Naukkarinen 2000.

Methods	Allocation: randomised. Blindness: double. Duration: 26 weeks + wash‐out (time not stated). Setting: multi‐centre, Finland.
Participants	Diagnosis: acute or chronic schizophrenia (DSM‐IV) with acute symptoms (at least ‘moderately ill’ using CGI 4). N = 60. Age: range 18 to 70 years; mean (olanzapine) 37.4 (perphenazine) 37.7 years. Sex: F 18, M 28 (olanzapine: F 10, M 13 – perphenazine: F 8, M 15). History: not stated. Included: acute symptoms present; 18‐70 years of age. Excluded: not described. Consent: not described.
Interventions	1. Perphenazine: 8mg to 32 mg/day, n = 30. 2. Olanzapine: 5mg to 29 mg/day, n = 30.
Outcomes	Mental state: PANSS; BPRS. Global state: CGI; PGI. Adverse events: UKU‐ConSat; UKU side‐effects scale. Leaving study early; adverse events. Additional medication. Change scores in weight increase. Unable to use ‐ CGI, PGI, UKU‐ConSat, UKU side effects (no mean or SD ‐ authors emailed for further information).
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomised: quote, “performed on the basis of clusters, or blocks, for each investigator, such that patients within a block for a given investigator were randomised in a 1:1 ratio” (p7).
Allocation concealment (selection bias)	Low risk	Randomisation schedule was, quote, “generated on block size, number of treatments, and number of investigative sites included in the study. The study drugs were assigned to medication kits by a computer‐generated random sequence in such a way that there was a 1:1 ratio of olanzapine to perphenazine‐ treated patients” (p7). Participants were assigned on their first visit and those who met inclusion criteria were randomly allocated to double‐blind at the second visit by assignment of a unique kit number.
Blinding (performance bias and detection bias) All outcomes	Low risk	Medication kit contained double‐blind study medication for each participant; blinding of study was preserved by allowing a “minimum number of Lilly personnel” to see the randomisation table and codes before completion of the study (p7).
Incomplete outcome data (attrition bias) All outcomes	High risk	High attrition, with 38% completing the study. Four separate reports provide different numbers of participants included. N = 60 originally randomised, however only n = 46 made reference to follow‐up reports. Of the n = 46, only n = 23 (n = 12 in olanzapine and n = 11 in perphenazine) completed the study. N = 53 included in the safety analysis, and n = 46 included in efficacy analysis.
Selective reporting (reporting bias)	High risk	Data not provided for CGI, PGI, UKU‐ConSat, UKU side effects – authors contact 13/11/13 for request of full‐published version, no response.
Other bias	High risk	Funding: involvement of Eli Lilly, unclear based on provided study report to what extent. Rating scales: raters not stated to be independent of treatment.