Methods to reduce infectious morbidity at caesarean birth

Sepsis

• No evidence was available for this outcome

Surgical site infection

• One randomised controlled trial (n=543) provided very low quality evidence to show that those who received a hydroactive dressing experienced a clinically important decrease in the number of surgical site infections as compared to those who received a standard dressing.

Need for antibiotics

• One randomised controlled trial (n=543) provided very low quality evidence to show that those who received a hydroactive dressing experienced a clinically important decrease in the need for antibiotics as compared to those who received a standard dressing.

Adverse skin events from techniques

• No evidence was available for this outcome

Endometritis

• No evidence was available for this outcome

Women’s experience

• No evidence was available for this outcome

Readmission into hospital

• One randomised controlled trial (n=543) provided very low quality evidence to show that there was no clinically important difference in readmission into hospital between those who received hydroactive or standard dressing.

Sepsis

• One randomised controlled trial (n=1606) provided very low quality evidence to show that for women with raised BMI (≥30 kg/m²), there was no clinically important difference in sepsis between those who received negative pressure wound therapy or standard dressing.

Wound infection/ surgical site infection

• Seven randomised controlled trials (n=3380) provided very low quality evidence to show that, for women with raised BMI (≥30 kg/m²), those who received negative pressure wound therapy may have experienced a clinically important decrease in the number of wound infections or surgical site infections as compared to those who received standard dressing.

  • One of the five randomised controlled trials (n=876) reported its results separately by BMI (women with a BMI between 30 and 34.9 kg/m², and women with a BMI of 35 kg/m² and greater) in both subgroups the point estimate suggested there was a clinically important decrease in the number of surgical site infections for those who received negative pressure wound therapy. However, for the BMI 30–34.9 kg/m² subgroup, the effect was not statistically significant (see appendix M for details).

Need for antibiotics

• Two randomised controlled trials (n=602) provided very low quality evidence to show that, for women with raised BMI (≥30 kg/m²), there was no clinically important difference in the need for antibiotics between those who received negative pressure wound therapy or standard dressing.

Adverse skin events from techniques

• Four randomised controlled trials (n=2303) provided very low quality evidence to show that, for women with raised BMI (≥30 kg/m²), there was no clinically important difference in adverse skin events between those who received negative pressure wound therapy or standard dressing.

Endometritis

• One randomised controlled trial (n=876) provided very low quality evidence to show that, for women with raised BMI (≥30 kg/m²), there was no clinically important difference in the occurrence of endometritis between those who received negative pressure wound therapy or standard dressing.

Women’s experience: reported pain score (days 1 to 7)

• One randomised controlled trial (n=89) provided low quality evidence to show that, for women with raised BMI (≥35 kg/m²), women who received negative pressure wound therapy had a clinically important reduction in pain on days 1–7 post-operatively (score of ≥2 on the Wong Baker faces score) as compared to those who received standard dressing.

Women’s experience: sharp pain at postoperative day 2

• One randomised controlled trial (n=119) provided very low quality evidence to show that, for women with raised BMI (≥30 kg/m²), there was no clinically important difference in sharp pain score on the second postoperative day between those who received negative pressure wound therapy or standard dressing.

Women’s experience: self-rated health status; measured with EQ-VAS

• One randomised controlled trial (n=876) provided low quality evidence to show that, for women with raised BMI (≥30 kg/m²), there was no clinically important difference in self-rated health status between those who received negative pressure wound therapy or standard dressing.

Women’s experience: satisfaction (0–10, higher is better)

• One randomised controlled trial (n=1604) provided low quality evidence to show that, for women with raised BMI (≥30 kg/m²), there was no clinically important difference in satisfaction between those who received negative pressure wound therapy or standard dressing.

Women’s experience: satisfaction (would use this dressing again)

• One randomised controlled trial (n=411) provided low quality evidence to show that, for women with raised BMI (≥30 kg/m²), there was no clinically important difference in satisfaction between those who received negative pressure wound therapy or standard dressing.

Readmission into hospital

• Four randomised controlled trials (n=2297) provided very low quality evidence to show that, for women with raised BMI (≥30 kg/m²), there was no clinically important difference in readmission into hospital between those who received negative pressure wound therapy or standard dressing.

