Advertisement

Factors associated with infection after operative vaginal birth—a secondary analysis of a randomized controlled trial of prophylactic antibiotics for the prevention of infection following operative vaginal birth

Open AccessPublished:August 23, 2022DOI:https://doi.org/10.1016/j.ajog.2022.08.037

      Background

      A recent randomized controlled trial of prophylactic antibiotics for the prevention of infection following operative vaginal birth showed that women allocated prophylactic intravenous amoxicillin and clavulanic acid had a significantly lower risk of developing confirmed or suspected infection within 6 weeks after operative vaginal birth (risk ratio [RR], 0.58; 95% confidence interval [CI], 0.49–0.69; P < .001). Some international and national guidelines have subsequently been updated to include prophylactic antibiotics after operative vaginal birth. However, the generalizability of the trial results may be limited in settings where the episiotomy rate is lower (89% of women in the trial had an episiotomy). In addition, there was a high burden of infection in the prophylactic antibiotics group despite the administration of prophylactic antibiotics. It is essential to identify modifiable risk factors for infection after operative vaginal birth, including the timing of antibiotic administration.

      Objective

      This study aimed to evaluate if the effectiveness of the prophylactic antibiotic in reducing confirmed or suspected infection was independent of perineal trauma, identify risk factors for infection after operative vaginal birth, and investigate variation in efficacy with the timing of antibiotic administration.

      Study Design

      This study was a secondary analysis of 3225 women with primary outcome data from the prophylactic antibiotics for the prevention of infection following operative vaginal birth randomized controlled trial. Women were divided into subgroups according to the perineal trauma experienced (episiotomy and/or perineal tear). The consistency of the prophylactic antibiotics in preventing infection across the subgroups was assessed using log-binomial regression and the likelihood ratio test. Multivariable log-binomial regression was used to investigate factors associated with infection. The multivariable risk factor model was subsequently fitted to the group of women who received amoxicillin and clavulanic acid to investigate the timing of antibiotic administration.

      Results

      Of the 3225 women included in the secondary analysis, 2144 (66.5%) had an episiotomy alone, 726 (22.5%) had an episiotomy and a tear, 277 (8.6%) had a tear alone, and 78 (2.4%) had neither episiotomy nor tear. Among women who experienced perineal trauma, amoxicillin and clavulanic acid administration was protective against infection in all subgroups compared with placebo with no significant interaction between subgroup and trial allocation (P=.17). Moreover, 2925 women were included in the multivariable risk factor analysis. The following were associated with adjusted risk ratios of infection: episiotomy, 2.94 (95% confidence interval, 1.62–5.31); forceps, 1.37 (95% confidence interval, 1.12–1.69) compared to vacuum extraction; primiparity, 1.34 (95% confidence interval, 1.05–1.70); amoxicillin and clavulanic acid administration, 0.60 (95% confidence interval, 0.51–0.72); body mass index of 25.0 to 29.9 kg/m2, 1.21 (95% confidence interval, 1.00–1.47), and body mass index of ≥30 kg/m2, 1.22 (95% confidence interval, 0.98–1.52) compared to body mass index of <25 kg/m2. Each 15-minute increment between birth and antibiotic administration was associated with a 3% higher risk of infection (adjusted risk ratio, 1.03; 95% confidence interval, 1.01–1.06).

      Conclusion

      Timely prophylactic antibiotics should be administered to all women after operative vaginal birth, irrespective of the type of perineal trauma. The use of episiotomy, forceps birth, primiparity, and overweight were associated with an increased risk of confirmed or suspected infection after operative vaginal birth.

      Key words

      Introduction

      Sepsis remains a major cause of maternal deaths globally, accounting for approximately 5% and 11% of deaths in high-income and low-income countries, respectively.
      • Say L.
      • Chou D.
      • Gemmill A.
      • et al.
      Global causes of maternal death: a WHO systematic analysis.
      The World Health Organization (WHO) Global Maternal Sepsis Study reported that for every woman with a severe maternal outcome (death or near miss) attributed to infection, a further 7 pregnant or recently pregnant women were hospitalized with an infection.
      WHO Global Maternal Sepsis Study (GLOSS) Research Group
      Frequency and management of maternal infection in health facilities in 52 countries (GLOSS): a 1-week inception cohort study.
      An increased risk of sepsis in association with cesarean birth has been recognized for many years and, more recently, for operative vaginal birth (OVB) (forceps or vacuum extraction).
      • Acosta C.D.
      • Bhattacharya S.
      • Tuffnell D.
      • Kurinczuk J.J.
      • Knight M.
      Maternal sepsis: a Scottish population-based case-control study.
      ,
      • Acosta C.D.
      • Kurinczuk J.J.
      • Lucas D.N.
      • et al.
      Severe maternal sepsis in the UK, 2011-2012: a national case-control study.
      Based on substantial randomized controlled trial evidence of effectiveness, including data from a recent large trial of antibiotic prophylaxis after OVB,
      • Knight M.
      • Chiocchia V.
      • Partlett C.
      • et al.
      Prophylactic antibiotics in the prevention of infection after operative vaginal delivery (ANODE): a multicentre randomised controlled trial.
      prophylactic antibiotics for both procedures are now indicated and have been incorporated into some national and international guidelines.
      • Murphy D.J.
      • Strachan B.K.
      • Bahl R.
      Royal College of Obstetricians and Gynaecologists
      Assisted vaginal birth: Green-top Guideline No. 26.
      ,
      Although the use of OVB has decreased in many countries, it remains an important obstetrical practice,
      Operative vaginal birth: ACOG Practice Bulletin, Number 219.
      particularly in settings where access to obstetrical care may be limited.
      • Bailey P.E.
      • van Roosmalen J.
      • Mola G.
      • Evans C.
      • de Bernis L.
      • Dao B.
      Assisted vaginal delivery in low and middle income countries: an overview.

      Why was this study conducted?

