The effect of early versus delayed postcaesarean feeding on women’s satisfaction: a randomised controlled trial



This article is corrected by:

  1. Errata: Errata Volume 115, Issue 5, 674, Article first published online: 7 March 2008

Dr GH Izbizky, Division of Obstetrics, Hospital Italiano de Buenos Aires, Potosí 4135 (C1199ACI), Buenos Aires, Argentina. Email


Objective  To evaluate the effect of early versus delayed feeding after caesarean section on the woman’s satisfaction.

Design  Randomised, controlled trial.

Setting  Tertiary care hospital.

Population  Healthy pregnant women were enrolled for the study during antenatal care visits.

Methods  Uncomplicated singleton pregnancies undergoing a planned or intrapartum caesarean section performed under regional anaesthesia were randomly assigned to either (1) a ‘delayed feeding’ group who started oral fluids 4 hours after surgery with diet introduced at 24 hours; or (2) an ‘early feeding’ group who were offered a regular diet within the first 8 hours.

Main outcome measures  Primary outcome was the woman’s satisfaction measured with a visual analogue scale (VAS) before their hospital discharge. The secondary outcomes were: pain, anorexia, abdominal distension, persistent nausea and/or vomiting, time to the first bowel movement and passage of flatus.

Results  Two hundred women were recruited, with 103 randomised to delayed feeding and 97 to early feeding. The woman’s satisfaction (mean VAS ± SD) was similar in both groups; 73 ± 17 mm in the delayed feeding group and 77 ± 13 mm in the early feeding group (P= 0.12). A statistically significant difference was observed in mean postoperative pain: 29 ± 13 mm in the delayed feeding group versus 24 ± 11 mm in the early feeding group (P= 0.008). No other significant differences in postoperative variables were recorded, and there were no major postoperative complications observed in either group.

Conclusion  Early feeding after uncomplicated caesarean in low-risk women is equivalent in terms of the woman’s satisfaction and the reduced perceived pain.


Caesarean section is the most common major hospital surgical procedure performed in the industrialised world, accounting for more than one-fourth of all deliveries in the USA in 20031 and 21% of all deliveries in England in 2001.2 Even higher rates have been reported in Latin America.3,4 An increasing number of women are, therefore, being exposed to this procedure.

Traditionally, women who had a caesarean section had solid food withheld for the first 24 hours in the belief that this would prevent gastrointestinal complications. However, several clinical trials5–12 and a systematic reviews13 have shown that early feeding is as safe as the traditional progressive approach. Moreover, some additional benefits have been reported such as a more rapid return of bowel sounds and regular oral diet and a shorter hospital stay.12,13 Although information about the safety of early feeding after caesarean section appears conclusive, the effect of the different postoperative feeding approaches on women’s satisfaction has not been well evaluated with the only data available being from one quasi-randomised trial14 and an audit.15 In addition, in 2007 Teoh et al. published a randomised trial in which satisfaction pertaining to the feeding regimen was measured as a secondary outcome on a visual analogue scale (VAS).16 The aim of the present study was to evaluate the effect of early versus delayed feeding after uncomplicated caesarean section on the woman’s satisfaction.

Materials and methods

Between December 2003 and 2004, 625 women were enrolled for the study during antenatal care visits at the Obstetric Division of Hospital Italiano de Buenos Aires, and those who finally underwent a caesarean section were randomised in the immediate postoperative period if they were eligible (Figure 1). The study protocol and informed consent were approved by the Institutional Review Board of the Hospital.

Figure 1.

Study profile.

The inclusion criteria for randomisation were a term singleton pregnancy with a planned or intrapartum uncomplicated caesarean section performed under regional anaesthesia. Exclusion criteria were use of tocolytic drugs, general anaesthesia, history of bowel surgery, prenatal diagnosis of fetal anomaly, maternal disease and intraoperative or immediate postoperative major complications. The allocation sequence was generated by a computer random number list in permuted blocks of ten by staff not involved in patient care. Allocation concealment was achieved by placing the allocation in sequentially numbered, opaque, sealed envelopes. The envelopes were kept by operating room personnel and opened, once participant data were written on it, after the end of the caesarean section.