Sepsis

• No evidence was available for this outcome

Wound infection

• One randomised controlled trial (n=320) provided very low quality evidence to show that there was no clinically important difference in wound infection rates between those whose dressing was removed at 6 hours or 24 hours.

Need for antibiotics

• No evidence was available for this outcome

Adverse skin events from techniques

• No evidence was available for this outcome

Endometritis

• No evidence was available for this outcome

Women’s experience: women who were satisfied with the intervention

• One randomised controlled trial (n=320) provided moderate quality evidence to show a clinically important increase in satisfaction with the intervention for those whose dressing was removed at 6 hours compared to those whose dressing was removed at 24 hours.

Readmission into hospital

• One randomised controlled trial (n=320) provided very low quality evidence to show that there was no clinically important difference in readmission into hospital between those whose dressing was removed at 6 or 24 hours.

Sepsis

• No evidence was available to inform this outcome

Surgical site infection

• Four randomised controlled trials (N=3059) provided low quality evidence to show a clinically important decrease in the number of surgical site infections for those who received chlorhexidine-based alcohol skin preparation compared to those who received iodophor-based skin preparation (including alcohol and aqueous based preparations).

Need for antibiotics

• No evidence was available for this outcome

Adverse skin reaction

• Two randomised controlled trials (N=2079) provided very low quality evidence to show that there was no clinically important difference in adverse skin reactions between those who received chlorhexidine-based alcohol skin preparation or iodophor-based aqueous/alcohol skin preparation.

Endometritis

• Two randomised controlled trials (N=2079) provided very low quality evidence to show that there was no clinically important difference in the occurrence of endometritis between those who received chlorhexidine-based alcohol skin preparation or iodophor-based aqueous/alcohol skin preparation.

Women’s experience

• No evidence was available for this outcome

Readmission into hospital

• Two randomised controlled trials (N=2079) provided low quality evidence to show that there was no clinically important difference in readmission into hospital between those who received chlorhexidine-based alcohol skin preparation or iodophor-based aqueous/alcohol skin preparation.

Sepsis

• No evidence was available for this outcome

Wound infection

• Seven randomised controlled trials (N=2639) provided very low quality evidence to show that there was no clinically important difference in the number of wound infections between those who received iodophor-based aqueous vaginal preparation or no vaginal/saline vaginal preparation.

Need for antibiotics

• No evidence was available for this outcome

Adverse skin events from techniques

• No evidence was available for this outcome

Endometritis

• Eight randomised controlled trials (N=3069) provided low quality evidence to show a clinically important decrease in the occurrence of endometritis for those who received iodophor-based aqueous vaginal preparation compared to those who received no vaginal/saline vaginal preparation.

Women’s experience

• No evidence was available for this outcome

Readmission into hospital

• No evidence was available for this outcome

Sepsis

• No evidence was available for this outcome

Wound infection

• One randomised controlled trial (N=200) provided very low quality evidence to show that there was no clinically important difference in wound infections between those who received chlorhexidine-based aqueous vaginal preparation or no vaginal cleansing/sterile water.

Need for antibiotics

• No evidence was available for this outcome

Adverse skin events from techniques

• No evidence was available for this outcome

Endometritis

• Two randomised controlled trials (N=214) provided moderate quality evidence to show a clinically important decrease in the number of episodes of endometritis for those who received chlorhexidine-based aqueous vaginal preparation compared to those who received no vaginal cleansing/sterile water.

Women’s experience

• No evidence was available for this outcome

Readmission into hospital

• No evidence was available for this outcome

Sepsis

• No evidence was available for this outcome

Wound infection

• Two randomised controlled trials (N=626) provided very low quality evidence to show that there was no clinically important difference in wound infections between those who received saline intra-abdominal irrigation or no irrigation.

Need for antibiotics

• No evidence was available for this outcome

Adverse skin events

• No evidence was available for this outcome

Endometritis

• Three randomised controlled trials (N=862) provided very low quality evidence to show that there was no clinically important difference in the occurrence of endometritis between those who received saline intra-abdominal irrigation or no irrigation.

