OBJECTIVE: To determine whether the current clinical criteria for diagnosing bacterial vaginosis can be simplified by using 2 clinical criteria rather than the standard 3 of 4 criteria (Amsel's criteria).
METHODS: This was a prospective observational study of 269 women undergoing a vaginal examination in the Women's Primary Care Center, Division of Research, or Colposcopy Clinic at Women & Infants Hospital. All 4 clinical criteria for diagnosing bacterial vaginosis were collected, and Gram stain was used as the gold standard. Sensitivity and specificity were calculated for each individual criterion, combinations of criteria, and a colorimetric pH and amine card. Receiver operating characteristic curve was generated to estimate the preferred pH and percentage of clue cells for diagnosing bacterial vaginosis.
RESULTS: The prevalence of bacterial vaginosis in our study population was 38.7%. Vaginal pH was the most sensitive of all the criteria, at 89%, and a positive amine odor was the individual criteria with the highest specificity, at 93%. Similar specificity was seen with combinations of 2 criteria and Amsel's criteria. Receiver operating characteristic curve analysis yielded a preferred pH and percentage of clue cells of 5.0 and 20%, respectively. However, a pH of 4.5 or greater improves sensitivity with minimal loss of specificity.
CONCLUSION: The clinical criteria for diagnosing bacterial vaginosis can be simplified to 2 clinical criteria without loss of sensitivity and specificity.
LEVEL OF EVIDENCE: II-2
The clinical criteria for diagnosing bacterial vaginosis can be simplified without loss of sensitivity or specificity.
From the *Division of Research, Department of Obstetrics and Gynecology, Women & Infants Hospital, and the †Center for Statistical Sciences, Brown University, Medical School, Providence, Rhode Island.
Preliminary data presented at the Infectious Disease Society of Obstetrics and Gynecology (IDSOG) Annual Meeting in Quebec City, Canada, August 9, 2001.
The authors thank CooperSurgical for the donation of the FemExam cards used in this study.
Reprints are not available. Address correspondence to: Robert E. Gutman, MD, 4940 Eastern Avenue, Room 125, Baltimore, MD 21224-2780; e-mail: email@example.com.
Received June 3, 2004. Received in revised form September 7, 2004. Accepted October 7, 2004.
Bacterial vaginosis is the most common vaginal infection, with a prevalence of 9–37%, depending on the population studied.1,2 Bacterial vaginosis has been associated with increased risk of preterm delivery,3 premature rupture of membranes,4 amniotic fluid infection,5 chorioamnionitis,6 postpartum endometritis,7–9 pelvic inflammatory disease,10–12 and postoperative infections.13,14
Currently, the clinical criteria for diagnosing bacterial vaginosis requires the presence of 3 of the following 4 criteria: 1) a thin, homogeneous discharge, 2) vaginal pH greater than 4.5, 3) a positive “whiff” test or release of amine odor with the addition of base, and 4) clue cells on microscopic evaluation of saline wet preparation. These criteria are based on the original work of Amsel et al.15 The Gram stain is believed by many to be the gold standard for diagnosing bacterial vaginosis. However, interpreting the Gram stain requires experience, and it is often difficult to get timely results for the clinical diagnosis of bacterial vaginosis. Thus, it is important to have simple and reliable clinical criteria that clinicians can use in practice. Because many clinicians do not routinely evaluate patients for all 4 of Amsel's criteria, it would be helpful to know if different combinations of criteria can be used to accurately diagnose bacterial vaginosis.
Although many research studies have looked at the sensitivity and specificity of Amsel's criteria, few have looked at these criteria individually and in various combinations using the Gram stain as the gold standard. We conducted a MEDLINE search from 1966 to November 2003, using the terms “bacterial vaginosis,” “BV,” and “nonspecific vaginitis.” We were unable to find any studies that evaluate all of the different combinations of clinical criteria for the purpose of simplifying and possibly improving the accuracy of current clinical methods for diagnosing bacterial vaginosis.
We hypothesize that the current clinical criteria for diagnosing bacterial vaginosis can be simplified by using 2 clinical criteria, rather than the standard 3 of 4 criteria (Amsel's criteria), without loss of sensitivity or specificity. Furthermore, we believe that the addition of the FemExam card (CooperSurgical Inc, Trumbull, CT), a colorimetric pH and amine card, will have better sensitivity and specificity when compared with Amsel's criteria.