Women were randomly assigned to one of two interventions. The ‘delayed feeding’ group started oral fluids 4 hours after surgery, followed by regular diet at 24 hours. The ‘early feeding’ group was offered regular diet within the first 8 hours. Both groups were offered the same analgesic regimen (dextropropoxyphene napsylate 98 mg plus dipyrone 400-mg tablets, 8-hourly), but had the option not to take the medication if they did not need it. Analgesic use was reported as the percentage of the total prescribed dose that was used.

The primary outcome was patient satisfaction before discharge from hospital, measured using a VAS.17,18 A 10-cm-long scale was used, with adjectival descriptions at the ends and intermediate positions. Participants were provided with a questionnaire containing the VAS, which was presented with a statement explaining what was intended to measure. This ‘framing’ statement was introduced after initial testing of the scale yielded a high variability. The variables evaluated as secondary outcomes were: self-reported pain, time to first passage of flatus and bowel movements, incidence of anorexia, nausea/vomiting and abdominal distension. Pain was measured and recorded daily by the woman on a 100-mm VAS (0 = minimum pain, 100 = maximum pain)19,20 and was expressed as the mean of the repeated daily measurements for each woman. Anorexia was defined as a lack of appetite or reluctance to accept food in the first 24 hours after surgery. Time to first flatus passage and bowel movement was recorded in hours after the caesarean section, as defined by Patolia et al.8 The women were examined by a staff physician and an obstetrics/gynaecology resident twice daily. Although they were not told of the woman’s allocation, true blinding was not always possible because of the nature of the intervention.

From a previous pilot study, the prior estimate of the variance of the VAS satisfaction score in the delayed feeding women was 1600. After framing the question, with a clear statement of what was intended to measure, the variance was reduced to 250. A difference of 10 mm to each side between groups was considered clinically relevant according to this experience. The trial was designed to demonstrate the equivalence between the two groups (defined as <10 mm). With a power of 90%, a two-sided type I error of 0.05, and considering a 10% loss, the sample size was estimated to be 94 women per group.21,22 The two treatments were considered equivalent if the 95% CI of the difference between the means was within the limits of −10 to +10 and contained the bound zero.

Normality was checked through graphic methods and the Shapiro–Wilk test. Differences between continuous variables were tested for significance by an unpaired t test if normally distributed, and the Wilcoxon rank–sum test otherwise. For categorical variables, the chi-square test or Fisher’s exact test was used. Subgroup analysis by type of caesarean section was planned. The primary analysis was based on treatment received (per protocol), and a secondary ‘intention-to-treat’ analysis was performed. For data entry MS Excel (Microsoft Corporation, One Microsoft Way, Redmond, WA, USA) was used and STATA version 8.0 (StataCorp., TX, USA) for data analysis.


During the study period, 625 women were enrolled for the study antenatally. Two hundred were randomised immediately following their caesarean section: 103 were assigned to the delayed feeding group and 97 to early feeding group (Figure 1). The baseline characteristics of women in the two groups were similar (Table 1).

Table 1.  Women’s characteristics, surgical procedures and neonatal outcomes according to groups
 Delayed feeding (n= 99)Early feeding (n = 101)
Maternal age, mean years (±SD)31.9 (±5)32.2 (±5)
Nulliparity (%)5240
Gestational age at delivery, mean weeks (±SD)39.3 (±1)39.3 (±1)
Previous abdominal surgery (%)48 (48)54 (54)
Surgical procedure
Planned caesarean section (%)64 (64)69 (68)
Intrapartum caesarean section (%)35 (35)32 (32)
Fasting, mean hours (±SD)11.6 (±4.6)11.8 (±4.6)
Surgery length, mean minimum (±SD)36.4 (±12)39.6 (±14)
Severe abdominal adhesions (%)4 (4)8 (8)
Visceral peritoneum closure (%)22 (22)18 (18)
Parietal peritoneum closure (%)31 (31)36 (36)
Indication for caesarean section as reported by clinicians
Repeat caesarean section (%)43 (44)54 (53)
Malpresentation (including breech) (%)6 (6)6 (6)
Failure to progress (induction/in labour) (%)37 (37)26 (26)
Presumed fetal compromise (%)6 (6)5 (5)
Other (%)7 (7)10 (10)

Four women in the delayed feeding group received food in the first eight postoperative hours. No crossover was observed in the opposite direction. Mean satisfaction ± SD was 73 ± 17 mm in the delayed feeding group and 77 ± 13 mm in the early feeding group. The difference was not statistically significant (−3.6, 95% CI −7.57 to −0.96, Figure 2). A statistically significant difference was observed in the mean pain expressed as VAS: 29 ± 13 mm in the delayed feeding group versus 24 ± 11 mm in the early feeding group (P= 0.008, Figure 3). Use of the prescribed analgesia was similar in the two groups. (70 versus 71%, P= 0.48). There were no significant differences in other postoperative variables recorded, including hospital stay (Table 2).

