The number of caesarean sections performed in the UK and other developed countries has increased steadily over the past 20 years. The most recent Department of Health maternity statistics bulletin reported that 22.7% of all deliveries in England and Wales were by caesarean section during the period 2003–04 (Department of Health 2005). This rate had increased from 18.2% in 1994–95, having been 10.4% in 1985. The National Sentinel Caesarean Section Audit found that the caesarean section rate (CSR) in the UK increased with maternal age and was higher in black African or black Caribbean women compared to white women (Thomas & Paranjothy 2001). Previous surveys of caesarean section rates found considerable variation between hospitals in England and Wales, with a range from 6.2% to 21.5% (Francome et al 1993), and that women who attended teaching hospitals were more likely to have a caesarean section than women in other hospitals, but there was also variation in rates between teaching hospitals (Savage & Francome 1993).

The trend towards earlier postnatal discharge from hospital for women delivered vaginally has also been seen following caesarean section. The average hospital stay following caesarean section in 2003–04 was between 3 and 4 days, having been between 4 and 5 days during the 6-year period up to 1994–95 (Department of Health 1997), and 8 days in a large national study conducted in 1983 (Moir-Bussy et al 1984). Only one randomised controlled trial (RCT) included in the Cochrane review of early postnatal discharge (Brown et al 2002) had a study population of 162 US women who had (unplanned) caesarean section (Brooten et al 1994). Outcomes were compared between those who were discharged early and those randomised to usual hospital discharge. The 61 women in the early discharge group were sent home a mean of 30.3 hours earlier and were significantly more satisfied with care than the usual care group.

It is likely that all but the most immediate requirements for postoperative care of the woman and her wound will be in the community and more complications and infections will be diagnosed there (Beattie et al 1994). Midwives in the community will therefore continue to see an increasing number of women who have had a caesarean section, and at an earlier stage of their postoperative course.

There is no universally agreed definition of wound infection following caesarean section. Signs of inflammation such as erythema and induration are not specific to infection and irritation from the suture material can cause similar appearances. It is possible that serous discharge from the wound may occur without infection but infection is likely when such discharge is purulent. Additional evidence of infection, in the form of positive bacteriological culture in material from the wound, is not always obtained. Studies vary in the criteria used to classify the wound as infected. The definition used in the National Survey of Infection in Hospitals published in 1981 was that of inflammation or discharge of pus and positive bacteriological culture (Meers 1981). This was the definition used in a national survey of wound infection following caesarean section in 1984 (Moir-Bussy et al 1984). It was also the basis for the definition used in a prospective study following caesarean section by Parrott et al (1989), but in this study a wound swab or pus sample was taken for bacteriological culture from all patients with signs of wound infection or with postoperative pyrexia. A positive culture was accepted as evidence of wound infection with or without signs of inflammation or sepsis. In another study of postcaesarean women, the diagnosis was made on clinical features alone (Pirwany & Mahmood 1997). Leigh et al (1990) included cases with either clinical or bacteriological findings of infection. In one survey of general surgical patients, follow-up extended into the community, and the definition of wound infection included cases where antibiotics were used to treat a previously uninfected but inflamed wound (Weigelt et al 1992).

Moir-Bussy et al (1984) undertook the first major prospective study of the incidence of wound infection after caesarean section in the UK. The method of selection of the 31 participating hospitals is not entirely clear so some reservations about generalisability of the findings remain. Survey questionnaires were completed for 2370 women having a section over a 12-week period in 1983. An overall infection rate of 6% (n = 141) was reported, with a range of zero to 20.5% between hospitals. In the study, definitions of classification terms used were clear but there is no evidence that objectivity, consistency and completeness between and within centres were achieved. Indeed, although the definition of wound infection used was inflammation or sepsis and a positive bacteriological culture in material taken from the wound, not all centres involved in the study collected swabs for culture on all women with clinical presentation. Culture from inflamed wounds was positive in 80% of cases and, in a later paper, the team suggest that if this rate were applied to the wounds recorded as inflamed but not cultured, the infection rate would have been 14% (Thompson et al 1987).