      The prophylactic antibiotics for the prevention of infection following operative delivery (ANODE) randomized controlled trial demonstrated the benefit of prophylactic antibiotics after operative vaginal birth (OVB). However, there was a high burden of infection in the ANODE trial despite prophylactic antibiotics. It was unclear whether the protective effect of prophylactic antibiotics was limited to women who underwent episiotomy.

      Key findings

      Prophylactic antibiotics seemed to protect against infection in all women who have perineal trauma after OVB. The risk factors for infection included the use of episiotomy, forceps birth, primiparity, overweight, and timing of antibiotic administration.

      What does this add to what is known?

      Prophylactic antibiotics should be given to all women who have OVBs as soon as possible after they have given birth. This study has identified additional risk factors for infection after OVB, which have the potential to be modified.
      The prophylactic antibiotics for the prevention of infection following operative delivery (ANODE) trial (trial number ISRCTN11166984) reported that women allocated amoxicillin and clavulanic acid had a statistically significantly lower rate of confirmed or suspected infection (180 of 1619 [11%]) than women allocated placebo (306 of 1606 [19%]) (risk ratio [RR], 0.58; 95% confidence interval [CI], 0.49–0.69; P<.001).
      • Knight M.
      • Chiocchia V.
      • Partlett C.
      • et al.
      Prophylactic antibiotics in the prevention of infection after operative vaginal delivery (ANODE): a multicentre randomised controlled trial.
      Nevertheless, the 11% infection rate among women receiving antibiotics indicates that identifying and tackling other modifiable risk factors remains vital in reducing the burden of maternal infection in women who have OVB. Almost 90% of women in the ANODE trial had an episiotomy, which may limit the generalizability to settings where the episiotomy rate is lower, and antibiotics were administered up to 6 hours after women gave birth. Previous studies have found that the use of episiotomy was significantly associated with infection up to 8 weeks after birth for all birth modalities.
      • Axelsson D.
      • Blomberg M.
      Maternal obesity, obstetric interventions and post-partum anaemia increase the risk of post-partum sepsis: a population-based cohort study based on Swedish medical health registers.
      • Axelsson D.
      • Brynhildsen J.
      • Blomberg M.
      Postpartum infection in relation to maternal characteristics, obstetric interventions and complications.
      • Gommesen D.
      • Nohr E.A.
      • Drue H.C.
      • Qvist N.
      • Rasch V.
      Obstetric perineal tears: risk factors, wound infection and dehiscence: a prospective cohort study.
      • Macleod M.
      • Strachan B.
      • Bahl R.
      • et al.
      A prospective cohort study of maternal and neonatal morbidity in relation to use of episiotomy at operative vaginal delivery.
      On this basis, the administration of prophylactic antibiotics is not recommended in a recently updated American College of Obstetricians and Gynecologists Practice Bulletin.
      Operative vaginal birth: ACOG Practice Bulletin, Number 219.
      Therefore, determining whether the benefit of amoxicillin and clavulanic acid is independent of perineal trauma status (perineal tear and/or episiotomy) is essential. The evaluation of the timing of prophylactic antibiotic administration is needed to direct guidelines that currently do not have a specified administration time.
      • Murphy D.J.
      • Strachan B.K.
      • Bahl R.
      Royal College of Obstetricians and Gynaecologists
      Assisted vaginal birth: Green-top Guideline No. 26.
      The primary objectives of this secondary (posthoc) analysis of the ANODE trial were to (1) evaluate whether the efficacy of prophylactic amoxicillin and clavulanic acid in reducing confirmed or suspected infection was independent of perineal tear and episiotomy, (2) identify risk factors for confirmed or suspected infection among women who have OVB and determine the population attributable fraction, and (3) investigate the optimum timing of prophylactic amoxicillin and clavulanic acid administration to minimize the risk of confirmed or suspected infection.

      Materials and Methods

      Study population

      Between March 13, 2016, and June 13, 2018, 3427 women who underwent a forceps or vacuum assisted birth at ≥36 weeks of gestation were randomly assigned (1:1) to receive a single intravenous dose of prophylactic amoxicillin and clavulanic acid or placebo as soon as possible after birth and no more than 6 hours later. This secondary analysis included all women in the ANODE trial cohort except for those with missing primary outcome data (n=195) and those who withdrew consent (n=7).

      Exposure assessment

      Maternal obstetrical and demographic data were collected from clinical records at trial entry by a research midwife and entered into electronic case report forms in an OpenClinica database. The height and weight measurements used for this analysis were those first recorded for the current pregnancy, which largely corresponded to the first contact with obstetrical services. The risk factors (for infection) included were based on previous literature,
      • Acosta C.D.
      • Bhattacharya S.
      • Tuffnell D.
      • Kurinczuk J.J.
      • Knight M.
      Maternal sepsis: a Scottish population-based case-control study.
      ,
      • Acosta C.D.
      • Kurinczuk J.J.
      • Lucas D.N.
      • et al.
      Severe maternal sepsis in the UK, 2011-2012: a national case-control study.
      ,
      • Axelsson D.
      • Blomberg M.
      Maternal obesity, obstetric interventions and post-partum anaemia increase the risk of post-partum sepsis: a population-based cohort study based on Swedish medical health registers.
      ,
      • Axelsson D.
      • Brynhildsen J.
      • Blomberg M.
      Postpartum infection in relation to maternal characteristics, obstetric interventions and complications.
      ,
      • Mohamed-Ahmed O.
      • Hinshaw K.
      • Knight M.
      Operative vaginal delivery and post-partum infection.
      and their availability in the dataset (Appendix, Supplemental Table 1).

      Outcome assessment

      The primary outcome was confirmed or suspected maternal infection within 6 weeks after birth, defined by either a new prescription of antibiotics for presumed perineal wound-related infection, endometritis or uterine infection, urinary tract infection with systemic features (pyelonephritis or sepsis), or other systemic infections (clinical sepsis), confirmed systemic infection on culture, or endometritis as defined by the US Centers for Disease Control and Prevention.