Figure 2.

Mean satisfaction according to treatment received. The confidence interval comparison approach is illustrated, in which equivalence is established when the bounds of the two-sided 95% CI lies within the equivalence zone. Top, margin of equivalence established according to previous experience. Middle, equivalence is established for the absolute difference (mm VAS) between early feeding and delayed feeding groups. Bottom, subgroup analysis shows that in the planned caesarean section group the confidence interval covers at least some points that lie outside the equivalence range, so that differences of potential clinical importance remain a real possibility and equivalence cannot safely be concluded.

Figure 3.

Box plot showing mean pain perception (mm VAS) according to treatment received. The line inside the box represents the median of the distribution. The extremes of that box are the 25th and 75th percentile. The extremities of the whiskers are the minimum and maximum values of the data.

Table 2.  Trial outcomes
 Delayed feeding (n= 99)Early feeding (n=101)P
  • *

    Wilcoxon rank–sum test.

  • **

    t test.

  • ***

    Chi-square test.

Satisfaction level, mean minimum (±SD)73 (±17)77 (±13)0.12*
Pain, mean minimum (±SD)29 (±13)24 (±11)0.008*
Time to bowel movement, mean hours (±SD)15 (±11)12 (±11)0.50**
Time to passage of flatus, mean hours (±SD)23 (±12)22 (±14)0.72**
Hospital stay, mean days (±SD)2.5 (±0.5)2.4 (±0.5)0.16**
Abdominal distension (%)16 (17)16 (16)0.95***

Subgroup analysis by type of caesarean section showed that the early feeding group had a higher satisfaction level than the late feeding group when they undergo planned caesarean section (absolute difference −6.4, 95% CI −11.7 to 1.1), but no difference in the experience of pain. Conversely, in the emergency intrapartum caesarean section group, in those who had early feeding there was no difference in satisfaction compared with the late feeding group but significantly less pain. (Table 3, Figure 2).

Table 3.  Trial outcomes stratified by type of caesarean section (planned versus intrapartum)
 Delayed feeding (n= 99)Early feeding (n= 101)P
  • *

    Wilcoxon rank–sum test.

Satisfaction level, mean mm (±SD)
Planned caesarean section (n= 133)71.5 (13)78.0 (17)0.01*
Intrapartum caesarean section (n= 67)76.1 (14)73.8 (15)0.51*
Pain, mean mm (±SD)
Planned caesarean section (n= 133)27.3 (±14)24.5 (±11)0.22*
Intrapartum caesarean section (n= 67)30.8 (±11)22.9 (±11)0.005*

No major postoperative complications were observed prior to discharge in either group. In the early feeding group, one obese woman with three previous abdominal operations and a midline incision was readmitted on tenth day with an incisional hernia, and a second woman was readmitted due to appendicitis on the 15th postoperative day.

The intention-to-treat analysis showed a small but statistically significant difference in the woman’s satisfaction in favour of early feeding group (difference −4.29, 95% CI −8.55 to −0.04). This did not, however, reach the ten points that was defined a priori as clinically relevant. The rest of the variables showed similar results to the per protocol analysis.


This study shows that the satisfaction rates obtained following early feeding after caesarean section are equivalent to those obtained following delayed feeding. However, those in the early feeding group reported lower levels of postoperative pain, especially when the caesarean section was conducted as an emergency. To our knowledge this is the first randomised control equivalence trial to evaluate the impact of postpartum medical care recommendations on women’s satisfaction and perceptions as a primary outcome.

We are aware that satisfaction is a complex and multidimensional psychological response to life events.23 Although there is no such thing as a single index of satisfaction, the approach used in this trial has been previously validated. Several studies have assessed satisfaction using VAS16,24,25 and these have also been adopted in quality-of-life instruments.18 Health professionals are increasingly encouraged to involve women in treatment decisions, recognising them as experts with a unique knowledge of their own health and their preferences for treatments,26 and recent evidence supports this view.27,28 Although women may wish to be more involved in the decisions concerning their obstetric interventions, information is lacking on the experiences and perceptions concerning postoperative caesarean birth care.

