Cesarean delivery is the most common major abdominal surgery performed in the United States.1 Endometritis and wound infection are common and significant postoperative morbidities with respect to patient cost, prolonged hospital stay, use of parenteral antibiotics, and patient discomfort. The risk of postcesarean infectious morbidity is reported to range from 5–85%, with the highest rate among indigent patients.2,3 The most recognized risk factors for developing postcesarean endometritis involve pathways that introduce large quantities of bacteria into the uterine cavity. These include a large number of vaginal examinations in labor, prolonged duration of active labor, prolonged membrane rupture, and failure to use antimicrobial prophylaxis.4 Other reported risk factors include nulliparity, use of internal monitors in labor, adolescence, presence of intrapartum bacterial vaginosis, and the presence of a comorbid immunocompromised state such as diabetes mellitus or human immunodeficiency virus infection.4–6 The pathophysiology is theoretically an ascending polymicrobial infection of cervical and vaginal organisms into the uterus, with hematogenous spread through exposed edges of incised myometrium. The bacterial species implicated in postcesarean endometritis and wound infection include gram-negative bacilli, aerobic and anaerobic gram-positive cocci, and anaerobic bacilli associated with bacterial vaginosis.7–9
There is evidence in the literature to support the use of preoperative vaginal scrub with povidone-iodine before a hysterectomy to decrease the incidence of postoperative infectious morbidity.10 Vaginal preparation has been shown to decrease the quantitative load of vaginal microorganisms as well as to remove certain species of bacteria.11,12 There is limited information in the literature regarding the use of preoperative vaginal preparation with povidone-iodine as a prophylactic measure against postcesarean infectious morbidity.13 In this randomized controlled trial, we prospectively investigated the hypothesis that vaginal preparation with povidone-iodine scrub before a cesarean delivery will decrease the incidence of postoperative endometritis, wound infection, and overall postoperative febrile morbidity.
MATERIALS AND METHODS
Eligible patients were recruited at Chicago Lying-In Hospital from November 1997 through March 2000. All women who were to undergo nonemergent cesarean delivery were eligible for recruitment, excluding those with placenta previa or a diagnosis of chorioamnionitis. Eligible patients were approached for the study at the time of admission for an elective cesarean delivery or, in the case of laboring patients, at the time the decision for cesarean delivery was made. After informed consent was obtained, each patient was assigned to receive either a standard abdominal skin preparation with povidone-iodine solution or the standard abdominal preparation plus an additional 30-second vaginal scrub with povidone-iodine solution. The allocation sequence was determined by using a random digit table, and the assignments were placed in sequentially numbered, opaque, sealed envelopes. The envelopes were opened after the patient agreed to participate in the study. This study was approved by the Institutional Review Board of the University of Chicago Hospitals.
The power calculation used to determine the minimum sample size assumed α = 0.05 and 80% power, based on our institutional rate of postcesarean infectious morbidity of 20%. A sample size of 390 was chosen as large enough to detect a 50% decrease in the incidence of postcesarean infectious morbidity.
Vaginal preparation was performed at the time of the abdominal scrub with 2 prepackaged foam sponge sticks saturated with povidone-iodine solution (EZ Prep 220, Clinipad Corporation, Charlotte, NC). Each sponge stick was rotated 360° in the vagina such that the entire process lasted about 30 seconds. All patients received a single dose of parenteral antibiotic prophylaxis with either cefazolin or clindamycin at the time of umbilical cord clamping.
After the patients were discharged from the hospital and completed a 6-week puerperal period, their charts were reviewed by 1 physician (R.V.S.) for development of postoperative febrile morbidity, diagnosis of endometritis, or diagnosis of wound infection. Febrile morbidity was defined as any postoperative temperature greater than 38°C. Endometritis was defined by a temperature elevation greater than 38.4°C persisting beyond the first postoperative day, in association with uterine tenderness and foul lochia, in the absence of physical or laboratory evidence of other infection. The diagnosis of endometritis was given at the time of clinical evaluation and not assigned by the chart reviewer. Wound infection was a clinical diagnosis, evidenced by erythema or wound edge separation with purulent drainage. This last diagnosis also included such entities as wound dehiscence and necrotizing fasciitis. Cases of skin separation without evidence of cellulitis were not included, because this information was not consistently available in the hospital charts. Analysis of the number of patients with endometritis was as a subgroup of patients with febrile morbidity. Because febrile morbidity, endometritis, and wound infection were defined as the primary outcomes for this study, data were not collected with respect to other sources of postoperative fever such as urinary tract infection, pneumonia, atelectasis, or drug fever. Appropriate antibiotic therapy was initiated for patients with a clear diagnosis of infection, whether pelvic or nonpelvic in origin. At our institution, outpatients who are diagnosed with postoperative wound infections are admitted to the hospital for initial intravenous antibiotic therapy and wound care, therefore documentation of this complication is available in their hospital charts.