This study demonstrates the difficulty arising from the lack of a universal method of classifying wounds as infected when determining its incidence. Estimates of wound infection rates from studies which required positive bacteriological culture in the classification are: 4.1% (Hawrylyshyn et al 1981) and 7% (Hillan 1995). The rate was 11.3% from the study where wound infection was classified on positive culture with or without clinical evidence of inflammation or sepsis (Parrot et al 1989).

Higher rates were reported when clinical findings alone were taken as evidence of infection in the absence of a positive bacterial culture, with rates of between 15.8% and 29.2% (Henderson and Love, 1995 and Leigh and Emmanuel, 1990). The wound infection rate was 25.3% in Beattie et al’s (1994) study which, in addition to clinical or bacteriological evidence, included cases where antibiotics were prescribed for wound infection without the study team having any further information of the clinical features. Follow-up in this study was also extended beyond discharge from hospital which will further inflate the rate. In the report of a self-completion questionnaire survey at 3 months postpartum, of 444 women who had had a caesarean section, 121 (27%) reported having had a wound infection at some point (Hillan 1992). It seems that only for 43 of these were antibiotics prescribed.

The trend is for rates to be higher when the diagnosis is made without requiring positive culture. Estimates based on positive culture alone are also subject to error. In the Moir-Bussy survey, Staphylococcus albus was the commonest organism cultured (Moir-Bussy et al 1984). The authors themselves acknowledge that this organism is an unlikely pathogen. If the cases where this was the sole organism cultured were removed from the numerator, the wound infection rate would have been 4.5% rather than 6%, but many of the cases where this was the sole organism cultured had clinical evidence of a wound infection. Classification for study purposes may differ from routine management, however, and in practice antibiotic therapy may well be instituted without culturing the pus or taking a swab of an inflamed wound.

As noted earlier, wound infection rates differed with time in Leigh et al’s study of 202 women having caesarean sections in 1985 and 196 women in 1987 (Leigh et al 1990). All women in this observational study had their wounds assessed daily (excluding weekends) by the study laboratory nurse until discharge from hospital. Swabs were taken from all inflamed wounds, whether or not there was discharge from the wound. Wound infections were classified into two groups: clinical, where there was inflammation and discharge but culture did not show any pathogenic bacteria; and bacteriological, where recognised pathogens were present. The incidence of clinical wound infection was 20% in 1985 and 15.8% in 1987, but bacteriologically proven infections were 12.5% and 5.1% respectively. Since the same methodology was used for each series, with no significant difference in the caesarean section rate, it is surprising that the reported incidence of wound infection is so different. No explanations are offered by the authors. Other audits of different time periods have largely endeavoured to show an effect of antibiotic prophylaxis on infection rate (Nice et al., 1996 and Pirwany and Mahmood, 1997).

The onset of wound infection following caesarean section has been described between the first and 15th postoperative day, most commonly manifested on day 4 or 5 (Leigh and Emmanuel, 1990 and Moir-Bussy et al., 1984). Since some women will not develop the infection until after discharge from hospital, the true incidence of wound infection is probably even higher than that reported in these studies. The majority of studies of wound infections were hospital-based retrospective chart reviews, leaving little scope for follow-up to identify infections after discharge. In almost all cases surveillance was limited to the hospital inpatient stay and although in one study observation was stated to be for 6 weeks, no indication was given of the completeness of hospital case-notes regarding the period following discharge (Pirwany & Mahmood 1997). Hulton et al (1992) showed that the overall rates of infection following caesarean section were underestimates if based on inpatient data alone. They ascertained the number of infections which occurred following discharge by writing to the woman’s surgeon at 6 weeks postpartum. The wound infection rate rose from 0.3% in 318 women who had caesarean sections in the 5-month period before this postdischarge method of ascertainment to 3.9% in the 5-month period after it was initiated (n = 500).

Three studies identified women prospectively (Beattie et al., 1994, Nice et al., 1996 and Parrott et al., 1989) but in one, surveillance was completed on day 6 post section (Parrot et al 1989). Nice et al’s study (1996) of 628 postsection women was prospective and follow-up was continued after discharge by the community midwife but surveillance stopped at the 10th postoperative day. Beattie et al (1994) also identified 428 women prospectively and followed them up beyond hospital discharge but it is not stated for how long. Of the 83 wound infections diagnosed in the 328 women on whom the results are based, 30 (36.2%) were diagnosed after discharge from hospital.

