Bacterial vaginosis as a risk factor for upper genital tract infection

Jeffrey F. Peipert, MD, MPH, Andrea Boyd Montagno, RN, Amy Sedlacek Cooper, MSN, CRNP, and C. James Sung, MD
Providence, Rhode Island OBJECTIVE: The objective of this study was to determine whether the clinical diagnosis of bacterial vaginosis is associated with objective evidence of acute upper genital tract infection. STUDY DESIGN; Women who either met the Centers for Disease Control and Prevention's minimal criteria for acute pelvic inflammatory disease or had other "nonclassic" signs of upper genital tract infection (i.e., atypical pelvic pain, abnormal uterine bleeding, or cervicitis) were evaluated with either an endometrial biopsy or a laparoscopy with endometrial and fimbrial biopsies for objective evidence of upper genital tract infection. Bacterial vaginosis was considered present if three of the four following criteria were found: (1) homogeneous gray-white vaginal discharge, (2) vaginal pH >4.5, (3) positive "whiff" test result, and (4) the presence of >20% of epithelial cells classified as clue cells. Patients were considered to have upper genital tract infection if they had histologic, microbiologic, or laparoscopic evidence of upper tract infection. RESULTS: One hundred sixteen women were evaluated between August 1993 and March 1997 with complete evaluations. Objective evidence of upper tract infection was present in 56% (14/25) of women with the clinical diagnosis of bacterial vaginosis compared with 30% of women (27/91) who did not meet the clinical criteria (p = 0.015). Using logistic regression to control for confounding variables, we found that the presence of bacterial vaginosis was associated with a threefold increased risk of upper genital tract infection (adjusted odds ratio = 3.0, 95% confidence interval 1.2 to 7.6). CONCLUSIONS: Bacterial vaginosis is associated with an increased risk of objective evidence of acute upper genital tract infection. Future prospective studies are needed to determine whether treatment of bacterial vaginosis can reduce the risk of ascending infection. (Am J Obstet Gynecol 1997;177:1184-7.)

Key words: Bacterial vaginosis, adnexitis, endometritis, risk factors

Pelvic inflammatory disease is broadly defined as infection of the female upper genital tract, including endometritis, salpingitis, oophoritis, pelvic peritonitis, or tuboovarian abscess. 1 Diagnosis and treatment of pelvic inflammatory disease are major public health problems in the United States, with considerable economic consequences. Elucidating risk factors for this disease is important to guide preventive and therapeutic strategies. Several recent studies have demonstrated an association between bacterial vaginosis mad pelvic inflammatory disease or histologic evidence of endometritis. 2-5 Because
From the Departments of Obstetrics and Gynecology and Pathology, Women & Infants' Hospital, Brown University School of Medicine. Supported by the Rhode Island Foundation and American Association for A1DS Research (AmFAR) grant No. 02110~15-RG. Presented in part at the annual meeting of District I, The American College of Obstetricians and Gynecologists, Charleston, South Cmvlina, October 10413, 1996. Received for publication March 11, 1997; revised April 22, 1997; accepted May 28, 1997, Reprint requests:Jeffrey F. Peipert, MD, MPH, Department of Obstetrics and Gynecology, Women & Infants' Hospital, 101 Dudley St., Providence, 1:7I02905. Copyright 1997 by Mosby-Year Book, Inc. 0002-9378/97 $5.00 + 0 6/1/83578
1184

the clinical diagnosis of pelvic inflammatory disease is fraught with a high n u m b e r of false-negative and falsepositive diagnoses, 6 we believe that objective evidence of upper tract infection is necessary in epidemiologic studies of peMc inflammatory disease. The purpose of our study was to evaluate the association of bacterial vaginosis witla objective evidence of acute upper genital tract infection in women who are first seen with classic and "nonclassic" symptoms of pelvic inflammatory disease.
Material and methods

A cross-sectional study was performed to evaluate currently available diagnostic tests and risk factors for acute upper genital tract infection. Because some authorities believe that the term pelvic inflammatory disease is synonymous with acute salpingitis, we use the term upper genital tract infection throughout this manuscript to refer to infection of the endometrium, fallopian tubes, ovaries, and pelvis to avoid controversy. We previously reported the methods used for this study in an initial evaluation of laboratory testing for upper genital tract infection,v Briefly, informed consent was obtained from women who chose to participate in

Volume 177, Number5 AmJ Obstet Gynecol

Peipert et al.