Other information abstracted from the chart review included patient demographics, parity, gestational age, indication for cesarean delivery, history of antenatal genitourinary infection, presence of Group B Streptococcus colonization, number of vaginal examinations, evidence of prolonged active labor (> 10 hours)14 and prolonged ruptured membranes (> 18 hours),15 presence of meconium-stained amniotic fluid, use of internal monitors in labor, and postoperative length of stay. Antenatal genitourinary infections included urinary tract infections, bacterial vaginosis, or documented infection with Neisseria gonorrhoeae or Chlamydia trachomatis. Additionally, hospital charts were reviewed for the presence of maternal comorbid conditions including severe anemia (hematocrit < 30%), diabetes, and morbid obesity (> 114 kg).
Measurement bias was minimized because the physicians who evaluated all patients with postoperative fever were unaware of any patient's participation in the study. This was possible because, at any given time, the postpartum service was a completely different team of physicians from that which staffed the labor and delivery unit, where patient enrollment took place. Reporting bias was minimal, because all hospital charts were reviewed without knowledge of patient assignment to either arm of the study.
Data were analyzed using the SAS System (SAS Institute Inc., Cary, NC). Due to the number of patients, proportions were compared using Fisher exact test or continuity-adjusted χ2. Logistic regression was used for the multivariate modeling of the data.
Four hundred patients were recruited into the study. Thirty-three subjects were excluded because their assignment envelopes that contained patient identification were lost, and no patient information was available. Of these unidentified subjects, 21 had been randomly assigned to receive a vaginal scrub, as indicated by the envelope assignment number. Hospital charts were unable to be located for 53 additional patients, 28 of whom had been assigned to the treatment group. Additionally, 6 subjects were excluded due to violation of inclusion criteria (5 had intrapartum chorioamnionitis, and 1 patient had a vaginal delivery). In this last group, 5 patients were assigned to receive a vaginal scrub. Ultimately, 92 subjects were excluded from the final analysis (Fig. 1). Primary analysis was intention-to-treat and involved all participants who were randomly assigned and for whom all data were available.
Complete data were available for 308 subjects. One hundred sixty-six subjects were assigned to receive a standard abdominal scrub, and 142 subjects were assigned to receive an additional vaginal preparation. Comparison of the control group to the treatment group scrub did not demonstrate a significant difference in patient demographics, pregnancy history, obstetric characteristics, or maternal comorbid conditions (Table 1).
Postcesarean endometritis occurred in 24 of 166 (14%) of control patients and in 10 of 142 (7%) patients who received a preoperative vaginal scrub (P < .05). Forty-seven of 166 (28%) control patients and 34 of 142 (24%) patients in the treatment group developed postoperative febrile morbidity. (P = .437) Wound infection was an infrequent complication that developed in 2 (1.2%) control patients and 1 (0.7%) patient in the treatment group. (P = .403) Postoperative length of stay was similar in the 2 groups. A postoperative stay that exceeded 4 days was seen in 14 (8.4%) control patients and 7(5.0%) of treatment patients (P = .224) (Table 2).
Multivariate analysis indicated several factors that affected the risk for developing postcesarean endometritis. The adjusted odds ratio (OR) for developing endometritis after a vaginal scrub was 0.44. (95% confidence interval [CI] 0.193–0.997). The risk of developing endometritis was significantly increased in association with severe anemia (adjusted OR 4.26, 95% CI 1.568–11.582), the use of internal monitors (adjusted OR 2.84, 95% CI 1.311–6.136), or a history of antenatal genitourinary infection (adjusted OR 2.89, 95% CI 1.265–6.595) (Table 3).