It is likely, however, that findings from the Weigelt et al (1992) general surgical wound surveillance programme are paralleled for caesarean section patients. For 30 days after surgery the programme followed 16,453 consecutive patients operated on in various surgical specialties in Dallas, Texas, during a 6-year period between 1983 and 1989. Surveillance was by infection control practitioners (ICPs) and clinic nurses instructed in wound evaluation by the ICPs. Surgeons were not involved with the classification of wounds to reduce any observer bias. Of all surgical wound infections that were diagnosed, 35% (516) first became apparent after discharge from hospital. Less than half (47%) of wound infections were manifest by the seventh postoperative day, 78% were diagnosed by day 14 and 90% by day 21. A small number of infections continued to be first diagnosed between days 21 and 30.

Of the wounds that did become infected, those which occurred in obese patients were more likely to be first diagnosed after discharge rather than in hospital.

As discussed previously, there is no standard definition of what constitutes a wound infection, but the signs and symptoms suggestive of wound infection have been described above: pyrexia, localised pain and erythema, local oedema (induration), excess exudate, pus and offensive odour (Meers, 1981 and Morrison, 1992). The diagnosis will therefore be made on the basis of patient reporting of pain, discharge, separation or fever and by visual inspection of the wound.

If a wound infection is suspected, a swab should be taken for identification of the organism and antibiotic sensitivity testing. The sample should be collected before the wound is cleaned, avoiding the surrounding skin, which may be colonised by different organisms from the one(s) causing the wound infection (Dealey 1994).

Apart from the need to relieve pain and deal with discharge, the concern with any wound infection is the possibility of breakdown. Dehiscence (separation of the wound edges) may be superficial, involving only the skin layer, or may extend to the deep layers. In severe cases there may be protrusion of the abdominal cavity contents (burst abdomen).

Most caesarean sections are performed using a Pfannenstiel skin and lower uterine segment incisions. In a series of 1300 caesarean sections, the majority of which were performed through Pfannenstiel incisions, none dehisced, although six cases of dehiscence occurred in the series following midline incisions (Donald 1979). Denominators for these figures were not quoted. Where the incision is midline, there is a greater risk of both dehiscence and herniation, but no published evidence of actual rates of occurrence following caesarean section were found. In abdominal laparotomy incisions in general surgical patients, the rate of burst abdomen (complete dehiscence) is of the order of 1–2% (Ausobsky et al., 1985 and Bucknall, 1983).

Due to these concerns, women in whom wound infection is detected while inpatients are often kept under observation for longer. In the Moir-Bussy study (1984) average inpatient stay increased by 2 days for those who developed postcaesarean wound infection in hospital. The average stay for a woman who did not develop a wound infection in their study was 8 days. In the Canadian study by Henderson & Love (1995), the mean duration of hospital stay was increased in women with wound infection by 0.8 days in first section cases (from a mean of 5.9) and by 0.4 days in second or more sections (from a mean of 5.5 days). Mean stay was inflated more by certain other types of postcaesarean infection, but because of its relative frequency, wound infection had the greatest overall impact on extra bed days required.

As stated earlier, follow-up data on caesarean section patients after discharge from hospital are scarce. Of 30 women in Beattie’s survey who had a wound infection diagnosed after discharge from hospital, two required readmission (Beattie et al 1994). No details are given of the indication or of subsequent treatment requirements.

In relation to all infectious morbidity following caesarean section (endometritis, urinary tract infection, wound infection, etc.), associations with many factors have been suggested (Hawrylyshyn et al., 1981 and Smaill and Hofmeyr, 2002). There is less evidence relating to the risk factors specific to wound infection. There are no systematic reviews of the evidence and, as has been discussed, most of the studies are retrospective case-note reviews relating only to complications identified during the inpatient period.