      Centers for Disease Control and Prevention (US). CDC/NHSN surveillance definitions for specific types of infections. 2013. Available at: http://www.cdc.gov/nhsn/pdfs/pscmanual/17pscnosinfdef_current.pdf. Accessed April 17, 2018.

      The primary outcome was ascertained from medical records at hospital discharge and by a telephone interview at 6 weeks after birth, following which women were sent a questionnaire for collection of data on secondary outcomes.
      • Knight M.
      • Chiocchia V.
      • Partlett C.
      • et al.
      Prophylactic antibiotics in the prevention of infection after operative vaginal delivery (ANODE): a multicentre randomised controlled trial.

      Statistical analysis

      Subgroup analysis of infection by perineal trauma

      Women were divided into subgroups according to the perineal trauma experienced during the OVB: solely a perineal tear, solely an episiotomy, both a perineal tear and an episiotomy, and neither perineal tear nor episiotomy. Perineal tears included both first- and second-degree tears.
      The consistency of the effect of amoxicillin and clavulanic acid in preventing confirmed or suspected infection vs placebo was assessed across perineal trauma subgroups using log-binomial regression. The RRs and 95% CIs have been presented for each subgroup. Likelihood ratio testing was used to assess statistical interaction.

      Risk factors for infection: both trial arms

      Women with missing baseline data were excluded from the risk factor analysis. Women were categorized by the most invasive mode of OVB that was attempted, with forceps classed as more invasive than vacuum extraction.
      • Murphy D.J.
      • Strachan B.K.
      • Bahl R.
      Royal College of Obstetricians and Gynaecologists
      Assisted vaginal birth: Green-top Guideline No. 26.
      The reasons for OVB were classed as “time critical” and “nontime critical” based on clinical judgment, as aseptic techniques may have been compromised in “time-critical” situations.
      The characteristics of the study population were summarized by the presence or absence of confirmed or suspected infection, with numbers (and percentages) presented for binary and categorical variables with comparison using chi-squared tests of association, means (with standard deviations) for normally distributed continuous variables with comparison using the Student t test, and medians (with interquartile ranges) with comparison using the Wilcoxon rank-sum for nonnormally distributed continuous data.
      All factors that had a P value of <.2 in univariable analysis were included in the initial multivariable log-binomial regression model. A backward elimination approach was used to select variables for the final model to avoid overfitting and selection bias and to provide correct estimates of standard error.
      • Harrell F.E.
      Regression modelling strategies with applications to linear models, logistic regression, and survival analysis.
      Where there was significant preexisting evidence of association, an a priori decision was taken to retain specific variables in the model; these variables included mode of birth, trial allocation, and body mass index (BMI) category. Likelihood ratio testing was used to establish whether additional variables significantly contributed to model fit. Overall model fit was assessed using the Akaike Bayesian information criterion and Schwarz Bayesian information criterion. Collinearity between variables used in the multivariable analysis was tested using Spearman correlation with a value of 0.8 considered significant.
      Based on biological plausibility, statistical interaction was tested for episiotomy and mode of birth because of the more invasive nature of forceps extraction vs vacuum extraction.
      • Murphy D.J.
      • Strachan B.K.
      • Bahl R.
      Royal College of Obstetricians and Gynaecologists
      Assisted vaginal birth: Green-top Guideline No. 26.
      The interaction between the use of episiotomy and BMI was also tested, as BMI has been found to be associated with the use of episiotomy.
      • Yamasato K.
      • Kimata C.
      • Burlingame J.M.
      Associations Between maternal obesity and race, with obstetric anal sphincter injury: a retrospective cohort study.
      A sensitivity analysis was undertaken using multiple imputation for missing baseline characteristics for the multivariable risk factor model. Of note, 300 participants had some missing baseline data. Parity, mode of birth, and episiotomy were found to be significant predictors of “missingness” using univariable log-binomial regression, suggesting the data were likely to be missing at random. All baseline variables in the Table were included in the imputation model, using the method of predictive mean matching because of the nonnormal distribution of variables. Ten imputations were performed, and the previous multivariable model was fitted to this multiple imputation dataset.
      TableBaseline demographic, obstetrical, and birth characteristics of participants by development of confirmed or suspected infection
      Baseline characteristicDeveloped infectionDid not develop infectionP valueaRR (95% CI)
      (n=452)(n=2473)
      Mother’s age at randomization (y),
      Mean (standard deviation) with comparison by t test
      mean (SD)
      30.5 (5.1)30.4 (5.4).711.00 (0.99–1.02)
      Gestational age at randomization,
      Median (interquartile range) with comparison by Wilcoxon rank-sum test
      median (interquartile range)
      40.1 (39.3–41.0)40.3 (39.3–41.0).48
      Gestational age category (wk)
       <375 (10.0)45 (90.0)1
       37 to <3828 (16.5)142 (83.5)1.65 (0.67–4.04)
       38 to <3958 (17.9)266 (82.1)1.79 (0.75–4.25)
       39 to <40105 (16.9)518 (83.1)1.69 (0.72–3.94)
       40 to <41130 (14.0)801 (86.0)1.40 (0.60–3.26)
       41 to <42112 (15.2)625 (84.8)1.52 (0.65–3.55)
       ≥4214 (15.6)76 (84.4)1.56 (0.60–4.07)
      Maternal ethnicity.43
       White393 (15.7)2115 (84.3)1
       BAME59 (14.1)358 (85.9)0.90 (0.70–1.16)
      BMI, median (interquartile range)
      Median (interquartile range) with comparison by Wilcoxon rank-sum test
      25.1 (22.6–28.7)24.5 (21.9–28.2).019
      Maternal BMI category (kg/m2)
       <25.0221 (14.0)1360 (86.0)1
       25.0–29.9136 (17.0)665 (83.0)1.21 (1.00–1.48)
       ≥30.095 (17.5)448 (82.5)1.25 (1.00–1.56)
      Multiple pregnancy.9
       Single pregnancy450 (15.5)2461 (84.5)1
       Twin pregnancy2 (14.3)12 (85.7)0.92 (0.26–3.34)
      Parity<.001
       Multiparous68 (11.1)542 (88.9)1
       Primiparous384 (16.6)1931 (83.4)1.49 (1.17–1.90)
      Timing of membrane rupture before birth (h).41
       <24386 (15.2)2147 (84.8)1
       24 to <4853 (16.0)278 (84.0)1.05 (0.81–1.37)
       ≥4813 (21.3)48 (78.7)1.40 (0.86–2.28)
      Induction of labor.42
       Induced230 (16.0)1207 (84.0)1
       Not induced222 (14.9)1266 (85.1)1.07 (0.91–1.27)
      Mode of birth<.001
       Vacuum extraction106 (10.9)865 (89.1)1
       Forceps346 (17.7)1608 (82.3)1.62 (1.32–1.99)
      Multiple modes of operative vaginal birth.071
       No423 (15.2)2363 (84.8)1
       Yes29 (20.9)110 (79.1)1.37 (0.98–1.92)
      Reason for operative vaginal birth.052
       Nontime critical239 (16.8)1185 (83.2)1
       Time critical213 (14.2)1288 (85.8)0.85 (0.71–1.00)
      Episiotomy during birth<.001
       No episiotomy11 (4.5)231 (95.5)1
       Episiotomy441 (16.4)2242 (83.6)3.62 (2.02–6.48)
      Tear during birth.36
       No perineal tear301 (15.0)1701 (85.0)1
       Perineal tear151 (16.4)772 (83.6)1.09 (0.91–1.30)
       Perineal wound sutured452 (15.5)2473 (84.5)NANA
      Suturing location.35
       Operating theater168 (16.3)863 (83.7)1
       Labor ward284 (15.0)1610 (85.0)0.92 (0.77–1.10)
      Suture material.33
       Vicryl15 (14.9)86 (85.1)1
       Vicryl Rapide427 (15.7)2289 (84.3)1.06 (0.66–1.70)
       Other10 (9.3)98 (90.7)0.62 (0.29–1.32)
      Allocation at randomization<.001
       Placebo285 (19.4)1187 (80.6)1
       Active (amoxicillin and clavulanic acid)167 (11.5)1286 (88.5)0.59 (0.50–0.71)
       Received the trial intervention
      Women who did not receive the intervention were excluded from this analysis because of missing data on time of antibiotic administration.
      452 (15.5)2473 (84.5)NANA
      Data are presented as number (percentage), unless otherwise indicated. P values correspond to comparison by chi-squared test of association unless otherwise stated. The absolute numbers and percentage are shown. Univariable RRs of confirmed or suspected infection are also shown.
      BAME, Black, Asian and minority ethnic; BMI, body mass index; CI, confidence interval; NA, not applicable; RR, risk ratio.
      Humphreys. Factors associated with infection following operative vaginal birth. Am J Obstet Gynecol 2022.
      a Mean (standard deviation) with comparison by t test
      b Median (interquartile range) with comparison by Wilcoxon rank-sum test
      c Women who did not receive the intervention were excluded from this analysis because of missing data on time of antibiotic administration.

