El‐Shora 2018.

Methods	Parallel RCT Ethics committee: approved by the ethics committee Informed consents: written informed consents obtained Site: Egypt Setting: 2 centres from university hospital Dates of data collection: from March 2016 to March 2017 Funding: departmental/institutional Registration: PACTR201603001502110 (www.pactr.org).
Participants	145 participants: mean age 43.5 years; sex distribution: 68 females and 77 males Inclusion criteria Patients who underwent elective cardiac surgery for valvular or coronary artery disease through full median sternotomy as a primary procedure Exclusion criteria Refused to participate in the research With preoperative coagulopathy Severe organ insufficiency (e.g. serum creatinine > 3 mg and/or liver dysfunction) Needed reoperation within 24 hours
Interventions	Intervention Epidural analgesia (N = 75) Comparator Bilateral paravertebral (N = 70) Induction: midazolam, fentanyl, propofol, lidocaine, and pancuronium Maintenance: isoflurane and pancuronium Surgery: valve or CABG with CPB
Outcomes	Relevant to this review Risk of mortality Risk of neurological complications Pain scores Tracheal extubation Others Urinary retention Vomiting Acute kidney injury Re‐exploration for bleeding ICU length of stay Hospital length of stay
Notes	Correspondence: email sent 18 November 2018; study authors asked us to extract the information from the trial Conflict of interest: none DOI: 10.1055/s‐0038‐1668496
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Blocked stratified randomization was used to assign participants to 2 groups via 1:1 allocation Randomization sequence was generated randomly online using https://www.randomizer.org/; block size ranged from 4 to 6 participants. Randomization was stratified by participating centres
Allocation concealment (selection bias)	Low risk	On the day before surgery, participants were sorted to 1 of the 2 groups based on blocked single‐blinded randomization
Blinding of participants and personnel (performance bias) All outcomes	High risk	Single‐blinded
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Single‐blinded Cardiac anaesthesia specialist, not participating in data collection or patient follow‐up, performed the block designated for each participant (either bilateral thoracic paravertebral or thoracic epidural block) A nurse collected the data without pre‐knowledge of participants' assigned groups
Incomplete outcome data (attrition bias) All outcomes	Low risk	5 participants were excluded from the paravertebral group: 3 for in‐hospital mortality and 2 for reoperation within 24 hours One participant in the epidural group died in hospital, but data for this participant were included in the analysis
Selective reporting (reporting bias)	Low risk	All results were reported
Other bias	Low risk	Groups well balanced “intention‐to‐treat” principle