Using simple univariate comparisons between groups of patients in their survey, Moir-Bussy et al (1984) suggested that the following were statistically significant risk factors for wound infection following caesarean section: higher weight/height ratio (as recorded in early pregnancy); higher numbers of vaginal examinations during labour; longer duration of labour; not being nursed in a ward where only caesarean section patients were nursed; alcoholic chlorhexidine not being used for skin preparation preoperatively; the skin incision being paramedian or midline rather than Pfannenstiel; having a drain of any sort; skin closure issues; having a plastic dressing rather than a material one; and being delivered by section in a unit where less than 50 caesarean sections were performed over the 12-week period of the survey (Moir-Bussy et al 1984).

Factors also considered in the Moir-Bussy survey, but found not to be statistically significant included: social class (proportionately more women from social classes IV and V were affected, but this difference was not statistically significant); emergency versus elective caesarean; duration of ruptured membranes; epidural or general anaesthesia; anaemia; and being delivered in a teaching or non-teaching hospital.

In their prospective study of 428 women who had a caesarean section, Beattie et al (1994) also described a link between obesity and wound infection, although they used last recorded weight as the index of obesity. Complete data were only available for 76.6% (n = 328) of the population in this study, and 44.2% of these women had some form of antibiotic prophylaxis. The study also showed a statistically significant, negative association between maternal age and wound infection but failed to show a statistically significant association between number of vaginal examinations and wound infection. Although there was a proportionately higher rate of infection for patients who had a drain inserted at the operation, this was not statistically significant. The authors state that a multivariate analysis method was used and the protective effect of prophylactic antibiotics was very strong so that demonstration of independent effects of other factors may be prohibited.

In general, studies have not attempted to clarify whether postoperative factors beyond the first day influence development of infection in the wound.

In a retrospective chart review of 1335 caesarean section cases between 1985 and 1988, Henderson & Love (1995) found that incisional wound infection was higher in first caesarean sections than in repeat sections, and was higher in elective section compared to emergency. Possible explanations suggested for this were decreased use of prophylactic antibiotics in elective sections or that the interval between preoperative shaving and surgery in elective section may predispose to wound infection. Leigh et al (1990) found a higher rate of bacteriologically confirmed infections after elective caesarean section, but no difference for clinical infection.

The evidence from 66 randomised controlled trials was examined in a systematic review undertaken to determine whether prophylactic antibiotic treatment given to women undergoing caesarean section decreased the incidence of febrile morbidity, wound and other infections, or serious infectious complications (Smaill & Hofmeyr 2002). The review included trials that compared any prophylactic antibiotic regimen with placebo or no treatment. The lack of consistent application of a standard definition for wound infection (and endometritis) was noted in the background to the review and, in the trials, endometritis and wound infection were usually defined clinically, without bacteriological confirmation.

The main outcome measure reported was all infectious morbidity and the findings of the review were that ‘the use of prophylactic antibiotics at caesarean section resulted in a major, clinically important and statistically significant decrease in the incidence of fever, endometritis, wound infection, urinary tract infection and serious infection after caesarean section’. Only for endometritis was a measure of effect given, with its rate of occurrence reduced by 73% (95% confidence interval (CI) 0.68–0.77) for all women undergoing the procedure. Prior to the review it was controversial whether prophylaxis should be reserved for cases in which the risk of infection was higher, but the size of the reduction in endometritis was consistent even when elective and non-elective sections were considered separately.

Hospital stay was reduced by 0.34 days (95% CI 0.10–0.51) in those who received prophylaxis, but there was too little information in general to allow costs of the two strategies to be compared.

In the data from trials where elective caesarean sections were identified, no statistically significant reduction in wound or urinary tract infection was seen with prophylaxis in this group. Although this could be because the trials were too small, even collectively, to show a positive effect, the reviewers had to conclude that while prophylaxis is recommended based on the reduction in endometritis, the rate of wound infections following elective caesarean section may not be seen to fall.

The review recommended that all units should have a policy of administering prophylactic antibiotics at caesarean section. A policy of not treating all women undergoing caesarean should only be considered in units that have prospectively confirmed a low rate of infection in women using methods where follow-up was long enough to include late complications.