      Risk factors for infection and timing of administration: amoxicillin and clavulanic acid arm only

      To reflect the current recommended clinical practice of amoxicillin and clavulanic acid administration after OVB,
      • Murphy D.J.
      • Strachan B.K.
      • Bahl R.
      Royal College of Obstetricians and Gynaecologists
      Assisted vaginal birth: Green-top Guideline No. 26.
      ,
      subsequent analyses were restricted to the 1453 women who received amoxicillin and clavulanic acid. The final multivariable log-binomial risk factor model was fitted to this group of women, and the timing of antibiotic administration was added to the model, categorized as 15-minute intervals to aid clinical interpretation. To determine the population attributable fraction for each identified adjusted risk factor, the Miettinen formula was used.
      • Miettinen O.S.
      Proportion of disease caused or prevented by a given exposure, trait or intervention.
      All statistical tests used a significance level of 0.05, and analyses were completed using Stata statistical software (version 16; StataCorp, College Station, TX).

      Ethics committee approval

      The ANODE trial was approved by the Health Research Authority National Research Ethics Service Committee South Central – Hampshire B (study reference number 15/SC/0442). Further Research Ethics Committee approval was not required for this analysis of anonymized trial data.

      Results

      Subgroup analysis of infection by perineal trauma

      A total of 3225 women with primary outcome data from the original trial were eligible for the perineal trauma subgroup analysis (Figure 1). Of those women, 2144 (66.5%) had an episiotomy only, 726 (22.5%) had both an episiotomy and a tear, and 277 (8.6%) had a tear only. Only 78 women (2.4%) had neither an episiotomy nor a tear.
      Figure thumbnail gr1
      Figure 1Flow of participants
      Humphreys. Factors associated with infection following operative vaginal birth. Am J Obstet Gynecol 2022.
      The proportion of women who developed a confirmed or suspected infection increased with the extent of perineal trauma in both trial arms (Figure 2). There was no statistically significant difference in the incidence of infection in the amoxicillin and clavulanic acid arm compared with the placebo arm among women who had no perineal trauma (RR, 1.18; 95% CI, 0.11–12.49), noting the very limited power of this analysis. Among women who experienced perineal trauma, amoxicillin and clavulanic acid administration was protective against infection in all subgroups compared with placebo administration: tear only (RR, 0.20; 95% CI, 0.04–0.87), episiotomy only (RR, 0.66; 95% CI, 0.53–0.81), and episiotomy with tear (RR, 0.50; 95% CI, 0.37–0.69) (Figure 2). There was no significant statistical interaction between the perineal trauma subgroup and the trial allocation on the risk of developing infection (P=.17).
      Figure thumbnail gr2
      Figure 2Risk of confirmed or suspected infection by perineal trauma subgroup
      CI, confidence interval; RR, risk ratio.
      Humphreys. Factors associated with infection following operative vaginal birth. Am J Obstet Gynecol 2022.