Women’s experience

• No evidence was available for this outcome

Readmission into hospital

• No evidence was available for this outcome

The outcomes that matter most

The aim of this review was to identify which interventions reduced infectious morbidity in women undergoing caesarean birth. The committee therefore designated 3 critical outcomes: sepsis, wound infection/surgical site infection and need for antibiotics. These outcomes were selected as the most direct indicators for the efficacy and safety of the different interventions considered to reduce infectious morbidity.

The committee identified 4 further outcomes as important: endometritis, readmission into hospital, adverse skin events from techniques or interventions, and women’s experience. These outcomes were important because endometritis may occur after caesarean birth, readmission may indicate the presence of a wound-related problem, and some of the skin preparations and wound dressings may lead to adverse skin events so including this allowed the benefits and harms of the interventions to be balanced. As post-operative wound problems can have a detrimental impact on quality of life, it was also thought important to include women’s experience.

The quality of the evidence

Twenty-seven RCTs (18 of which were incorporated from 3 previously published systematic reviews) were included in this review. The quality of the evidence ranged from very low to moderate as assessed by GRADE.

The main reason for downgrading the evidence was the risk of bias due to studies not reporting how randomisation was performed or concealed, or because women, investigators and assessors were aware of treatment allocation. Other reasons for downgrading the quality of the evidence included sponsorship bias, where studies were funded by the manufacturers of the intervention under investigation, or indirectness (as some studies were conducted in low or middle income countries). Additionally, studies were also downgraded because of imprecision, as the trials had few women included, and therefore the confidence intervals around the estimate for each of the outcomes were wide.

The analysis comparing efficacy of NPWT in different BMI categories was a post-hoc subgrouping of an RCT. As such there is an additional risk of bias as these subgroups did not appear to be pre-specified or stratification that occurred prior to randomisation. However, the thresholds chosen (BMI 30–34.9 and 35 kg/m² or above) were reasonable and therefore the likelihood they were selected to emphasise a certain outcome is limited.

Benefits and harms

Although the use of prophylactic antibiotics is standard practice for women undergoing caesarean birth, there is still a risk of infection during any surgical procedure. Infections complicate recovery after surgery, may require a protracted hospital stay or intensive monitoring, and can have an important, detrimental effect on the woman’s quality of life and emotional state. The committee’s priority with these recommendations was to minimise maternal morbidity through the use of specific interventions.

The committee made the recommendations about choice of skin and vaginal preparation based on the evidence in this report, which suggested that these interventions reduce the risk of surgical site infections and endometritis, respectively.

Skin preparation for the abdomen is standard practice for a caesarean birth and the evidence indicated that the use of alcohol-based chlorhexidine skin preparation of the abdomen offered an important reduction in wound/surgical site infection compared to iodine skin preparations. The committee noted that this evidence, specific to women undergoing caesarean birth, is also in keeping with the recommendations for the general surgical population, contained in the NICE guideline on the prevention and treatment of surgical site infections. However, the committee noted that there was no difference in the rates of adverse events, endometritis or readmission between alcohol-based chlorhexidine preparations and iodine preparations, and so suggested that iodine preparations could be used as an alternative if alcohol-based chlorhexidine skin preparations were not available. This hierarchy is also in line with the NICE guideline on the prevention and treatment of surgical site infections.

The evidence showed a clinically important reduction in the occurrence of endometritis when antiseptic vaginal preparation (cleansing solution) was used, as compared to no vaginal preparation, or the use of saline only. Aqueous iodine vaginal solutions were shown to result in a clinically important reduction in endometritis, as compared to no preparation/saline preparation. On subgroup analysis according to membrane status, this difference was found to be most marked for women with ruptured membranes. The data regarding aqueous chlorhexidine vaginal preparation were more limited (2 studies), but also demonstrated a clinically important reduction in endometritis with the use of this solution. Therefore the committee decided that it would be appropriate to recommend aqueous iodine solution but to state that aqueous chlorhexidine vaginal preparation could be used as an alternative solution if the woman has allergies to iodine or if an iodine preparation is not available. The evidence for aqueous chlorhexidine vaginal preparation was not specific for women with ruptured membranes.