1185

the study if they were first seen with either "classic" or "nonclassic" signs of upper genital tract infection. "Classic" signs of upper genital tract infection included abdominal, cervical motion, and adnexal tenderness. "Nonclassic" signs included atypical pain (i.e., <30 days' duration) that did not meet the minimal criteria of the Centers for Disease Control and Prevention (CDC) for acute pelvic inflammatory disease, abnormal uterine bleeding of unclear etiology (in women <40 years old, to avoid those who were perimenopausal), or cervicitis as demonstrated by a mucopurulent cervical discharge or a positive cervical culture for either Neisseria gonorrhoeae or Chlamydia trachomatis, or both. The choice of this broad range of signs of upper genital tract infection was deliberate; the intention was t o include a full spectrum of cases required to evaluate the chosen diagnostic tests, s All participants were <46 years old. Exclusion criteria included pregnancy, delivery, or operative procedure within the preceding 4 weeks, an inability to give informed consent, or a history of hysterectomy or bilateral salpingectomy. Evaluation began with a comprehensive enrollment interview reviewing medical, obstetric and gynecologic, and sexual history. Physical examination and laboratory testing were then performed. Vaginal secretions were screened for white blood cells and evaluated for evidence of vaginitis as documented by the presence of clue cells, trichomonads, and pseudohyphae or spores. Bacterial vaginosis was clinically diagnosed if three of the four following criteria were found: (i) homogeneous graywhite vaginal discharge, (2) vaginal pH >4.5, (3) positive "whiff" test result, and (4) the presence of >20% of epithelial cells classified as clue cells. This criterion has been shown to have a high positive predictive value (92%) when compared with Gram stain criteria. 9' 10 Cervical and endometrial cultures were obtained for N. gono,rhoeae and C. trachomatis. All participants underwent either endometrial biopsy alone or a combination of endometrial biopsy and laparoscopy to obtain objective evidence of acute upper genital tract infection. This decision was left to the patient. However, laparoscopy was not offered if the patient did not have pain or tenderness. If a laparoscopy was performed, tubal and peritoneal cultures for N. gonorrhoeae and C. trachomatis together with a fimbrial minibiopsy were done. The objective criteria used for the diagnosis of upper genital tract infection included histologic evidence o f acute endometritis or salpingitis, laparoscopic evidence of salpingitis as demonstrated by purulent exudate in the pelvis with no other source, or microbiologic evidence of N. gonorrhoeae or C. trachomatis from the endometrium or fallopian tubes. We defined histologic endometritis as the presence of acute endometritis. For acute endometritis to be verified, neutrophils had to be present in one or more of the following areas in the absence of tissue

Table I. Demographic and reproductive characteristics of study population

Characteristic
Race White Black Hispanic Other Insurance Private Medicaid or uninsured Marital status Married Single Widowed, divorced, or separated Smoker History of sexually transmitted diseases History of pelvic inflammatory disease 43 37 35 1 25 91 26 78 12 57 58 23

% 37 32 30 1 22 78 22 67 10 49 50 20

disintegration such as menstrual changes: the surface epithelium, the subepithelial stroma, or the endometrial glands. 11 When plasma cells were present in addition to neutrophils, a histologic diagnosis of acute and chronic endometritis was made. 11q3 The study protocol was approved by the Women & Infants' Hospital Institutional Review Board before initiation of the research. Sample size calculations were performed on the basis of the following assumptions: prevalence of bacterial vaginosis of 33%, evidence of upper genital tract infection in 20% of patients without bacterial vaginosis, relative risk of 3.0 in patients with bacterial vaginosis, c~ = 0.05, and power of 80%. This yielded a total sample size of 81 patients. Statistical analysis was performed with Statistical Analysis Software (SAS, Cary, N.C.) and Epi-Info 6.0 (Centers for Disease Control and Prevention, Atlanta). The X2 and Fisher's exact tests were used for categoric variables. Odds ratios and 95% confidence intervals were also calculated. Logistic regression was performed to control for potential confounding variables in the association between bacterial vaginosis and upper genital tract infection. Values of p < 0.05 were considered statistically significant. Results Between August 1993 and March 1997, 134 women were evaluated for evidence of upper genital tract infection. Three women were excluded because of inadequate endometrial samples, and 15 women were excluded because of inadequate saline solution vaginal preparation (in association with heavy bleeding) or missing data. The remaining 116 patients are the subject of this report. The demographic characteristics of the study population are shown in Table I. The median gravidity of the study population was 2 (range 0 to 12), and the median n u m b e r of years of education was 12 (range 7 to 16). The median n u m b e r of sexual partners in the 6 months

1186

Peipert et al.

November 1997
Am j Obstet Gynecol

Table II. Characteristics of patients with and without evidence of bacterial vaginosis

Bacterial vaginosis present (n = 25) Characteristic

Married White Smoker History of sexually transmined diseases History of pelvic inflammatory disease Multiple partners past 6 mo Early onset of coitus >3 lifetime partners Age (yr) Mean SD

Bacterial vaginosis absent (n = 91) No.