      Risk factors for infection: both trial arms

      Following the exclusion of those with missing baseline data, 2925 women were included in the multivariable risk factor analysis (Figure 1), of which 452 (15.5%) developed confirmed or suspected infection. Infection occurred a median of 7 days postnatally (interquartile range [IQR], 4–11 days; data only available for 170 of 452 women). For women with systemic sepsis, infection occurred a median of 5 days postnatally (IQR, 3–8 days; data available for 15 of 15 women). The baseline demographic, obstetrical, and clinical characteristics of women who did and did not develop infections are presented in the Table. All women included in the study received either the placebo or amoxicillin and clavulanic acid intervention that they had been allocated to at randomization. Factors associated with developing infection in univariable analysis were maternal BMI (median, 25.1 vs 24.5 kg/m2; P=.019), primiparity (85.0% vs 78.1%; P<.001), forceps birth (76.5% vs 65.0%; P<.001), episiotomy (97.6% vs 90.7%; P<.001), receipt of amoxicillin and clavulanic acid (36.9% vs 52.0%; P<.001), multiple modes of OVB attempted (6.4% vs 4.4%; P=.071), and time-critical reason for OVB (47.1% vs 52.1%; P=.052).
      Neither time-critical reason for OVB nor attempting multiple modes of OVB contributed to model fit. The final model is shown in Figure 3, A. The use of episiotomy (aRR 2.94; 95% CI, 1.62–5.31) and the use of forceps (aRR, 1.37; 95% CI, 1.12–1.69) were significantly associated with infection. The administration of amoxicillin and clavulanic acid was protective (aRR, 0.60; 95% CI, 0.51–0.72). Compared with women with a BMI of <25 kg/m2, the risk of infection in women with a BMI of 25 of 29.9 kg/m2 was statistically significantly increased (aRR, 1.21; 95% CI, 1.00–1.47), but not in women with a BMI of ≥30 kg/m2 (aRR, 1.22; 95% CI, 0.98–1.52). No significant interactions were found between the mode of birth and episiotomy (P=.345) or episiotomy and BMI (P=.961).
      Figure thumbnail gr3
      Figure 3Factors associated with confirmed or suspected infection in both trial arms combined (aRRs) and in the amoxicillin and clavulanic acid trial arm only (aRRs)
      A, Both trial arms combined. B, Amoxicillin and clavulanic acid trial arm only.
      aRR, adjusted risk ratio; BMI, body mass index; CI, confidence interval; PAF, population attributable fraction.
      Humphreys. Factors associated with infection following operative vaginal birth. Am J Obstet Gynecol 2022.
      The imputed data minimally changed the final multivariable risk factor model with all aRRs for infection differing by under 5% (Supplemental Figure).

      Risk factors for infection and timing of administration: amoxicillin and clavulanic acid arm only

      The final multivariable risk factor model for infection restricted to women who received amoxicillin and clavulanic acid can be seen in Figure 3, B, where the aRRs were similar to the full cohort model. The population attributable fraction for infection associated with the use of episiotomy was 81.4%, 23.8% for primiparity and 22.1% for use of forceps.
      Women received amoxicillin and clavulanic acid a median of 192 (IQR, 135–270) minutes after birth. The timing was modeled as a continuous categorical variable, as a significant linear trend was found at univariable analysis (P=.043) with no evidence of departure from linearity. The addition of the timing of the intervention to the final multivariable risk factor model can be seen in Appendix, Supplemental Table 2. Each additional 15-minute increment between birth and antibiotic administration was associated with a 3% (95% CI, 1.01–1.06) higher risk of infection, when adjusting for BMI category, mode of birth, episiotomy, and parity. A woman receiving the amoxicillin and clavulanic acid 3 hours after birth had an almost 50% higher risk of infection (1.0312=1.43), and a woman receiving the amoxicillin and clavulanic acid 6 hours after birth was associated with a 2-fold higher risk (1.0324=2.03) than a woman receiving antibiotic within 15 minutes.

      Comment

      Principal findings

      The analysis of this large prospective cohort of women undergoing OVB suggested that prophylactic amoxicillin and clavulanic acid administration is protective against confirmed or suspected infection after perineal trauma whether or not an episiotomy is used. Nevertheless, episiotomy is confirmed as an important risk factor for developing infection within 6 weeks of OVB, conferring a 3-fold higher risk compared to women who do not have an episiotomy, with approximately 80% of cases on a population basis attributable to this factor. Other significant risk factors included the use of forceps vs vacuum extraction and primiparity, both associated with a 30% to 40% increased risk. Furthermore, there was tentative evidence that a BMI of ≥25 kg/m2 may result in an elevated risk. Importantly, findings from this study suggested that prophylactic amoxicillin and clavulanic acid should be administered as soon as possible after OVB and that the longer the delay, the greater reduction in the efficacy of the treatment.