The evidence suggested that negative pressure wound therapy (NPWT) is likely to be effective in reducing wound infections or surgical site infections in women with body mass index (BMI) of 30 kg/m² or more, although the outcome is on the cusp of statistical significance.The committee discussed the evidence relevant for this intervention and noted that the studies were not robust enough to make a strong recommendation in all women with a BMI of 30 kg/m² or above. The main issues that the committee noted were that 2 different brands of NPWT were used across the studies and, as a result, the negative pressure that women received varied substantially. Five of the included studies (Gunatilake 2017, Hussamy 2019, Ruhstaller 2017, Tuuli 2020, Wihbey 2018) used the PREVENA negative pressure wound therapy device, applying a negative pressure of 125 mmHg, whereas 2 of the included studies in this comparison (Chaboyer 2014, Hyldig 2018) used the PICO negative pressure wound therapy device, applying a negative pressure of 80 mmHg. Furthermore, some of these studies were funded by the manufacturer of the negative pressure wound therapy device, which introduced a potential risk of bias. The experience of the committee was that, in current practice, NPWT was more commonly used for women with a BMI of 40 kg/m² or more, but the inclusion criteria for the studies reviewed was often lower than this. In a health economic analysis of one of the larger trials (Hyldig 2018), the trial authors reported their results separately for the group of women with a BMI 30–34.9 kg/m² and those with a BMI of 35 kg/m² or greater. The direction and point estimate of the effect was similar between the two groups. However, the relative effect was not statistically significant in the BMI 30–34.9 kg/m² group and the absolute effect was smaller. The results of the economic analysis differed between these groups (see below). There was some inconsistent evidence on adverse skin events occurring with NPWT. Overall there appeared to be no clinically important difference in adverse skin events between NPWT and standard dressing, however in 2 of the larger studies there were far more adverse skin events in the NPWT arm. The committee noted it was difficult to determine the severity of these events and also queried whether the inconsistent results could be due to varying monitoring strategies or inclusion criteria in terms of allergies. Finally the committee also noted the NICE medical technologies guidance (MTG43) about PICO negative pressure wound dressings for closed surgical incisions, which recommended their use for people at high risk of wound infections. Taking all of this into account, the committee agreed that there was sufficient evidence to make a weak recommendation for the use of NPWT in women with a BMI of 35 kg/m² and above.

Some limited evidence suggested that there were no clinically important differences in early (6 hours) as compared to standard (24 hours) removal of wound dressings, and that women were more satisfied when the dressing was removed earlier. This was consistent with the committee’s experience, and the committee also noted that women included in this study were being treated in an inpatient setting, and their surgical wounds were examined prior to discharge, which would be standard care in the UK. The committee therefore considered that the methods of the study were robust. The previous guideline had recommended that dressings were removed after 24 hours so the committee amended this recommendation to state that dressings could be removed between 6 and 24 hours after the CB. The committee also made a new recommendation to advise women that the evidence showed no differences in the risk of wound infection when the dressing was removed 6 hours or 24 hours postoperatively.

There was very limited evidence on the use of different types of postoperative dressings. A single study was identified which considered two specific types of dressing. The committee acknowledged that there are many different types available, but could not recommend one dressing over another as there was not enough evidence to support the decision. However, as women may ask about different dressings, the committee made a recommendation to clarify that there was evidence to demonstrate that one type of wound dressing was better at reducing wound infections that another.

There was some evidence comparing saline intra-abdominal irrigation with no irrigation which found no difference for wound infection or endometritis, and the committee decided that it was not necessary to make any recommendations relating to this intervention.

Due to the paucity of evidence in the use of hair removal, incise drapes and diathermy, the committee were unable to make specific recommendations regarding these interventions. Instead, they noted the relevant recommendations in the NICE guideline on surgical site infections: prevention and treatment. These apply to the general population undergoing surgery, rather than specifically to women having a caesarean birth, but were in line with the committee’s experience.

Cost effectiveness and resource use

The committee discussed the three relevant studies that considered the cost-effectiveness of NPWT in obese women (BMI ≥ 30 kg/m²) having a caesarean birth.

The results of Heard 2017 and Tuffaha showed NPWT to be more effective and more costly than standard care. In both studies, the ICER result was interpreted as showing that NPWT is cost-effective (based on an Australian cost-effectiveness threshold). However, there was some uncertainty around the result in both models (largely as a result of uncertainty in the clinical evidence base). The committee also noted that these 2 studies are Australian and are therefore of limited applicability to the UK health care setting.

