Minig 2009b.

Methods	A prospective randomised controlled study in a single institution. Power calculation was performed a priori. Number of participants randomised: 167 (83 in the early group, 84 in the delayed feeding group) Number of participants analysed: 143 (71 in the early group, 72 in the delayed feeding group), In each group, 12 women were excluded after randomisation because of benign gynaecologic pathology, nongynaecologic pathology, and admission to the ICU for > 24 hours. Analysis: Available case analysis
Participants	Inclusion criteria: Gynecologic oncology patients aged 18 ‐ 75 years, undergoing laparotomy; mean age: 54 years (early group), 57 years (delayed feeding group). The majority of participants had ovarian malignancy, 59% in the early group and 57% in the delayed feeding group.Pelvic and aortic lymphadenectomy were performed in > 70% and in almost 50% of participants, respectively. No significant difference in participant characteristics between the two groups. Location: European Institute of Oncology (IEO), Milan, Italy Enrolment period: January 1, 2007 to November 15, 2007 Exclusion criteria: Preoperative (infections, intestinal obstruction, severe malnutrition, American Society of Anesthesiologists score > 4), intraoperative (total or anterior pelvic exenteration, bowel resection), postoperative (admission to the intensive care unit for > 24 h, final histopathologic diagnosis revealing benign or nongynaecologic disease)
Interventions	Early group: Participants were offered liquids, mineral water (no gas), tea, chamomile infusion, or apple juice during the first 24 hours. If no nausea and vomiting, a regular diet of boiled or grilled beef, chicken, or fish was given starting on day 1 and continued for the entire hospital stay. Delayed feeding group: Participants received nothing by mouth until the presence of bowel sound and the passage of flatus. Then, if no nausea and vomiting, an oral liquid diet was given for 24 hours. If well tolerated, a semisolid diet was given for 1 day before proceeding to a regular diet. Discharge criteria included the tolerance of a regular diet for at least 24 hours with recovery of bowel function, normal clinical parameters and physical examination. All participants underwent bowel preparation and preoperative antibiotics prophylaxis. In addition, a nasogastric tube was placed in all participants during surgery and was removed after the surgery finished. All participants received general anaesthesia. Postoperative analgesia was given via epidural catheter for 3 days (ropivacaine and fentanyl) or as intravascular continuous administration of ketorolac and tramadol in those without an epidural catheter.
Outcomes	Hospital stay Recovery of bowel activity (time to first passage of gas and stool, time to tolerance for solid diet) Intestinal morbidities (presence of ileus, intensity of abdominal pain, presence of nausea and vomiting) Other morbidities (wound infection, abdominal abscess, pneumonia, urinary tract infection, bacteraemia, wound dehiscence, marked postoperative bleeding, anastomotic leak, respiratory failure, cardiovascular instability, renal dysfunction, thromboembolic complications) Participants' satisfaction level and quality of life Analgesic and antiemetic drug requirements
Notes	Funding source: not reported Conflicts of interest: not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "patients were randomized by means of the Web‐based Tenalea randomization system (https://it.tenalea.net/ieo)." Comment: Computer‐generated sequence
Allocation concealment (selection bias)	Low risk	Quote: "patients were randomized by means of the Web‐based Tenalea randomization system (https://it.tenalea.net/ieo)." Comment: Central web‐based allocation
Blinding of participants and personnel (performance bias) All outcomes	High risk	Comment: Because of the study's context, the participants could not be blinded. The physicians taking care of the participants were aware of their study allocation (information from the study's author). It is possible that some outcomes (hospital stay, subjective intestinal morbidities, participants' satisfaction level and quality of life) were influenced by the lack of blinding.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Comment: Nursing staff, the primary outcome assessors, were aware of participants' study allocation (information from the study's author). However, the influence of the lack of blinding on study outcomes was unclear.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "Twenty‐four patients (12 for each group) were subsequently excluded as a result of benign gynecologic pathology, nongynecologic pathology, and admission to the intensive care unit for > 24 h." Comment: Reasons for missing outcome data unlikely to be related to true outcome. Also, the missing outcome data were balanced in numbers across intervention groups, with similar reasons for missing data across groups.
Selective reporting (reporting bias)	Low risk	Comment: The study protocol was published in www.clinicaltrials.gov. The reported outcomes corresponded to those listed in the registered protocol.