22 37 44 45 18 16 33 53 25.4 6.7

No.
4 6 13 13 5 5 8 15 24.3 5.5

%
16 24 52 52 20 20 32 60

%
24 41 48 49 20 18 37 62

Statistical significance
p= p= p= p= p= p= p= p= 0.76 0.13 0.75 0.82 0.98 0.78 0.64 0.88

p = 0.47

before study inclusion was 1 with a range of 0 to 5, and the median n u m b e r of lifetime sexual partners was 4 with a range from 1 to >100. Sixty-three of the 116 patients (54%) met the CDC's minimal criteria for pelvic inflammatory disease, 30 (26%) had pain and tenderness but did not meet the CDC's criteria, 10 patients (9%) had abnormal uterine bleeding, and 13 (11%) patients who were recruited had evidence of cervicitis (either mucopus on cotton swab test or positive cervical cultures). Sixty-two patients (53%) underwent endometrial biopsy whereas 54 (47%) had an endometrial biopsy and laparoscopic assessment. Of the 54 subjects who had a laparoscopic assessment, purulent exudate was noted in 17 (32%), 20 (37%) had edematous fallopian tubes, and 21 (39%) had tubal erythema. Eleven patients (20%) had the laparoscopic triad of edema, erythema, and purulent exudate. Of the total group (N = 116), acute endometritis was found in 22 (19%) and 11 (10%) had chronic endometritis. Five patients (4%) had evidence of acute and chronic endometritis. Four percent of subjects (5/116) were positive for N. gonorrhoeae from the cervix, and 22% (26/116) were positive for C. trachomatis. Four of 116 (3%) were positive for N. gonorrhoeae from the endometrium or fallopian tubes, and 11% (13/116) were positive for C. trachomatis from the upper genital tract. We first evaluated factors that may be associated with the clinical diagnosis of bacterial vaginosis. When compared with women without evidence of bacterial vaginosis, women meeting clinical criteria had similar ages, smoking history, history of sexually transmitted diseases or pelvic inflammatory disease, and sexual history (Table II). The only characteristic that approached statistical significance was race; patients with evidence of bacterial vaginosis were more likely to be nonwhite (76% vs 59%, p = 0.13). Fourteen of 25 patients with clinical evidence of bacterial vaginosis (56%) had objective evidence of upper genital tract infection compared with 27 of 91 (30%)

patients without bacterial vaginosis (p = 0.015). We then added cases of chronic endometritis to our definition of upper genital tract infection to see whether our results w o u l d vary. Fifteen of 25 (60%) subjects with bacterial vaginosis had upper tract infection according to this expanded definition, and 31 of 91 (34%) subjects without bacterial vaginosis had positive findings (p < 0.02). We then wanted to exclude the possibility of confounding as a cause for the noted association, so we performed multivariable analysis with logistic regression. We included race, bacterial vaginosis, and age in the model statement with objective evidence of upper genital tract infection as the dependent variable. Bacterial vaginosis continued to be statistically significantly associated with upper genital tract infection when we controlled for these covariates (adjusted odds ratio = 3.0, 95% confidence interval 1.2 to 7.6). Because our previous report identified elevated white blood cell count and C-reactive protein as important laboratory predictors of upper genital tract infection,7 we also did a separate logistic regression using these variables and the presence of bacterial vaginosis in the model. Bacterial vaginosis remained an important predictor of upper genital tract infection even after we controlled for these other laboratory variables (p = 0.03).

Comment
We found an association between the clinical diagnosis of bacterial vaginosis and the presence of objective evidence of upper genital tract infection in a group of women who had a wide spectrum of signs and symptoms of upper genital tract infection. The presence of bacterial vaginosis was associated with a threefold increase in the rate of positive objective criteria. Our findings are consistent with the few published studies in the literature evaluating this association. Paavonen et al. s found that 9 of 31 women with laparoscopically confirmed pelvic inflammatory disease had bacterial vaginosis (29%) compared with none of 14

Volume 177, Number 5 AmJ Obstet Oynecol

Peipert et al.

