Study	Reason for exclusion
Caglayan 2008	Study not truly randomised. Participants allocated to treatment groups using alternation.
Dryzer 1993	Unable to differentiate between participants with benign and malignant disease in the results section. Also high risk of bias from randomisation method (allocated to treatment groups based on the last digit of their hospital number).
Elayouty 2012	Unclear from text if truly randomised ‐ participants given number on entering study ‐ allocated to bleomycin if number was odd and allocated to povidone group if number was even. Study authors emailed for clarification but no response.
Engel 1981	Study not truly randomised. Participants allocated to treatment groups based on the day of the calendar month.
Gust 1990	Pilot data (not randomised) and randomised data presented grouped together. Unable to split out the pilot group.
Kwasniewska‐Rokicinska 1979	Participants with pleural effusions and ascites included, but unable to differentiate between them in the results section.
Lissoni 1995	Unable to differentiate between pleural, pericardial and peritoneal effusions in the results. No response from study authors.
Maiche 1993	Study not truly randomised. Participants allocated to bleomycin group if met a list of criteria, otherwise given mitoxantrone.
Manes 2000	Study not truly randomised. Participants allocated to treatment groups based on the month of their diagnosis with MPE.
Nio 1999	Participants with pleural and peritoneal effusions included in the study and unable to differentiate them in the results.
Tattersall 1982	Insufficient information in paper to extract required data (e.g. unclear how pleurodesis success was defined and at what time point).