The concept of vaginal antisepsis is not new to the field of gynecology. Since the 1970s, it has been demonstrated that a povidone-iodine vaginal scrub before vaginal surgery or abdominal hysterectomy is associated with lower postoperative infectious morbidity.16 Osborne and Wright11 showed that a preoperative povidone-iodine vaginal scrub decreased the total number of bacterial species in the vagina by at least 98%. A preoperative vaginal scrub with povidone-iodine was shown to remove anaerobic gram-positive bacilli and dramatically decrease the quantities of gram-negative bacilli and aerobic and anaerobic gram-positive cocci, especially Enterococcus species.12
Risk factors for developing postcesarean endometritis have been recognized, and all involve an increased risk of exposure of the upper genital tract to lower tract bacteria. A large number of vaginal examinations in labor has the strongest association with postoperative infection, likely due to an increased size and speed of entry of a bacterial inoculum from the vagina into the uterus.1–15 Other frequently described obstetric risk factors for postcesarean infection include prolonged labor, prolonged membrane rupture, and the use of internal monitoring.2,4,7,8,14 Maternal characteristics that seem to contribute to the development of postcesarean endometritis include age and socioeconomic status, with the highest rate of infection developing in indigent young women.3 Additionally, the presence of bacterial vaginosis or other concurrent vaginal infection predisposes the patient to an ascending infection.6,7 Watts et al7 found a 6-fold increase of postcesarean endometritis when bacterial vaginosis was associated with delivery.
Strategies to decrease the incidence of postcesarean endometritis are currently being used, including conscientious surgical technique and the routine use of antibiotic prophylaxis. Despite these interventions, infectious morbidity after a cesarean delivery remains significant.4,5,7
The effectiveness of prophylactic parenteral antibiotics has been well demonstrated in the literature, with cumulative data suggesting a reduction in postcesarean endometritis of approximately 50% in a given institution.2 The eradication of infection is thus incomplete, with a resultant postcesarean endometritis rate of 10–20% despite the use of prophylactic antibiotics.5,7 Prophylaxis failure with increased doses of cefazolin has been reported, and may be due to a shift in vaginal flora with a dominance of organisms resistant to cefazolin, such as Enterococcus.4,5,8
Recent studies have investigated specific preoperative interventions to decrease the risk of postcesarean endometritis. One published study also investigated the use of preoperative vaginal preparation with povidone-iodine before cesarean delivery.13 Reid et al13 reported that vaginal preparation did not affect the incidence of postoperative fever, endometritis, or wound infection. However, that study did not address several potential risks for increased exposure to infection, specifically, a history of antenatal genitourinary infection, use of intrapartum internal monitors, severe anemia, or presence of diabetes mellitus or obesity. Also, that study did not indicate whether all participants received parenteral prophylactic antibiotic at the time of umbilical cord clamping.
Pitt et al17 reported that the use of preoperative intravaginal metronidazole gel reduced the incidence of postcesarean endometritis, presumably by reducing the local exposure of anaerobic bacteria during a cesarean delivery. This finding supports our hypothesis that, similar to antisepsis before a hysterectomy, postcesarean infection rates may be improved by reducing the vaginal bacterial load preoperatively.
A general interpretation of our results suggests that a preoperative vaginal scrub decreases the risk of postcesarean endometritis. This intervention, however, does not seem to reduce the overall risk of developing postoperative febrile morbidity. Within our model, severe anemia, the use of internal monitors, and a history of antenatal genitourinary infections were independently associated with an increased risk of postcesarean endometritis. Differences in reported postoperative endometritis rates could be attributed to the technique and materials used for the vaginal preparation itself. Perhaps subtle differences in contact time and distribution of povidone-iodine within the vagina, or the amount of antiseptic used for the preparation might affect infectious outcomes.
Several limitations have been identified in the assignment of this randomized study. The total number and demographic data of patients who underwent cesarean delivery during the study period is unknown, therefore it is also unknown whether we obtained a valid representation of our eligible patient pool. There was a large amount of data that was lost due to the unavailability of hospital charts. These data were balanced with respect to the number of subjects assigned to each arm of the trial. However, there was an additional large imbalanced loss of information for the subjects excluded for violation of recruitment criteria and for the unidentified subjects for whom no data are available. In these latter 2 groups of patients, more information was lost for the treatment arm of the trial. Finally, the power calculation was not met in the final analysis of available charts. These shortcomings introduce the possibility of bias, with a possible overestimation of the treatment effect of a preoperative vaginal scrub.18 With respect to the assessment of primary outcomes, we did not collect specific data about those patients, if any, who were diagnosed with endometritis during the first postoperative day, nor those who may have developed fundal tenderness and foul lochia but did not have temperatures that exceeded 38.4°C, and the effect of vaginal preparation within this subgroup is unknown. Additionally, an examination of nonuterine sources of fever may be useful in interpreting the overall incidence of febrile morbidity.