MATERIALS AND METHODS
We performed a prospective observational study of 269 women recruited from various locations at Women & Infants Hospital over a 20-month period, beginning August 1, 2000. Any woman undergoing a speculum examination at the Women's Primary Care Center, Colposcopy Clinic, or Division of Research was eligible to participate in the study. Because all specimens were taken in patients who were already undergoing examinations, the study was approved by our Institutional Review Board as a residual tissue study. Thus, a separate, specific informed consent was not obtained, and information from the study was not used for clinical care or the management of the patient. Women were excluded if there was a large amount of vaginal bleeding on examination. Routine pelvic examinations were performed by select second- through fourth-year obstetric and gynecology residents, research nurses, or an attending gynecologist (J.F.P.). All of the examining staff were trained in the use of the FemExam card by a CooperSurgical representative.
Descriptive variables obtained during evaluation included age, pregnancy status, parity, ethnicity, presence or absence of symptoms, and a sexually transmitted diseases history. The following procedures were used to establish the presence or absence of the 4 Amsel criteria: vaginal discharge was characterized according to its color and the presence of a thin homogeneous discharge. Swabs of the vaginal secretions were taken from the sidewalls. We placed the secretions from one swab on ColorpHast pH indicator strips (EM Science, Gibbstown, NJ) with a pH range of 4.0–7.0 to determine vaginal pH. Vaginal secretions were then combined with 2 drops of normal saline on a slide and covered with a coverslip for the wet mount. Examination under high-powered microscopy identified the percentage of clue cells. We added 10% potassium hydroxide solution to the vaginal secretions from the second swab, and the release of a “fishy” amine odor signified a positive “whiff” test. We then rolled the third swab along a glass slide and allowed it to air dry for the Gram stain. An assistant placed vaginal secretions from the final swab on the FemExam card. This card has 2 colorimetric test circles for the detection of vaginal pH of 4.7 or greater and volatile vaginal fluid amines. The individual test circles form a blue plus (+) sign when the result is positive and blue minus (–) sign when the result is negative. To minimize bias of the clinical evaluation, the examiner read the FemExam card immediately after recording all other information.
Patients found to have bacterial vaginosis by clinical criteria were treated according to the current standard of care. We sent the Gram stain to Magee Women's Hospital research laboratory where a standardized 0–10 point score was assigned. According to Nugent's criteria,16 a score of 7 or greater defined the gold standard diagnosis of bacterial vaginosis.
We calculated the sensitivity, specificity, and 95% confidence intervals for each of the individual criterion, combinations of the criteria, Amsel's criteria, and the FemExam card. Patient characteristics were compared between women with and those without bacterial vaginosis by using the Student t test and Wilcoxon rank-sum test for continuous data and χ2 test for nonparametric categorical data. We generated a receiver operating characteristic curve to estimate the preferred pH and percentage of clue cells for the diagnosis of bacterial vaginosis.
The prevalence of bacterial vaginosis in our study population was 38.7%. Table 1 displays the demographic and clinical characteristics of the participants. Women with bacterial vaginosis had slightly higher mean age (25.4 versus 23.3 years, P = .03) and median parity (1.0 versus 0, P < .01). Black and Hispanic women had a higher prevalence of bacterial vaginosis than white women (P = .02). Our population consisted of 37.5% white women, 30.4% black women, and 27.6% Hispanic women. Women with bacterial vaginosis were more likely to be symptomatic, with individual symptoms of vaginal discharge and foul odor being more common (P < .01). Those with bacterial vaginosis were also more likely to have a history of Neisseria gonorrhoeae infection (P = .04) and trichomoniasis (P = .02), but a history of at least one sexually transmitted disease was not significantly increased in the women with bacterial vaginosis. Examination and recruitment location did not affect the prevalence of bacterial vaginosis.
The sensitivity and specificity and their respective 95% confidence intervals of individual criterion and combinations of criteria are shown in Table 2. Vaginal pH was the criteria with the highest sensitivity, at 89%. The sensitivity of the remaining 3 individual criteria ranged from 67% to 79%, and any combination of 2 criteria ranged from 61% to 69%. Amsel's criteria had a sensitivity of 69%. Positive amine odor was the individual criteria with the highest specificity, at 93%. Similar specificity was seen with combinations of 2 criteria and Amsel's criteria (86–95%). Thin, homogeneous discharge had the lowest specificity, at 54%.
The sensitivity, specificity, and 95% confidence intervals of the FemExam card criteria are shown in Table 3. The FemExam card had a similar sensitivity, but lower specificity for pH, and a similar specificity but lower sensitivity for the presence of amines when compared to the “whiff” test. When 1 of the 2 criteria was present on the card, the sensitivity was 89% and specificity was 61%. When both of the criteria were present, the sensitivity was 40% and the specificity was 95%.
