Karlsson 2000.

Methods	Twelve‐week, randomised, multicentre, double‐blind study.
Participants	In‐ or out‐patients being treated in psychiatric hospitals, psychiatric specialist or general practices, or geriatrics units. Patients were to have a diagnosis of major depression (DSM‐III‐R criteria), a Montgomery and Asberg Rating Scale for Depression (MADRS) total score of ≥ 20 and a Mini Mental State Examination (MMSE) total score of at least 16. For patients with a MMSE score of 16‐24, the DSM‐III‐R diagnosis forms for dementia were completed. Exclusion criteria: patients having schizophrenia or related psychotic disorder, neurological disease other than vascular or primary degenerative dementia, focal cortical deficit or chronic drug or alcohol abuse. Patients with severe somatic disorders, such as cardiac, renal, hepatic or endocrinological disorders or blood laboratory abnormalities, which, in the opinion of investigator, interfered with participation in the study. Patients were not to have received other antidepressants during previous 4‐7 days; irreversible MAO‐inhibitors (A or B), lithium or carbamazepine during the previous 2 weeks, or fluoxetine during the previous 5 weeks. Patients were also excluded if they had received electroconvulsive therapy within the previous 8 weeks, oral or parenteral neuroleptics during the previous week, depot neuroleptics during the previous 4 weeks, an investigational drug during the previous 3 months, or were known to be intolerant to or have had a non‐response to the study drugs. Patients at risk for suicide in the investigator's opinion and patients treated with oral anticoagulants. Age: 65 years or older.
Interventions	Citalopram: 163 participants. Mianserin: 173 participants. Citalopram dose range: 20‐40 mg/day Mianserin dose range: 30‐60 mg/day
Outcomes	Primary outcome: mean change at endpoint on MADRS. Secondary outcomes: mean change in Clinical Global Impression‐Severity (CGI‐S) of Illness and Clinical Global Impression‐ Improvement (CGI‐I) scales, Gottfries‐Brane‐Steen (GBS) sub‐scale 3 ("emotional functions") and sub‐scale 4 ("symptoms common in dementia disorders") and MMSE. In addition, a modified Well‐Being Questionnaire was completed at baseline and week 12.
Notes	This study was funded by Lundbeck (citalopram manufacturer).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No clear information about sequence generation. Quote: "Patients were randomly assigned".
Allocation concealment (selection bias)	Unclear risk	No information about allocation concealment.
Blinding (performance bias and detection bias) All outcomes	Low risk	Quote: "to ensure blinding, the citalopram and mianserin tablets were identical in appearance and were taken once daily, preferably in the evening".
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Quote: "two different populations were analysed: for primary efficacy analysis the efficacy population was chosen. (...) For secondary analysis, the Intention‐to‐treat (ITT) population was chosen. Primary and secondary efficacy variables were evaluated in both of these populations".
Selective reporting (reporting bias)	High risk	MADRS score and remission rate are missing.
Other bias	Unclear risk	Sponsorship bias cannot be ruled out.