      Results in the context of what is known

      Observational studies have reported that the use of episiotomy is associated with a 1.4- to 10-fold higher odds of infection.
      • Axelsson D.
      • Blomberg M.
      Maternal obesity, obstetric interventions and post-partum anaemia increase the risk of post-partum sepsis: a population-based cohort study based on Swedish medical health registers.
      • Axelsson D.
      • Brynhildsen J.
      • Blomberg M.
      Postpartum infection in relation to maternal characteristics, obstetric interventions and complications.
      • Gommesen D.
      • Nohr E.A.
      • Drue H.C.
      • Qvist N.
      • Rasch V.
      Obstetric perineal tears: risk factors, wound infection and dehiscence: a prospective cohort study.
      • Macleod M.
      • Strachan B.
      • Bahl R.
      • et al.
      A prospective cohort study of maternal and neonatal morbidity in relation to use of episiotomy at operative vaginal delivery.
      The results from the analysis presented here, showing a 3-fold increase in infection, most closely align with the results of a British cohort study of nulliparous women who had OVB, which found a 4-fold higher odds of perineal infection with episiotomy use; however, the study was limited to 10 days after delivery.
      • Macleod M.
      • Strachan B.
      • Bahl R.
      • et al.
      A prospective cohort study of maternal and neonatal morbidity in relation to use of episiotomy at operative vaginal delivery.
      Most observational studies assessing maternal infection have not categorized OVB by the instrument used
      • Acosta C.D.
      • Bhattacharya S.
      • Tuffnell D.
      • Kurinczuk J.J.
      • Knight M.
      Maternal sepsis: a Scottish population-based case-control study.
      ,
      • Acosta C.D.
      • Kurinczuk J.J.
      • Lucas D.N.
      • et al.
      Severe maternal sepsis in the UK, 2011-2012: a national case-control study.
      ,
      • Axelsson D.
      • Blomberg M.
      Maternal obesity, obstetric interventions and post-partum anaemia increase the risk of post-partum sepsis: a population-based cohort study based on Swedish medical health registers.
      • Axelsson D.
      • Brynhildsen J.
      • Blomberg M.
      Postpartum infection in relation to maternal characteristics, obstetric interventions and complications.
      • Gommesen D.
      • Nohr E.A.
      • Drue H.C.
      • Qvist N.
      • Rasch V.
      Obstetric perineal tears: risk factors, wound infection and dehiscence: a prospective cohort study.
      ,
      • Acosta C.D.
      • Knight M.
      • Lee H.C.
      • Kurinczuk J.J.
      • Gould J.B.
      • Lyndon A.
      The continuum of maternal sepsis severity: incidence and risk factors in a population-based cohort study.
      ,
      • Al-Ostad G.
      • Kezouh A.
      • Spence A.R.
      • Abenhaim H.A.
      Incidence and risk factors of sepsis mortality in labor, delivery and after birth: population-based study in the USA.
      or have used women undergoing cesarean delivery as the reference group.
      • Bailit J.L.
      • Grobman W.A.
      • Rice M.M.
      • et al.
      Evaluation of delivery options for second-stage events.
      ,
      • Halscott T.
      • Reddy U.
      • Landy H.
      • et al.
      722: Maternal and neonatal outcomes by attempted mode of operative delivery during the second stage of labor in term singleton gestations.
      A US retrospective cohort of more than 6800 OVBs reported an unadjusted RR of 2.7 for postpartum infection for the use of forceps relative to vacuum extraction,
      • Kabiru W.N.
      • Jamieson D.
      • Graves W.
      • Lindsay M.
      Trends in operative vaginal delivery rates and associated maternal complication rates in an inner-city hospital.
      whereas a review found that the absolute proportion of women who developed infection ranged from 50% to more than 4-fold higher after a forceps birth compared with vacuum extraction.
      • Mohamed-Ahmed O.
      • Hinshaw K.
      • Knight M.
      Operative vaginal delivery and post-partum infection.
      Evidence for the association of primiparity with infection is mixed
      • Acosta C.D.
      • Bhattacharya S.
      • Tuffnell D.
      • Kurinczuk J.J.
      • Knight M.
      Maternal sepsis: a Scottish population-based case-control study.
      ,
      • Acosta C.D.
      • Kurinczuk J.J.
      • Lucas D.N.
      • et al.
      Severe maternal sepsis in the UK, 2011-2012: a national case-control study.
      ,
      • Acosta C.D.
      • Knight M.
      • Lee H.C.
      • Kurinczuk J.J.
      • Gould J.B.
      • Lyndon A.
      The continuum of maternal sepsis severity: incidence and risk factors in a population-based cohort study.
      ,
      • Martínez-Galiano J.M.
      • Hernández-Martínez A.
      • Rodríguez-Almagro J.
      • Delgado-Rodríguez M.
      • Gómez-Salgado J.
      Relationship between parity and the problems that appear in the postpartum period.
      and may reflect a lack of adjustment for confounders in some studies.
      The association of postpartum infection with obesity has been consistently found by other observational studies where the odds of infection ranged from 29% higher for postpartum endometritis to more than 2-fold higher for sepsis.
      • Acosta C.D.
      • Bhattacharya S.
      • Tuffnell D.
      • Kurinczuk J.J.
      • Knight M.
      Maternal sepsis: a Scottish population-based case-control study.
      ,
      • Axelsson D.
      • Blomberg M.
      Maternal obesity, obstetric interventions and post-partum anaemia increase the risk of post-partum sepsis: a population-based cohort study based on Swedish medical health registers.
      ,
      • Axelsson D.
      • Brynhildsen J.
      • Blomberg M.
      Postpartum infection in relation to maternal characteristics, obstetric interventions and complications.

      Clinical implications

      Although no novel risk factor for infection has been identified by this study, the study demonstrates these associations with increased risk in this specific population of women undergoing OVB. Except for parity, all identified risk factors are potentially modifiable, and thus, these results have the potential to influence clinical practice and guidelines. Most importantly, the very clear association between earlier administration of amoxicillin and clavulanic acid and lower infection rates emphasizes the added benefit of administration as soon as possible after women have given birth.

      Research implications

      A Cochrane review on the role of prophylactic antibiotics in OVB has been updated to reflect the results of the ANODE trial but noted the need for further randomized controlled evidence in low-income settings,
      • Liabsuetrakul T.
      • Choobun T.
      • Peeyananjarassri K.
      • Islam Q.M.
      Antibiotic prophylaxis for operative vaginal delivery.
      where routine intravenous administration may not be possible. It is questionable whether such a trial would be considered ethical given the strong recommendation from the WHO concerning the use of prophylactic amoxicillin and clavulanic acid.
      Nevertheless, identifying and targeting modifiable risk factors in low- and middle-income countries are crucial as they disproportionately bear the burden of maternal mortality, accounting for 94% of global deaths.

      UNFPA, World Health Organization, UNICEF, World Bank Group, the United Nations Population Division. Trends in maternal mortality: 2000 to 2017: Estimates by WHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division. 2019. Available at: https://www.unfpa.org/featured-publication/trends-maternal-mortality-2000-2017. Accessed July 27, 2021.

