resided at the nursing home for ≥ 1 month, at least 1 of 2 modifiable risk factors for pneumonia (impaired oral hygiene, swallowing difficulty) Exclusion criteria: housing for short‐term rehabilitation, presence of a gastric or jejunostomy tube, presence of a tracheostomy, life expectancy < 3 months, current use of chlorhexidene, pneumonia within the previous 6 weeks, previous enrolment in the study, unwillingness to give informed consent, non‐English speaking, inappropriateness for the study in opinion of nursing home administration Multi‐component OHC group: 18 homes allocated (434 participants; of whom 100 had a stroke) Usual care group: 18 homes allocated (400 participants; of whom 92 had a stroke) Details of participants are shown in Table 5
Interventions	Multi‐component OHC intervention Intervention: multi‐component OHC intervention Materials: oral chlorhexidene Agent: 0.12% chlorhexidene oral rinse Training: study personnel trained nursing home staff about intervention procedures Procedures: manual tooth and gum brushing plus chlorhexidene oral rinse, plus upright positioning during feeding. Tailored to participants who could either perform self‐care or required assistance Provided by: nursing aides Training provider: not reported Delivery: face‐to‐face; 1:1; nursing home Regimen: twice per day Duration: unclear but study participants were followed up for 2.5 years Tailoring: intervention protocol was tailored to participants who could either perform self‐care or required assistance (detailed in Supplementary Appendix of paper) Modification: none Adherence: adherence to chlorhexidene and toothpaste usage were measured by comparing expected vs actual volumes (Table 7). No dropouts reported Usual care Intervention: standard care Materials: not reported Procedures: not reported Provided by: not reported Delivery: not reported Regimen: not reported Tailoring: not reported Modification: none Adherence: no dropouts reported
Outcomes	Primary outcomes: radiographically documented pneumonia (Table 8) Secondary outcomes: development of a first lower respiratory tract infection; adherence to chlorhexidine (compared expected vs actual chlorhexidine volume expenditure), oral brushing adherence (compared expected vs actual residual toothpaste tube weight), upright feeding positioning adherence was evaluated qualitatively once per month Data collection: baseline and participants were followed for up to 2.5 years for development of primary outcome
Funding	Study authors declared no conflicts of interest. Study funded by the National Institutes of Health, the National Institute on Aging (NIA) (K23AG028691, R01AG030093 and P30AG021342)
Notes	Dropouts are detailed in Table 9 Statistical data not included within the review meta‐analyses Note: the number of stroke participants were reported but we were unable to access the outcome data specific to participants who had a stroke and so they were not included in the meta‐analyses
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Comment: nursing homes were stratified into 2 groups by number of minutes that nursing aides spent with residents per day, > 140 aide minutes were high stratum, < 140 minutes per day were low stratum. Homes were randomised within stratum using a permuted block design.
Allocation concealment (selection bias)	Unclear risk	Comment: insufficient information.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Comment: insufficient information.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Comment: 2 investigators adjudicated all the outcomes, blinded to the randomisation status of the participants and the cumulative incidence during the trial.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Comment: dropouts accounted for. Missing outcome data balanced, with no significance differences for either outcome between the 2 groups. ITT employed.
Selective reporting (reporting bias)	Low risk	Comment: all prespecified outcomes were reported.
Baseline data comparable?	Low risk	Comment: no difference in age, sex, race or ethnicity, comorbidities, mental status, functional status except for 1 measure of behaviour (resists care).
A priori power calculation	Low risk	Comment: yes; target of 828 participants to detect a 25% reduction in pneumonia rate.
Other bias	Unclear risk	Comment; the study was terminated for futility as the conditional power under observed treatment difference was nearly 0.