Within our study population, this trial demonstrates a benefit of a preoperative vaginal scrub just before cesarean delivery. The incidence of postcesarean endometritis was significantly decreased in those subjects who were scrubbed with both abdominal and vaginal povidone-iodine compared with those who received a standard abdominal scrub alone. Used in conjunction with prophylactic antibiotics, a vaginal povidone-iodine preparation may further decrease the number of bacteria species exposed to the endometrium at the time of uterine incision, especially Enterococcus species that are resistant to cefazolin prophylaxis. Further studies are required to confirm these findings before a change is practice is instituted.
1. Soper DE. Infections following cesarean section. Curr Opin Obstet Gynecol 1993;5:517–20.
2. Gonik B, Shannon RL, Shawar R, Costner M, Seibel M. Why patients fail antibiotic prophylaxis at cesarean delivery: histologic evidence for incipient infection. Obstet Gynecol 1992;79:179–84.
3. Magann EF, Dodson MK, Ray MA, Harris RL, Martin JN, Morrision JC. Preoperative skin preparation and intraoperative pelvic irrigation: impact on post-cesarean endometritis and wound infection. Obstet Gynecol 1993;81:922–5.
4. Chang PL, Newton ER. Predictors of antibiotic prophylactic failure in post-cesarean endometritis. Obstet Gynecol 1992;80:117–22.
5. Watts DH, Hillier SL, Eschenbach. Upper genital tract isolates at delivery as predictors of post-cesarean infections among women receiving antibiotic prophylaxis. Obstet Gynecol 1991;77:287–92.
6. Gabbe SG, Niebyl JR, Simpson JL, editors. Obstetrics: normal and problem pregnancies. 3rd ed. London (UK):Churchill Livingstone, Inc.; 1996.
7. Watts DH, Krohn MA, Hillier SL, Eschenbach DA. Bacterial vaginosis as a risk factor for post-cesarean endometritis. Obstet Gynecol 1990;75:52–8.
8. Roberts S, Maccato M, Faro S, Pinell P. The microbiology of post-cesarean wound morbidity. Obstet Gynecol 1993;81:383–6.
9. Berenson AB, Hammill HA, Martens MG, Faro S. Bacteriologic findings of post-cesarean endometritis in adolescents. Obstet Gynecol 1990;75:627–9.
10. Eason EL, Sampalis JS, Hemmings R, Joseph L. Povidone-iodine gel vaginal antisepsis for abdominal hysterectomy. Am J Obstet Gynecol 1997;176:1011–6.
11. Osborne NG, Wright RC. Effect of preoperative scrub on the bacterial flora of the endocervix and vagina. Obstet Gynecol 1977;50:148–50.
12. Amstey MS, Jones AP. Preparation of the vagina for surgery: a comparison of povidone-iodine and saline solution. JAMA 1981;245:839–41.
13. Reid VC, Hartmann KE, McMahon M, Fry EP. Vaginal preparation with povidone iodine and postcesarean infectious morbidity: a randomized controlled trial. Obstet Gynecol 2001;97:147–52.
14. Cunningham FG, Gant NF, Leveno KJ, Gilstrap LC III, Hauth JC, Wenstrom KD, editors. Williams obstetrics. 21st ed. New York (NY): McGraw-Hill; 2001.
15. Schrag S, Gorwitz R, Fultz-Butts K, Schuchat A. Prevention of prenatal Group B Streptococcal disease. Revised guidelines from CDC. MMWR Recomm Rep 2002;51(RR-11):1–22.
16. Haeri AD, Kloppers LL, Forder AA, Baillie P. Effect of different pre-operative vaginal preparations on morbidity of patients undergoing abdominal hysterectomy. S Afr Med J 1976;50:1984–6.
17. Pitt C, Sanchez-Ramos L, Kaunitz AM. Adjunctive intravaginal metronidazole for the prevention of postcesarean endometritis: a randomized controlled trial. Obstet Gynecol 2001;98:745–50.
18. Schulz KF, Chalmers I, Hayes RJ, Altman DG. Empirical evidence of bias: dimensions of methodological quality associated with estimates of treatment effects in controlled trials. JAMA 1995;273:408–12.