Sharma 2017.

Study characteristics
Methods	RCT. Parallel design with 2 arms. Duration: 3 months. Location: single centre in Australia.
Participants	Inclusion criteria: 60 years and over presenting to the General Medicine Department of the Flinders Medical Centre between November 2014 and June 2016, malnourished according to PG‐SGA (classes B or C). Exclusion criteria: receiving palliative care, resident in rural areas, indigenous Australians, non‐English speaking and unable to give informed consent. Number randomised: 148 participants (intervention group, n = 78; control group, n = 70). Attrition: 103 analysed (intervention group, n = 57; control group, n = 46), 45 lost to follow‐up (intervention group, n = 21; control group, n = 24). Gender split: intervention group 29.7% males, control group 32.9% males. Age: mean (range), intervention group 82.0 (80.0 ‐ 83.9), control group 81.6 (79.5 ‐ 83.6). Nutritional status: n (%) PG‐SGA score, intervention PG‐SGA B 67 (90.5), PG‐SGA C 7 (9.5); control PG‐SGA B 60 (87), PG‐SGA C 9 (13).
Interventions	Intervention: participants received dietary advice plus ONS if required in the form of individualized nutrition care plan plus monthly post‐discharge telehealth follow‐up. Control: participants received no dietary advice and no ONS in the form of intervention only upon referral by their treating clinicians.
Outcomes	Length of stay, skinfolds*, MUAC*, grip strength* complications, QoL (EQ‐5D)*, mortality*, readmission rate*, nutritional status by PG‐SGA score.
Publication details	Language: English. Funding: not declared. Publication status: peer‐reviewed journal.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: An independent biostatistician prepared the randomisation schedule using random blocks of 8; treatment allocations were randomly permuted and balanced within blocks.
Allocation concealment (selection bias)	Low risk	Quote: After obtaining written informed consent, the researcher contacted central office to open these sealed envelopes to allocate participants to either control or intervention groups.
Blinding (performance bias and detection bias) Clinical outcomes	Low risk	Quote: After group allocation the participants and the ward dietitian, who provided nutrition intervention, were not blinded to group allocation but the research dietitian who conducted the final outcome assessment was blinded to participants' group allocation. In addition, the research person overseeing data entry and the biostatistician were blinded.
Blinding (performance bias and detection bias) Functional outcomes	Low risk	Quote: After group allocation the participants and the ward dietitian, who provided nutrition intervention, were not blinded to group allocation but the research dietitian who conducted the final outcome assessment was blinded to participants' group allocation. In addition, the research person overseeing data entry and the biostatistician were blinded.
Blinding (performance bias and detection bias) Nutritional outcomes	Low risk	Quote: After group allocation the participants and the ward dietitian, who provided nutrition intervention, were not blinded to group allocation but the research dietitian who conducted the final outcome assessment was blinded to participants' group allocation. In addition, the research person overseeing data entry and the biostatistician were blinded.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Not blinded and likely that researchers and participants were aware of group allocation as this was a nutritional intervention. It is possible that assessment of some outcomes was influenced by lack of blinding.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Outcomes were assessed by a research dietitian who was blinded to group allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	148 participants randomised (intervention group n = 78; control group n = 70), complete data were available for analysis for 103 participants (intervention group n = 57; control group n = 46 participants). The main reasons for participants being lost to follow‐up were loss of contact n = 7 (intervention group n = 1; control group n = 6), consent withdrawal n = 12 (intervention group n = 8; control group n = 4) and death n = 26 (intervention group n = 12; control group n = 14). Numbers and reasons similar between groups
Selective reporting (reporting bias)	High risk	No study protocol identified, thus unable to judge whether all planned outcomes were reported.
Other bias	Low risk	Baseline characteristics similar between groups, with both groups having a higher number of females, the majority of participants residing at home pre‐admission, a similar number of co‐morbidities, and similar Charlson Co‐morbidity Index and other baseline characteristics. There was no difference in the severity of malnutrition at baseline.