A receiver operating characteristic curve was generated for pH and clue cells on wet preparation. We visually estimated the point of inflection on the receiver operating characteristic curve. A pH of 5.0 or greater maximized sensitivity (83%) and specificity (82%). However, a pH of 4.5 or more improves sensitivity to over 89% while still maintaining a relatively high specificity of 74% (Fig. 1). Sensitivity and specificity of clue cells appears to be maximized with a cutoff of more than 20% clue cells on saline wet preparation (Fig. 2; sensitivity 74%, specificity, 86%).
The prevalence of bacterial vaginosis was higher in our study (39%) than in most studies,1,2,17 and thus represents a high-risk population. Although the ethnic diversity of this population is not representative of most U.S. communities, it provides us with a better sampling of black and Hispanic women, who appear to be at higher risk of having bacterial vaginosis than white women.17 We confirmed the findings seen in another large cross-sectional study, which showed that patients with bacterial vaginosis are more likely to have vaginal symptoms, specifically vaginal discharge and foul odor.17,18 A study by Larsson et al19 revealed similar risk factors for sexually transmitted diseases and bacterial vaginosis, including lower age of first intercourse and higher number of lifetime sexual partners. Yen et al17 displayed a lower rate of bacterial vaginosis among the “non–sexually experienced” and a higher rate among those with multiple sexual partners in the past 3 months. However, in our study and in Yen's, a history of a sexually transmitted disease was not associated with an increased prevalence of bacterial vaginosis.
We confirmed the findings of Eschenbach et al18 that vaginal pH had the highest sensitivity, but the lowest specificity, of all clinical diagnostic criteria. The low specificity is not surprising considering that elevated pH may be altered by semen, cervical mucus, and blood. An elevated pH is more common in postmenopausal women20 and can also be seen with trichomoniasis infections. The high sensitivity should serve as an asset in screening algorithms. Our findings for amine odor also support previous studies that found it to have the highest specificity and lowest sensitivity of the individual criteria.18,21 The combination of any 2 criteria decreased sensitivity and increased specificity. However, the sensitivity and specificity of each pair of criteria is similar to those observed for Amsel's criteria. This implies that there is no clear advantage to using Amsel's criteria (≥ 3 of the 4 criteria). This supports our hypothesis that the current clinical criteria for diagnosing bacterial vaginosis (Amsel's criteria) can be simplified by using 2 clinical criteria, without significant loss of sensitivity or specificity.
One study evaluating an objective colorimetric test for amines found a sensitivity of 87% and specificity of 98% compared with Gram-stain criteria.22 The authors concluded that these tests are more sensitive than the “whiff” test for detecting the presence of amines. We found the FemExam card to have less sensitivity than the “whiff” test for detecting amines and lower specificity for detecting an elevated vaginal pH. The difference in pH may be due to the different cutoff points: 4.5 for our study and 4.7 for the FemExam card. Overall, we felt the majority of the differences could be explained by the difficulty in reading and interpreting the results of the card. Although it is conceptually easy to distinguish between a plus sign and a minus sign, the results often had very faint blue signals that were difficult to interpret. The findings may also be biased because the card was read after the clinical criteria were collected and recorded, although we would have expected this to falsely raise the sensitivity and specificity. Also, the slight delay in reading the cards may have resulted in suboptimal outcomes. Ultimately, the pH measurements are not statistically different based on the area under the receiver operating characteristic curve and 95% CI, but the amine measurements appear to be worse for the card because of the reasons listed above. This difference persists when pH and amine are both positive because of the large discrepancy in the sensitivity of amine measurements. The findings of this study fail to support routine use of the FemExam card to improve sensitivity and specificity.
The receiver operating characteristic curve confirms previously set cutoff points for pH and percentage of clue cells. The presence of greater than 20% clue cells on saline wet preparation appears to optimize sensitivity and specificity (Fig. 2). Although a pH of 5.0 or greater appears to optimize sensitivity and specificity, we prefer a pH of 4.5 or greater to improve sensitivity with an acceptable decline in specificity (Fig. 1). This is consistent with the majority of studies that use a pH of 4.5 or 4.7 for the cutoff of clinically diagnosing bacterial vaginosis.
