Maternal morbidity related to infections after cesarean was eight-fold higher than after vaginal delivery.1 Surgical site infection is defined operationally as infection involving the abdominal incision or the uterus.2,3 Total cost in the United States, including indirect expenses related to this morbidity, could exceed $10 billion annually.4
Reported rates of postcesarean surgical site infection vary greatly, from 0.3% in Turkey,5 11.6% in Brazil,6 to 18.3% in Saudi Arabia.7 Despite numerous investigations, there is disagreement about risk factors of surgical site infection after cesarean delivery. Many factors affect infection rates in different settings. Confounding variables were not sufficiently controlled in many of those studies. Therefore, we conducted this prospective study to determine postoperative complications and to identify risk factors for surgical site infection after cesarean, by multivariate analysis. A better understanding of predictors might improve infection control by reducing clinical effects of postcesarean infections.
Materials and Methods
We prospectively studied a cohort of 969 women who had cesareans at Hungvuong Hospital in Ho Chi Minh City, Vietnam. It is a 450-bed, tertiary care obstetric and gynecologic hospital with an average of 1300 deliveries and 350 major operations per month. It serves the population of 2.5 million women in Ho Chi Minh City and is a referral center for 18 district hospitals and the obstetrics and gynecology departments of other hospitals in the city.
From May to August 1997, all women who had cesareans were recruited. The principal investigator visited each postoperative ward twice weekly and collected all pertinent data. Demographic information, putative factors, and surgical indications were recorded. Host-related variables included age, residence, parity, body mass index (BMI), preoperative stay, existing comorbidities, prior amniocentesis, labor induction, rupture of membrane duration, and preoperative condition. Surgery-related variables included emergency nature of the operation, cesarean hysterectomy, surgical duration, wound class, type of anesthesia, type of abdominal incision, experience of surgeon, volume of blood loss, and timing of prophylactic antibiotics. The variable “residence” was specified as urban or rural. Body mass index was calculated using postpartum weight and height, which were determined on postoperative day 3 by research assistants. Preoperative stay was the interval in days between hospital admission and surgery. Existing comorbidities investigated were diabetes mellitus, remote infection, preeclampsia, anemia, and chorioamnionitis. Preoperative condition was assessed by the American Society of Anesthesiologists preoperative assessment score.8 A woman with mild systemic disease was classified as class 2, whereas classes 3 and 4 were associated with severe systemic disease. We applied the modified wound classification to cope with difficulties in obstetric wound classification. Cesareans were classified class I if there was no rupture of membranes or labor, class II if there was less than 2 hours of rupture of membranes without labor or labor of any length with no rupture of membranes, class III for rupture of membranes greater than 2 hours, and class IV for purulent amniotic fluid.9 Rupture of membranes duration was the interval, in hours, between recorded timing of rupture of membranes and surgical incision. An emergency cesarean delivery was defined as an operation for compelling reasons that had not been planned, and an elective cesarean was defined as an operation planned and done when scheduled or sooner if labor accelerated the delivery time.10 We classified surgeon experience into three levels, supervisor, attending physician, and resident. Volume of blood loss was calculated by subtracting total irrigation fluid used and amount of amniotic fluid from the total volume of fluid in the suction container at the end of surgery, then adding the amount of blood on sponges, determined by weight. The true volume of blood loss was recorded after subtracting the volume of possible blood replacement. Timing of antibiotic prophylaxis was classified as early (2–24 hours before surgical incision), preoperative (0–2 hours before incision), perioperative (within 3 hours after incision), or postoperative (more than 3 hours after the incision).
Postoperatively women were monitored for signs of infection. Temperature was measured orally every 4, 6, and 12 hours for the first, second, and following postoperative days, respectively. In women who had fever (temperature 38.0C or greater) the temperature was taken every 4 hours until it was less than 37.5C on two consecutive measurements. Leukocyte count was done routinely when a woman's temperature was over 38.5C. Further tests such as urine analysis, urine culture, and chest x-ray or wound culture were not done routinely unless infection was suspected. Surgical sites were re-examined by research assistants when women returned to the outpatient clinic as scheduled after discharge.