      Strengths and limitations

      The ANODE trial initially highlighted the previously unknown burden of maternal infection after OVB, and this secondary analysis has identified key risk factors and their relative importance. Primary outcome data were missing for only 5.7% of women, and the large sample size allowed for reasonable statistical power to identify risk factors. Furthermore, the long duration of follow-up through the whole of the first 6 weeks after birth and the comprehensive definition of maternal infection indicate that the true incidence of infection was likely to have been captured. Many other observational studies have been limited by their lack of active follow-up, meaning cases of maternal infection may have been missed.
      • Mohamed-Ahmed O.
      • Hinshaw K.
      • Knight M.
      Operative vaginal delivery and post-partum infection.
      As this was a posthoc analysis, it cannot be ruled out that the study was statistically underpowered to detect variation in the association between subgroups. The analysis was also limited by the number of obstetrical, clinical, and demographic factors that were recorded for the trial. Therefore, there may be other significant predictors of infection, including past medical history, smoking, and socioeconomic status, that have been previously identified.
      • Acosta C.D.
      • Kurinczuk J.J.
      • Lucas D.N.
      • et al.
      Severe maternal sepsis in the UK, 2011-2012: a national case-control study.
      ,
      • Axelsson D.
      • Brynhildsen J.
      • Blomberg M.
      Postpartum infection in relation to maternal characteristics, obstetric interventions and complications.
      ,
      • Acosta C.D.
      • Knight M.
      • Lee H.C.
      • Kurinczuk J.J.
      • Gould J.B.
      • Lyndon A.
      The continuum of maternal sepsis severity: incidence and risk factors in a population-based cohort study.
      ,
      • Al-Ostad G.
      • Kezouh A.
      • Spence A.R.
      • Abenhaim H.A.
      Incidence and risk factors of sepsis mortality in labor, delivery and after birth: population-based study in the USA.
      It should also be noted that, for the observational analysis, there may be residual confounding, including confounding by indication, which may account for the observed results. As with many trials, the women recruited for the ANODE trial may not have been representative of the general population. Of note, 14% of women in the secondary analysis of the ANODE trial came from a Black and minority ethnic background. However, in 2016–2018, at least 20% of maternities in England were to women from a minority ethnic background.
      • Knight M.
      • Bunch K.
      • Tuffnell D.
      • et al.
      eds. Saving Lives, Improving Mothers’ Care - lessons learned to inform maternity care from the UK and Ireland Confidential Enquiries into Maternal Deaths and Morbidity 2016-18.

      Conclusions

      This study has found no evidence to suggest that prophylactic amoxicillin and clavulanic acid administration is less protective against confirmed or suspected infection after OVB with perineal trauma in the absence of episiotomy, which provides reassurance of the benefit of prophylactic antibiotic in settings where the episiotomy rate is lower. The findings suggested that the use of episiotomy, forceps birth, primiparity, and possibly obesity were associated with an increased risk of postpartum confirmed or suspected infection in women undergoing OVB. Importantly for clinical practice, the burden of infection may be further reduced by timely administration of the antibiotic to all women irrespective of the state of their perineum.

      Acknowledgments

      We would like to acknowledge the women who participated in the prophylactic antibiotics for the prevention of infection following operative delivery (ANODE) trial, the ANODE Collaborative Group, the site principal investigators, and the research staff.

      Supplementary Data

      Appendix. Supplementary web material

      Figure thumbnail fx1
      Supplemental FigureFactors associated with infection: sensitivity analysis using multiple imputation
      aRR, adjusted risk ratio; BMI, body mass index; CI, confidence interval.
      Humphreys. Factors associated with infection following operative vaginal birth. Am J Obstet Gynecol 2022.
      Supplemental Table 1The choice of demographic, clinical, and obstetrical variables that were investigated
      Demographic and clinical variablesObstetrical variables
      -Age-Gestational age at randomization
      -Ethnicity-Parity (primiparous vs multiparous)
      -BMI-If the pregnancy was a multiple pregnancy
      -Trial allocation-Rupture of membranes before birth

      -Timing of membrane rupture
      -Induction of labor
      -Multiple modes of operative vaginal birth attempted
      -Mode of birth
      -Episiotomy in current birth
      -Perineal tear in current birth

      -Perineal wound suturing
      -Suture material
      -Location of suturing
      -Reason for operative vaginal birth
      BMI, body mass index.
      Humphreys. Factors associated with infection following operative vaginal birth. Am J Obstet Gynecol 2022.
      Supplemental Table 2The final multivariable risk factor model for confirmed or suspected infection with timing of amoxicillin and clavulanic acid administration in women who received amoxicillin and clavulanic acid
      CharacteristicaRR (95% CI)
      Mode of birth
       Vacuum extraction1
       Forceps1.39 (0.99–1.95)
      Use of episiotomy
       No episiotomy1
       Episiotomy5.60 (1.39–22.52)
      Parity
       Multiparous1
       Primiparous1.39 (0.92–2.10)
      BMI category
       <25.0 kg/m21
       25.0–29.9 kg/m21.51 (1.10–2.08)
       ≥30.0 kg/m21.37 (0.94–2.01)
      Per 15-min increment between birth and amoxicillin and clavulanic acid administration1.03 (1.01–1.06)
      aRR, adjusted risk ratio; BMI, body mass index; CI, confidence interval.
      Humphreys. Factors associated with infection following operative vaginal birth. Am J Obstet Gynecol 2022.

