Within the prospectively collected, institutional review board-approved database, the authors included 632 consecutive patients who underwent elective open RRP in the Martini-Klinik Prostate Cancer Center affiliated with the University Medical Centre Hamburg-Eppendorf. The study was approved by the local ethics committee of the Medical Board Hamburg (WF-040/17, July 11, 2017), and the trial was registered at ClinicalTrial.gov (Identifier ‘NCT03565705’). All patients gave consent to participate in the prospectively collected database. The manuscript adheres to the applicable SQUIRE 2.0 (Revised Standards for Quality Improvement Reporting Excellence) guidelines from September 15, 2015.

With more than 2500 RRPs carried out annually, the Martini-Klinik is in terms of the number of operative procedures the world’s largest prostate cancer clinic. Twelve specialized surgeons perform RRP either by open or robotically assisted minimally invasive procedures. Before March 1, 2017, the standard anaesthesia technique for open RRP in the Martini-Klinik consisted of spinal anaesthesia combined with general anaesthesia. That standard anaesthesia technique was changed on March 1, 2017, from the combination of general anaesthesia preceded by spinal anaesthesia to general anaesthesia alone. Data were collected from the last 318 consecutive men who underwent open RRP before March 1, 2017, with spinal anaesthesia (SPA), and the cohort of the next 318 patients after March 1, 2017, under general anaesthesia only (GA).

We included only male patients aged 18 years or older, because the study investigated opioid consumption during and after open radical retropubic prostatectomy. Patients with chronic pain therapy (e.g., out-of-hospital opioid therapy) were excluded from the study. The authors included all patients who received only general anaesthesia without spinal anaesthesia after March 1, 2017, and all patients who received a combination of both until February 28, 2017.

All patients with and without intraoperative spinal anaesthesia received the same perioperative care following local standards. Upon arrival in the operation room, the anaesthesiologists started convective air-warming to minimize perioperative hypothermia. Routine monitoring included electrocardiography, non-invasive blood pressure measurement, pulse oximetry, Bispectral index monitoring (BIS), acceleromyography for train-of-four (TOF) counts, body temperature measurement (a probe placed in the oesophagus in the GA group, and above the laryngeal mask in the SPA group) and capnography in both groups.

The patients in the SPA group received spinal anaesthesia before induction of general anaesthesia. The sterile lumbar puncture was carried out on the awake patient sitting on the operating room table using a standard 26-gauge pencilpoint needle between L2 and L3 or between L3 and L4 after local anaesthesia on the puncture site. Spinal anaesthesia was achieved by injecting isobaric bupivacaine 0.5% with a volume between 2.8 and 3.5 ml (14 to 17.5 mg, depending on the patient’s height) in the cerebrospinal fluid. The placement and pharmacological effect were confirmed by the loss of motor function in the lower extremities. Dermatomal levels were determined by testing the patient’s ability to discriminate between heat and cold. After clear confirmation of the presence of spinal anaesthesia, the patients in the SPA group received a sufentanil bolus of 0.5 μg·kg−1 followed by a propofol bolus of 2 mg·kg−1 body weight. The airway was secured by inserting a third-generation laryngeal mask.

For the induction of general anaesthesia, the patients in the GA group received a sufentanil bolus of 0.5 μg·kg−1 followed by a propofol bolus of 2 mg·kg−1 body weight and rocuronium 0.5 mg·kg−1. The airway was secured with tracheal intubation.

After airway management positive pressure ventilation was started, and anaesthesia maintenance was performed with either sevoflurane 1.7–2.0 Vol% (minimal alveolar concentration MAC 1.0) or propofol infusion both with a target range of BIS (bispectral index, depth of sedation monitoring) values between 40 and 50. Further application of sufentanil was administered at the discretion of the attending anaesthesiologist using clinical signs of nociception (e.g., increase in heart rate or blood pressure). Patients received continuous low-dose norepinephrine administration compensating for the vasodilatation associated with narcosis to maintain the mean arterial pressure above 65 mmHg. At the end of surgery, patients in both groups received 1 g metamizole dipyrone i.v. and were transferred from the operation room into the post-anaesthesia care unit (PACU) after extubation.

During the postoperative period in the PACU, patients in both groups received postoperative care following the institution’s standardized treatment protocol. Upon arrival at the PACU, the patient’s level of analgesia was measured on a numerical pain rating scale (NRS) from 0 to 10. NRS ≤ 3 was considered to be no or mild pain, and NRS > 3 was considered to be moderate to intense pain (Hayhurst and Durieux 2016). The NRS was reassessed every 15 min. In cases of NRS > 3 or whenever the patients mentioned pain, patients received a bolus of 3.75 mg piritramide (dose equivalence to morphine is 1:0.7 according to 2.625 mg morphine) with no upper dose limit. On the ward, all patients from both groups received the same, regular and precisely scheduled, prophylactic analgesic treatment plan during the first 2 days after surgery. This multimodal therapy included scheduled, weight-adjusted non-opioid analgesics, anti-spasmic medication and mobilizing physiotherapy.

The primary endpoint of the study was the total amount of postoperative opioid consumption (morphine equivalents) in the PACU. Secondary endpoints were the extent of intraoperative opioid consumption (sufentanil during surgery), maximum postoperative pain measured with the highest score on the NRS, duration of postoperative recovery time (time interval between postoperative tracheal extubation and the patient fulfilling the fit-for-discharge criteria from the PACU to the ward), days in hospital and the occurrence of PONV or shivering in the PACU. Long-term parameters were cancer-free survival and persistence of pain 1 year after surgery. In addition, we collected further perioperative parameters (e.g., blood loss, need for vasoactive agents) to assess the comparability of the two treatment groups.

The sample size calculation was based on a chart review of routine data of the patients having had a prostatectomy between February and March 2017 in the author’s institution due to a lack of adequate published data. In this sample, the authors found a mean consumption of morphine equivalence in the PACU in patients with spinal anaesthesia of 7.7 mg and considered a difference of 20% (i.e., 1.5 mg) to be of clinical relevance. Based on the standard deviation of 5.5 mg in the random sample, the authors calculated that a total sample size of 636 patients (318 per group) has a 90% power to detect differences between the group means at a 5% significance level in a two-sample t-test using the software package PASS 2008 version 08.0.6 (NCSS LLC, Kaysville, UT, USA).

Differences in the primary endpoint were assessed by a two-sided t-test. Additionally, to adjust for potential confounders and group imbalances, the influence of the study group and 17 covariates as fixed effects on the postoperative opioid consumption in the PACU were evaluated with a general linear model. The model was adjusted for 17 baseline variables: age, BMI, ASA status and each patient’s preoperative long-term medication (in 14 categories). Data distributions were assessed graphically via histograms. Analysis of the demographic and procedural data was performed using Mann–Whitney U test or Fisher's exact test as appropriate. The secondary endpoints were analysed in an explorative manner by two-sided t-tests or Mann–Whitney U tests for continuous data or by χ2-tests for categorical data. Cancer-free survival was determined by Kaplan–Meier curves and group differences by log-rank tests. Effects of additional covariates were examined by multivariable Cox regression analyses. Statistical analyses were performed using SPSS 25.0 (IBM SPSS Statistics Inc., Armonk, NY, USA), and two-tailed p-values < 0.05 were considered significant for the primary endpoint. Secondary endpoints were compared in an explorative analysis.