Surgical complications included intraoperative hemorrhage necessitating transfusion, postoperative hemorrhage, and injury to adjacent organs. Any bleeding event requiring postoperative intervention was called postoperative hemorrhage complication. The standard criterion of postoperative febrile morbidity was an oral temperature of at least 38.0C on any two of the first 10 days postpartum, excluding the first 24 hours.11 Postoperative infections were diagnosed by using the Centers for Disease Control and Prevention (CDC) definitions.2,3 Surgical site infection included superficial, deep, and space-organ infection. Endometritis and vaginal cuff infections constituted organ surgical site infections.3 Surgical site infection was operationally identified by purulent discharge, positive culture, deliberate reopening of surgical wounds, evidence of abscess, or diagnosis by the attending physician.3
Data management and analysis were done with the statistical software Epi Info version 6.04b (CDC, Atlanta, GA) and STATA version 5.0 (StataCorp, College Station, TX). χ2, Fisher exact test, and Student t test or Mann-Whitney test were used for discrete and continuous variables, when appropriate. Multiple logistic regression analysis was done to obtain adjusted estimates of odds ratios (ORs) and to identify independent risk factors. Variables that were likely to be associated with outcome (P ≤ .2 in univariate analysis) or considered potential confounders were included in the multiple logistic regression model. P < .05 was considered statistically significant.
During the 4-month study, there were 969 cesareans among 5181 deliveries, a rate of 18.7%. There were five obstetric hysterectomies, two of which were selectively indicated for different extents of placenta increta. More than four fifths of the women were free from comorbidity (Table 1). Ovarian carcinoma was confirmed histologically in one cesarean with oophorectomy; radical hysterectomy was done 1 month later. Although 19.4% of subjects had repeat cesareans, only 15.4% had previous operation as the only indication for the current cesarean (Table 2). Placenta previa, rather than previous cesarean, was recorded as the indication if this placental abnormality resulted in the current operation. External fetal heart rate monitoring was used for fetal assessment.
The rate of surgical complications was low. Estimated volume of blood loss over 1 L occurred in 15 cases, 12 associated with abnormal implantation of placenta. Blood replacement was indicated for 27 cases (2.8%). There was only one postoperative hemorrhage, diagnosed 4 hours after surgery. Partial bladder injury occurred in two cases of repeat cesarean; one was complicated by placenta increta.
There were 157 women (16.2%) with febrile morbidity and 95 (9.8%) with surgical site infections, 19 with urinary tract infections, two with pneumonia, and three cases of gastroenteritis. Superficial and organ-space surgical site infection accounted for 52.9% and 24.4% of all postoperative infections, respectively. Infection of deep soft tissues was diagnosed in three cesareans. One vaginal cuff infection was identified after a cesarean hysterectomy. Ten infected wounds were documented after discharge.
We identified ten factors that were associated with surgical site infections by univariate analysis (P < .05), including nulliparity, BMI, preeclampsia, remote infection, chorioamnionitis, preoperative hospitalization, rupture of membranes–operation interval, American Society of Anesthesiologists score, cesarean hysterectomy, and volume of blood loss. Other variables likely related to outcomes (P < .20) were residence, experience of surgeon, kinds of abdominal incisions, wound contamination level, and surgical duration. Labor induction (P = .96) and indication for dystocia (P = .32) were not likely to be associated with postcesarean infection. All cesarean deliveries received parenteral antibiotics for prophylaxis, and timing of administration appeared not to be closely associated with subsequent infections (P = .28). Those variables were included in the initial multiple logistic regression model. Multiple logistic regression found seven predictors independently associated with postcesarean surgical site infection (Table 3). Host susceptibility had an essential effect on prediction of that complication.
