Sjöling 2003.

Methods	Randomised parallel group trial.
Participants	60 participants (36 women, 24 men; mean age 71 years, range 54 to 86) undergoing total knee replacement. Inclusion criteria: diagnosis of osteoarthritis, aged ≥ 50 years. Exclusion criteria: rheumatoid arthritis, postoperative state of confusion. Location: Sweden.
Interventions	Intervention (n = 30) and control (n = 30) groups had a 20‐ to 40‐minute private information session with one of the investigators within 4 days of surgery. Information of a procedural nature was given verbally and in a leaflet. Participants were informed of postoperative pain measurements and showed how to use the VAS. Intervention group received additional specific information (verbally and in a leaflet) that emphasised the participant's role in pain management and explained the benefits of well‐treated postoperative pain for improving recovery and reducing complications.
Outcomes	Pain (VAS with scores ranging from 0 to 10 where a higher score represented worse pain, postoperative analgesics); satisfaction with pain management and nursing care; state and trait anxiety (5‐point Likert scale where a higher score represented greater anxiety); length of hospital stay. Participants were assessed preoperatively and 1, 2 and 3 days postoperatively.
Notes	Trialists reported that "The mean length of hospitalisation was 11 days and it was the same for both groups". Attempts to retrieve missing SDs were unsuccessful.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "... the first patient was randomised to the treatment group and each subsequent following patient was altered (sic) allocated to either group." Comment: alternation is not a random method of generating a sequence.
Allocation concealment (selection bias)	High risk	Comment: unlikely that allocation was concealed using alternation.
Blinding (performance bias and detection bias) Self‐reported outcomes	High risk	Comment: no description of any attempt to blind participants and blinding is unlikely to have been possible. Knowledge of being in the intervention group may have influenced participants' pain measurements and satisfaction with care.
Blinding (performance bias and detection bias) Objective outcomes	Unclear risk	Quote: staff were aware that patients were taking part in a trial but "They were not informed that the patients were divided into two groups and participating in an intervention trial." Comment: staff assessed pain by getting participants to indicate their pain on the VAS and so it is possible that staff could have discovered which group a participant was in. The impact on the objective outcomes of staff being aware of group allocation is likely to be minor.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "All the patients with postoperative patient controlled analgesia (PCA), two in the treatment group and one in the control group, were excluded from the analysis of analgesic use postoperatively due to their markedly higher use of opioids, leading to disturbances in the data." Comment: the number of participants excluded from the analyses and the reasons why are described in full.
Selective reporting (reporting bias)	High risk	Comment: all outcomes listed in the methods were reported in the results, but without the trial protocol, it is unclear whether any other outcomes were measured but not reported based on the results. Outcome data were only reported as medians and IQRs, so data were unable to be included in meta‐analyses.
Other bias	Low risk	Comment: no other sources of bias identified.