      References

        • Say L.
        • Chou D.
        • Gemmill A.
        • et al.
        Global causes of maternal death: a WHO systematic analysis.
        Lancet Glob Health. 2014; 2: e323-e333
        • WHO Global Maternal Sepsis Study (GLOSS) Research Group
        Frequency and management of maternal infection in health facilities in 52 countries (GLOSS): a 1-week inception cohort study.
        Lancet Glob Health. 2020; 8: e661-e671
        • Acosta C.D.
        • Bhattacharya S.
        • Tuffnell D.
        • Kurinczuk J.J.
        • Knight M.
        Maternal sepsis: a Scottish population-based case-control study.
        BJOG. 2012; 119: 474-483
        • Acosta C.D.
        • Kurinczuk J.J.
        • Lucas D.N.
        • et al.
        Severe maternal sepsis in the UK, 2011-2012: a national case-control study.
        PLoS Med. 2014; 11e1001672
        • Knight M.
        • Chiocchia V.
        • Partlett C.
        • et al.
        Prophylactic antibiotics in the prevention of infection after operative vaginal delivery (ANODE): a multicentre randomised controlled trial.
        Lancet. 2019; 393: 2395-2403
        • Murphy D.J.
        • Strachan B.K.
        • Bahl R.
        • Royal College of Obstetricians and Gynaecologists
        Assisted vaginal birth: Green-top Guideline No. 26.
        BJOG. 2020; 127: e70-e112
      1. WHO recommendation on routine antibiotic prophylaxis for women undergoing operative vaginal birth. World Health Organization, Geneva, Switzerland2021
      2. Operative vaginal birth: ACOG Practice Bulletin, Number 219.
        Obstet Gynecol. 2020; 135: e149-e159
        • Bailey P.E.
        • van Roosmalen J.
        • Mola G.
        • Evans C.
        • de Bernis L.
        • Dao B.
        Assisted vaginal delivery in low and middle income countries: an overview.
        BJOG. 2017; 124: 1335-1344
        • Axelsson D.
        • Blomberg M.
        Maternal obesity, obstetric interventions and post-partum anaemia increase the risk of post-partum sepsis: a population-based cohort study based on Swedish medical health registers.
        Infect Dis (Lond). 2017; 49: 765-771
        • Axelsson D.
        • Brynhildsen J.
        • Blomberg M.
        Postpartum infection in relation to maternal characteristics, obstetric interventions and complications.
        J Perinat Med. 2018; 46: 271-278
        • Gommesen D.
        • Nohr E.A.
        • Drue H.C.
        • Qvist N.
        • Rasch V.
        Obstetric perineal tears: risk factors, wound infection and dehiscence: a prospective cohort study.
        Arch Gynecol Obstet. 2019; 300: 67-77
        • Macleod M.
        • Strachan B.
        • Bahl R.
        • et al.
        A prospective cohort study of maternal and neonatal morbidity in relation to use of episiotomy at operative vaginal delivery.
        BJOG. 2008; 115: 1688-1694
        • Mohamed-Ahmed O.
        • Hinshaw K.
        • Knight M.
        Operative vaginal delivery and post-partum infection.
        Best Pract Res Clin Obstet Gynaecol. 2019; 56: 93-106
      3. Centers for Disease Control and Prevention (US). CDC/NHSN surveillance definitions for specific types of infections. 2013. Available at: http://www.cdc.gov/nhsn/pdfs/pscmanual/17pscnosinfdef_current.pdf. Accessed April 17, 2018.

        • Harrell F.E.
        Regression modelling strategies with applications to linear models, logistic regression, and survival analysis.
        Springer, New York, NY2001
        • Yamasato K.
        • Kimata C.
        • Burlingame J.M.
        Associations Between maternal obesity and race, with obstetric anal sphincter injury: a retrospective cohort study.
        Hawaii J Med Public Health. 2019; 78: 8-12
        • Miettinen O.S.
        Proportion of disease caused or prevented by a given exposure, trait or intervention.
        Am J Epidemiol. 1974; 99: 325-332
        • Acosta C.D.
        • Knight M.
        • Lee H.C.
        • Kurinczuk J.J.
        • Gould J.B.
        • Lyndon A.
        The continuum of maternal sepsis severity: incidence and risk factors in a population-based cohort study.
        PLoS One. 2013; 8e67175
        • Al-Ostad G.
        • Kezouh A.
        • Spence A.R.
        • Abenhaim H.A.
        Incidence and risk factors of sepsis mortality in labor, delivery and after birth: population-based study in the USA.
        J Obstet Gynaecol Res. 2015; 41: 1201-1206
        • Bailit J.L.
        • Grobman W.A.
        • Rice M.M.
        • et al.
        Evaluation of delivery options for second-stage events.
        Am J Obstet Gynecol. 2016; 214: 638.e1-638.e10
        • Halscott T.
        • Reddy U.
        • Landy H.
        • et al.
        722: Maternal and neonatal outcomes by attempted mode of operative delivery during the second stage of labor in term singleton gestations.
        Am J Obstet Gynecol. 2015; 212: S351-S352
        • Kabiru W.N.
        • Jamieson D.
        • Graves W.
        • Lindsay M.
        Trends in operative vaginal delivery rates and associated maternal complication rates in an inner-city hospital.
        Am J Obstet Gynecol. 2001; 184: 1112-1114
        • Martínez-Galiano J.M.
        • Hernández-Martínez A.
        • Rodríguez-Almagro J.
        • Delgado-Rodríguez M.
        • Gómez-Salgado J.
        Relationship between parity and the problems that appear in the postpartum period.
        Sci Rep. 2019; 911763
        • Liabsuetrakul T.
        • Choobun T.
        • Peeyananjarassri K.
        • Islam Q.M.
        Antibiotic prophylaxis for operative vaginal delivery.
        Cochrane Database Syst Rev. 2020; 3: CD004455
      4. UNFPA, World Health Organization, UNICEF, World Bank Group, the United Nations Population Division. Trends in maternal mortality: 2000 to 2017: Estimates by WHO, UNICEF, UNFPA, World Bank Group and the United Nations Population Division. 2019. Available at: https://www.unfpa.org/featured-publication/trends-maternal-mortality-2000-2017. Accessed July 27, 2021.

        • Knight M.
        • Bunch K.
        • Tuffnell D.
        • et al.
        eds. Saving Lives, Improving Mothers’ Care - lessons learned to inform maternity care from the UK and Ireland Confidential Enquiries into Maternal Deaths and Morbidity 2016-18.
        National Perinatal Epidemiology Unit, University of Oxford, Oxford, United Kingdom2020