The short duration and CDC definitions of the present study allowed uniformity in diagnoses of various sources of postcesarean infections, highlighting the importance of complications and risk factors for surgical site infection. Surgical complications were rare in the present study. Our blood transfusion rate was within the limits of other reports, which varied from 1.2%12 to 6.3%.13 Incidence of reoperation because of intraabdominal hemorrhage in the study by Nielsen and Hokegard14 (0.3%) was triple that in ours. The likelihood of bladder injury in our data was identical to findings of Nielsen and Hokegard12 and Eisenkop et al.15 There was no ureter injury in our study, whereas Eisenkop et al reported a rate of 0.09%.15
Our postcesarean infection rate compared favorably with those of other settings. Febrile morbidity was reported from 15.3%16 to 23.3%13 of abdominal deliveries. The overall rate of postoperative infection in the present study was 12.4%, which is consistent with the rate of 13.9% reported in a prospective study of 1319 cesareans in Denmark.14 Our rate of surgical site infection, contributing four fifths of all postoperative infections, was within the range reported by others who also used CDC definitions.5–7,17 Neither the rate of Yalcin et al5 nor Eltahawy et al7 was determined from a large-scale study, so it might be distorted. Rates comparable to ours were reported from Canada (8.8%17 and 9.6%13) and Brazil (11.6%).6
Multiple logistic regression showed seven variables independently associated with postcesarean surgical site infection. Preoperative remote infection and chorioamnionitis, although they occurred in only a few of our subjects, appeared strongly to predispose women to postcesarean surgical site infection. Remote infection not only compromises the immune status of the patient but can increase the innoculum of microorganisms contaminating the surgical site. A remote infection indicates heavy abnormal bacterial colonization that can readily contaminate the surgical site. Garibaldi et al18 found that distant infections independently carried risks 2.8 times higher. Intrauterine,10 pathologic,19 and clinical intra-amniotic infections20,21 were found to be risk factors for subsequent endometritis. Prompt and aggressive antibiotic therapy should be started as soon as suspected infection is confirmed, to reduce subsequent postoperative infections.10
The American Society of Anesthesiologists physical status classification is a standardized, reproducible numeric determination that is used routinely to stratify severity of illness for surgical patients and is known to be a good indicator of host susceptibility to infection.8,18 A concept of antepartum risk factors, a complex proxy of host susceptibility, was proved closely correlated with postcesarean endometritis,1 but the association was not examined by multivariate analysis. Our data supported the finding of Garibaldi et al18 that the association between preoperative health status and postoperative wound infection remained valid even after multivariate modeling analysis. Severe systemic disease (American Society of Anesthesiologists class of 3 or more) can increase risk of infection by five times.
Preeclampsia increases risk of postcesarean infection by a factor of two. Asymptomatic bacteriuria was documented more frequently among preeclamptic women (19%) compared with controls (4.5%) (P < .005).22 That remote infection was not intentionally identified or included in analysis, so its contribution to the risk of postoperative infection could not be specified, but it could partially explain the association between preeclampsia and surgical site infection after cesarean.
Obesity as a risk of postcesarean wound infection has been known for years.9,14,23,24 Unlike general patients, it is usually difficult to measure precisely the weight of obstetric patients. Neither prepregnancy24 nor delivery data14 reflect the current condition of postpartum subjects. Patients' weight measured at postoperative day 3 would be a better indicator because it was not yet substantially affected by diuresis.25 We attempted to identify the association between increments of BMI, rather than an arbitrarily dichotomous variable of obesity, and postcesarean infection. The risk of infection doubled for every five-unit increment of BMI. That risk could be due to the relative avascularity of adipose tissue or technical difficulties of handling adipose tissue that can result in more traumas to the abdominal wall or difficulties to obliterate dead space in the fat abdominal wall.
Regardless of other risk factors, we found nulliparity and high volume of blood loss as determinants of postcesarean surgical site infection. Nulliparity was reported to be independently associated with increased risk of postcesarean endometritis.26 The precise mechanism by which nulliparity increases the risk of infection is not fully understood. The present data showed the risk of surgical site infection reduced by 39% and 60% when women had one or more children, respectively. Likewise, risk of postoperative infection has been shown to be proportional to volume of blood loss during cesarean.1,10 Risk of surgical site infection increased 30% for every 100-mL increment of blood loss. Precise measurement of blood loss during cesarean is almost impossible because of difficulties quantifying amniotic fluid. Blood loss volume was recorded without knowledge of outcome, so potential measurement bias would be minimal in our study. The association between each 100-mL increment of blood loss, instead of its precise volume, and postcesarean infection was examined, increasing the accuracy of the analysis. A high volume of blood loss is usually associated with poor control of bleeding, increased tissue damage from prolonged retraction and manipulation, and more sutures. Suture, a foreign body, can promote contamination and reduce local resistance mechanisms.
We could not confirm some variables that had been predictors of postcesarean wound infection, namely duration of rupture of membranes,1,9,14,23,24 and surgical duration.10,20,26 Confounding effects were controlled sufficiently by multiple logistic regression in only two studies.20,23 Prolonged rupture of membranes increased the likelihood of an infection ascending from vagina into uterine cavity. However, chorioamnionitis was closely related to prolonged rupture of membranes (P < .001). In a multivariate model this strong predictor would have masked the hypothesized association between prolonged rupture of membranes and postcesarean infection.
Although the cesareans that lasted longer than 1 hour had 2.4 times the risk of postoperative infection, by univariate analysis, a larger sample is needed to confirm its independent predictive role. Assuming an α of .05, power of .80, and assumed infection rate in controls of 8.5%, at least 264 subjects would be needed for each study arm.