This clinical trial places the focus upon equivalence between the randomised groups. The next step is to determine whether the observed results are clinically relevant. The interpretation of ‘satisfaction’ or ‘quality-of-life’ scores is not always straightforward, and it is not easy to identify a single value that will serve as a clinically relevant measurement. However, it is important that in assessing the benefits of healthcare interventions, researchers should not only include clinical endpoints but also incorporate individuals’ preferences in order to offer a more individually oriented outcome. In this context, the definition of a ‘clinically significant benefit’ is a matter of a woman’s personal judgement as well as that of clinicians.

Caesarean section rates have been rising progressively during recent decades all over the world. Many reasons have been described for the increasing rate, including women’s choice.20 Facilities and human resources necessary for perinatal care could be significantly influenced by this trend, in addition to healthcare costs. It is important, therefore, to try and reduce morbidity and increase satisfaction rates from this common procedure. Teoh et al.16 randomised 196 women undergoing caesarean section under spinal anaesthesia, to compare the incidence of ileus in early and late feeding groups. As a secondary outcome, they measured maternal satisfaction. Although they used a slightly different outcome measurement from ours, they found a higher rate of satisfaction in the early-fed group (90 versus 60 on the VAS scale P < 0.001). They also reported a very low incidence of ileus (3%).

The strengths of this trial are the noninferiority design, the comparison of two active policies,21,22 the high acceptance rate (which minimised selection bias) and the minimal crossover between the study branches. To focus only on variables of physical discomfort is a limitation of this study. Some other attributes of woman’s satisfaction would deserve evaluation in further trials. Another limitation is the external validity of our findings because our participants may represent a limited socio-economic and cultural population. This trial was performed on low-risk, middle-income, educated women.

The primary estimate of equivalence results from a per protocol analysis; however, the ‘intention-to-treat’ analysis showed a small but statistical significant difference in favour of early feeding. Our rationale for using a per protocol analysis was that in a comparative trial, where the aim is to decide if two treatments are different, an intention-to-treat analysis is generally conservative: the inclusion of protocol violators and withdrawals will tend to make the results from the two treatment groups more similar. However, for an equivalence trial this effect is no longer conservative: any blurring of the difference between the treatment groups will increase the chance of declaring equivalence.

This trial shows that a programme of early feeding after a caesarean delivery in low-risk women is clinically equivalent in terms of women satisfaction, is as safe as the traditional approach and has beneficial effects on women’s perceived pain. These findings add to the already robust literature on the safety and benefits of early feeding after uncomplicated caesarean section. We believe that there is now enough evidence for clinicians worldwide to safely implement early feeding following caesarean section.29


This trial was supported with the Division’s own funds for research.


The authors thank the effort of the Residents of Obstetrics and Gynecology, Department of Nurses and labour and delivery room personnel of the Hospital Italiano in the execution of this trial.

Journal club

It has been a strongly ingrained part of the hospital doctor’s ritual to listen for bowel sounds in the immediate postoperative days following any abdominal surgery. The thought of giving women food on the same day as their caesarean irrespective of whether bowel sounds are present or not will horrify many traditionalists. But there is now increasing evidence that early feeding is not only safe but also may actually improve outcomes. This randomised trial from Brazil adds further weight to that argument. Although the two regimens were equivalent in terms of maternal satisfaction, there were some secondary beneficial effects.

Discussion points

  • 1Background: In your hospital, how long is the average woman starved preoperatively for a routine caesarean? What effect does perioperative starving have on maternal and fetal/neonatal health? What are the potential risks of early postoperative feeding of women who have had caesarean sections?
  • 2Technical: What were the two regimens compared in this study? Would you have chosen the same two regimens to compare? The researchers chose the women’s satisfaction at their main outcome: do you think that this is the most important question? What is an equivalence trial? Do you think that its use was appropriate in this study?
  • 3Clinical practice: What is the current practice in your hospital regarding restarting fluids and diet after a caesarean? Do you know of any national or local recommendations? Do you think that the evidence is now strong enough for you to give routine early feeding?
  • 4Hypothetical: Do you think that day case caesareans will ever be possible? What would be the advantages and what are the obstacles in the way of achieving this?

Correspondence: Dr A Weeks, School of Reproductive and Developmental Medicine, University of Liverpool, Crown Street, Liverpool L8 7SS, UK. Email