1187

w o m e n without laparoscopically c o n f i r m e d pelvic inflammatory disease. Korn et al. 4 also investigated the association between u p p e r genital tract infection and bacterial vaginosis and f o u n d that plasma cell endometritis was present in 10 of 22 w o m e n with bacterial vaginosis c o m p a r e d with o n e of 19 control subjects. Eschenbach et al. 5 d o c u m e n t e d a n i n e f o l d increased risk of the clinical diagnosis of pelvic inflammatory disease in w o m e n with bacterial vaginosis. In a r e c e n t study by Hillier et al. 2 using c o m p r e h e n s i v e microbiologic testing, anaerobic gram-negative rods were f o u n d to be associated with histologic endometritis. O u r findings support the role of bacterial vaginosis as a risk factor for u p p e r genital tract infection. O u r study, however, has several limitations. Because our study is a cross-sectional analysis, we cannot establish that the bacterial vaginosis flora p r e c e d e d the u p p e r genital tract infection. In o t h e r words, temporal sequence c a n n o t be clearly d e t e r m i n e d . Second, c o m p r e h e n s i v e microbiology data including anaerobic cultures were not perf o r m e d in a research laboratory e q u i p p e d to p e r f o r m these sensitive analyses. T h e r e are several strengths of this study that deserve m e n t i o n . Many o t h e r studies evaluating risk factors for u p p e r genital tract i n f e c t i o n rely on the clinical diagnosis o f pelvic i n f l a m m a t o r y disease, which is known to be inaccurate. O u r study relies on objective criteria i n c l u d i n g histologic features, m i c r o b i o l o g i c characteristics, a n d laparoscopy. A s e c o n d i m p o r t a n t s t r e n g t h is o u r b r o a d i n c l u s i o n criteria. We d e l i b e r a t e l y chose b o t h "classic" a n d "nonclassic" symptoms o f u p p e r tract i n f e c t i o n to i n c l u d e a wide r a n g e of patients. This, in turn, increases the e x t e r n a l validity of o u r findings. T h e question that remains unanswered is whether t r e a t m e n t of either symptomatic or asymptomatic bacterial vaginosis will result in fewer cases of u p p e r genital tract infection or o t h e r gynecologic complications. Future r a n d o m i z e d trials should evaluate w h e t h e r treatm e n t of bacterial vaginosis can result in a decrease in

u p p e r genital tract infectious and subsequent reproductive sequelae.

REFERENCES

1. Centers for Disease Control and Prevention. 1993 Sexually transmitted diseases treatment guidelines. MMWR Morb Mortal Wkly Rep 1993;42:75-8. 2. Hillier SL, Kiviat NB, Hawes SE, Hasselquist MB, Hanssen PW, Eschenbach DA, et al. Role of bacterial vaginosisassociated microorganisms in endometritis. Am J Obstet Gynecol 1996; 175:435-41. 3. Paavonen J, Teisala K, Heinonen PK, A_ine R, Laine S, Lehtineu M, et al. Microbiologic and histopathologicai findings in acute pelvic inflammatory disease. Br J Obstet Gynaecol 1987;94:454-60. 4. Korn AP, Bolan G, Padian N, Ohm-Smith M, Schacter J, Landers DV. Plasma cell endometritis in women with symptomatic bacterial vaginosis. Obstet Gynecol 1995;85:387-90. 5. Eschenbach DA, Hillier SL, Critchlow CW, Stevens C, DeRouen T, Holmes KK. Diagnosis and clinical manifestations of bacterial vaginosis. AmJ Obstet Gyneco11988;158:819-28. 6. Jacobson L, Westrom L. Objectivized diagriosis of acute pelvic inflammatory disease. AmJ Obstet Gynecol 1969;105: 1088-98. 7. Peipert JF, Boardman L, Hogan JW, Sung J, Mayer K. Laboratory evaluation of acute upper genital tract infection. Obstet Gynecol 1996;87:730-6. 8. PeipertJF, Sweeney PJ. Diagnostic testing in obstetrics and gynecology: A clinician's guide. Obstet GynecoI 1993;82: 619-23. 9. Amsel R, Totten PA, Spiegel CA, Chen KCS, Eschenbach D, Holmes KK. Nonspecific vaginitis: diagnostic criteria and microbial and epidemiological associations. AmJ Med 1983; 74:14-22. 10. SchwebkeJR, Hillier SL, SobelJD, McGregorJA, Sweet RL. Validity of the vaginal gram stain for the diagnosis of bacterial vaginosis. Obstet Gynecol 1996;88:573-6. 11. Kurman RJ, Mazur MT. Benign diseases of the endometrium. In: Kurman RJ, editor. Blaustein's pathology of the female genital tract. 4th ed. New York: Springer-Verlag; 1994. p. 369-70. 12. Poropatich C, Rojas M, Silverberg SG. Polymorphonuclear leukocytes in the endometrium during the normal menstrual cycle. IntJ Gynecol Pathol 1987;6:230-4. 13. Kiviat NB, Wolner-Hanssen P, Eschenbach DA, Wasserheit IN, PaavonenJA, Bell TA, et al, Endonietrial histopathology in patients with culture-proved upper genital tract infection and laparoscopically diagnosed acute salpingitis. Am J Surg Pathol 1990;14:167-75.