1. Ott WJ. Primary cesarean section: Factors related to postpartum infection. Obstet Gynecol 1981;57:171–6.
2. Garner JS, Javis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections. 1988. Am J Infect Control 1988;16:128–40.
3. Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG. CDC definitions of nosocomial surgical site infections. 1992: A modification of CDC definitions of surgical wound infections. Infect Control Hosp Epidemiol 1992;13:606–8.
4. Wong ES. Surgical site infection. In: Mayhall CG, ed. Hospital epidemiology and infection control. Baltimore: Williams & Wilkins, 1996:154–75.
5. Yalcin N, Bakir M, Dokmetas I, Sabir N. Postoperative wound infections. J Hosp Infect 1995;29:305–9.
6. Starling CE, Cutto BR, Pinheiro SM. Applying the Centers for Disease Control and Prevention and National Nosocomial Surveillance system methods in Brazilian hospitals. Am J Infect Control 1997;25:303–11.
7. Eltahawy AT, Mokhtar AA, Khalar RMF, Bahnassy AA. Postoperative wound infection at a university hospital in Jeddah, Saudi Arabia. J Hosp Infect 1992;21:79–83.
8. Owens WD, Felts JA, Spitznagel EL. ASA physical status classifications: A study of consistency of ratings. Anesthesiology 1978;49:239–43.
9. Emmons SL, Krohn M, Jackson M, Eschenbach DA. Development of wound infections among women undergoing cesarean section. Obstet Gynecol 1988;72:559–64.
10. Hagglund L, Christensen K, Christensen P, Kamme C. Risk factors in cesarean section infection. Obstet Gynecol 1983;62:145–50.
11. Cunningham FG, MacDonald PC, Gant NF, Levento KJ, Gilstrap LC III. Puerperal infection. In: Williams obstetrics. 19th edition. Norwalk, Connecticut: Appleton & Lange, 1993:627–42.
12. Nielsen TF, Hokegard KH. Cesarean section and intraoperative surgical complications. Acta Obstet Gynecol Scand 1984;63:103–8.
13. Baskett TF, McMillen RM. Cesarean section: Trends and morbidity. Can Med Assoc J 1981;125:723–6.
14. Nielsen TF, Hokegard KH. Postoperative cesarean section morbidity: A prospective study. Am J Obstet Gynecol 1983;146:911–5.
15. Eisenkop SM, Richman R, Platt LD, Paul RH. Urinary tract injury during cesarean section. Obstet Gynecol 1982;60:591–6.
16. Guldholt I, Espersen T. Maternal febrile morbidity after cesarean section. Acta Obstet Gynecol Scand 1987;66:675–9.
17. Gravel-Tropper D, Oxley C, Memish Z, Garber GE. Underestimation of surgical site infection rates in obstetrics and gynecology. Am J Infect Control 1995;23:22–6.
18. Garibaldi RA, Cushing D, Lerer T. Predictors of intraoperative-acquired surgical wound infections. J Hosp Infect 1991;18(Suppl A):289–98.
19. Newton ER, Prihoda TJ, Gibbs RS. A clinical and microbiologic analysis of risk factors for puerperal endometritis. Obstet Gynecol 1990;75:402–6.
20. Suonio S, Saarikoski S, Vohlonen I, Kauhanen O. Risk factors for fever, endometritis and wound infection after abdominal delivery. Int J Gynaecol Obstet 1989;29:135–42.
21. Dinsmoor MJ, Gibbs RS. Previous intra-amniotic infection as a risk factor for subsequent peripartal uterine infections. Obstet Gynecol 1989;74:299–301.
22. Hill JA, Devoe LD, Bryans CI Jr. Frequency of asymptomatic bacteriuria in preeclampsia. Obstet Gynecol 1986;67:529–32.
23. Pelle H, Jepsen OB, Larsen SO, Bo J, Christensen F, Dreisler A, et al. Wound infection after cesarean section. J Infect Control 1986;7:456–61.
24. Martens M, Kolrud B, Faro S, Maccato M, Hammill H. Development of wound infection or separation after cesarean delivery. Prospective evaluation of 2,431 cases. J Reprod Med 1995;40:171–5.
25. Sheikh GN. Observation of maternal weight behavior during the puerperium. Am J Obstet Gynecol 1971;111:244–50.
© 2000 The American College of Obstetricians and Gynecologists
26. Chang PL, Newton ER. Predictors of antibiotic prophylactic failure in post-cesarean endometritis. Obstet Gynecol 1992;80:117–22.