Hyldig 2019 found NPWT to be dominant when compared to standard dressing but neither the cost saving or QALY benefit were found to be statistically significant. Nevertheless, probabilistic sensitivity analysis suggested there was a 65% probability that NPWT was cost saving. In addition, the committee noted that any cost savings appeared to be driven by the sub-group of women with BMI ≥ 35 kg/m².

The results of an economic study conducted as part of a recent NICE medical technology guidance on NPWT using PICO dressings (MTG43) were also discussed by the committee. The report included a cost analysis submitted by the manufacturer which was subsequently revised by the external assessment centre (EAC). The revised EAC cost analysis showed that, in comparison to standard dressings, PICO dressings resulted in modest cost savings when considering all surgery types. However, this overall result was driven by the large cost savings seen in highly invasive surgery (such as colorectal cancer) and PICO dressings were unlikely to be cost saving when used for surgeries undertaken on healthier patients such as caesarean birth and orthopaedic surgery.

On the basis of the economic evidence, the committee considered that a weak recommendation to consider NPWT was justified in women with a BMI of 35 kg/m² or above. An original economic analysis undertaken for this guideline suggested that although unlikely on the balance of probabilities, NPWT might be cost saving in this population due to a reduced incidence of surgical site infections when compared to standard dressings. The committee also thought that this was reflective of NHS practice where NPWT following caesarean birth would normally be reserved for this population. The committee also considered that this analysis finding was consistent with the MTG43 view that cost savings were more likely in less healthy patients. The committee agreed that no recommendation to consider NPWT in women with a BMI ≥ 30 kg/m² to BMI < 35 kg/m² was warranted from the economic evidence presented.

The committee identified that considering the use of NPWT in women with a BMI of 35 kg/m² or above having a caesarean birth, will be a change of practice for many units, who currently do not use it all at or who may use it at higher BMI thresholds, and may have resource implications, particularly in areas where a higher proportion of pregnant women will meet this criterion.

i. Cost-minimisation analysis

A probabilistic sensitivity analysis (PSA) with 10,000 simulations was undertaken for each sub-group (BMI ≥ 30 kg/m² to BMI < 35 kg/m²; BMI ≥ 35 kg/m²). The baseline risk was sampled using a Beta distribution and relative risk was sampled using a log-normal distribution. For women with a BMI ≥ 30 kg/m² to BMI < 35 kg/m² NPWT led to a mean net increase in costs of £77 when compared to standard dressing. The PSA suggested that there was only a 2.2% chance that NPWT was cost saving relative to standard dressing. In the sub-group of women with a BMI ≥ 35 kg/m² the ad-hoc analysis suggested that NPWT resulted in a £32 increase in mean net costs and had a 28.2% probability of being cheaper than standard dressing. The estimated probability distribution for the increase in costs with NPWT relative to standard dressing for each of the sub-groups is given in Figure 16 and Figure 17 respectively.

Figure 16Probability distribution for net increase in costs with NPWT relative to standard dressing in women with a BMI ≥ 30 kg/m² to BMI < 35 kg/m²

Figure 17Probability distribution for net increase in costs with NPWT relative to standard dressing in women with a BMI ≥ 35 kg/m²

ii. Cost-utility analysis

A PSA was undertaken for each of the sub-groups (BMI ≥ 30 kg/m² to BMI < 35 kg/m²; BMI ≥ 35 kg/m²) and the results are summarised in Table 15 and the cost-effectiveness analysis curves in Figure 18 and Figure 19.

Figure 18Cost-effectiveness acceptability curve for NPWT compared to standard dressing in women with BMI ≥ 30 kg/m² to BMI < 35 kg/m²

Figure 19Cost-effectiveness acceptability curve for NPWT compared to standard dressing in women with BMI ≥ 35 kg/m²

The committee were aware that that a NICE medical technology guidance (MTG43) considered Hyldig 2019 a weak publication, based on the method for eliciting QALYs and concerns around missing data for costs in the base case analysis. However, these limitations were not relevant to the findings of the ad-hoc analysis undertaken.