Before initiating the study, we did a power and sample size calculation. Based on our desire to have relatively narrow 95% confidence intervals (± 7%) with an estimated prevalence of 25%, we calculated a sample size of approximately 500 patients. Our population, however, had a higher prevalence of bacterial vaginosis and ultimately resulted in a diagnosis of bacterial vaginosis by Gram-stain criteria in 104 patients. Consequently, despite having 131 fewer patients than our target, we had only 21 fewer patients with bacterial vaginosis. This resulted in slightly larger 95% confidence intervals than anticipated. There is overlap between the confidence intervals for sensitivity and specificity and the area under the curve among individual criteria and combinations of criteria (Table 2). Because of these similarities, a much larger sample size would be required to detect a statistically significant difference. The relatively large number of patients with bacterial vaginosis, the use of Gram stain as the gold standard diagnosis of bacterial vaginosis, and careful analytic methods, including receiver operating characteristic curve analysis, are strengths of this study. The high prevalence of bacterial vaginosis, however, may limit the generalizability of the results.
Because of the superior sensitivity of pH for the detection of bacterial vaginosis, we recommend an algorithm that involves initial diagnostic testing for bacterial vaginosis with pH, using test paper with an appropriate pH range (pH 4.0–6.0). Testing for one of the other 3 criteria should then be performed. If the first 2 tests are positive for bacterial vaginosis, the diagnosis can be made and appropriate treatment given. If one or both tests are negative, then additional tests should be performed until a diagnosis is made. This algorithm is only indicated for the detection of bacterial vaginosis and cannot be used to rule out other infectious processes that may occur in conjunction with bacterial vaginosis. Appropriate testing, including microscopy, should be performed as clinically indicated to rule out other infectious etiologies. In more complicated cases of recurrent or persistent bacterial vaginosis, a Gram stain should be considered to corroborate the diagnosis.
In conclusion, we believe that the evidence in this study supports simplifying the clinical diagnosis of bacterial vaginosis. We recommend using any 2 of the 4 clinical criteria to achieve a test performance similar to that achieved by using the criteria established by Amsel et al.15
1. Goldenberg RL, Klebanoff MA, Nugent R, Krohn MA, Hillier S, Andrews WW. Bacterial colonization of the vagina during pregnancy in four ethnic groups. Vaginal Infections and Prematurity Study Group. Am J Obstet Gynecol 1996;174:1618–21.
2. Hill LH, Ruparelia H, Embel JA. Nonspecific vaginitis and other genital infections in three clinic populations. Sex Transm Dis 1983;10:114–8.
3. Leitich H, Bodner-Adler B, Brunbauer M, Kaider A, Egarter C, Husslein P. Bacterial vaginosis as a risk factor for preterm delivery: a meta-analysis. Am J Obstet Gynecol 2003;189:139–47.
4. Mercer BM, Golderberg RL, Meis PJ, Moawad AH, Shellhaas C, Das A, et al. The Preterm Prediction Study: prediction of preterm premature rupture of membranes through clinical findings and ancillary testing. The National Institute of Child Health and Human Development Maternal–Fetal Medicine Units Network. Am J Obstet Gynecol 2000;183:738–45.
5. Silver HM, Sperling RS, St Clair PJ, Gibbs RS. Evidence relating bacterial vaginosis to intraamniotic infection. Am J Obstet Gynecol 1989;161:808–12.
6. Hillier SL, Martius J, Krohn M, Kiviat N, Holmes KK, Eschenbach DA. A case-control study of chorioamnionic infection and histologic chorioamnionitis in prematurity. N Engl J Med 1988;319:972–8.
7. Jacobsson B, Pernevi P, Chidekel L, Jorgen Platz-Christensen J. Bacterial vaginosis in early pregnancy may predispose for preterm birth and postpartum endometritis. Acta Obstet Gynecol Scand 2002;81:1006–10.
8. Newton ER, Prihoda TJ, Gibbs RS. A clinical and microbiologic analysis of risk factors for puerperal endometritis. Obstet Gynecol 1990;75:402–6.
9. Watts DH, Krohn MA, Hillier SL, Eschenbach DA. Bacterial vaginosis as a risk factor for post-cesarean endometritis. Obstet Gynecol 1990;75:52–8.
10. Sweet RL. Role of bacterial vaginosis in pelvic inflammatory disease. Clin Infect Dis 1995;20(suppl):S271–5.
11. Peipert JF, Montagno AB, Cooper AS, Sung CJ. Bacterial vaginosis as a risk factor for upper genital tract infection. Am J Obstet Gynecol 1997;177:1184–7.
12. Wiesenfeld HC, Hillier SL, Krohn MA, Amortegui AJ, Heine RP, Landers DV, et al. Lower genital tract infection and endometritis: insight into subclinical pelvic inflammatory disease. Obstet Gynecol 2002;100:456–63.
13. Lin L, Song J, Dimber N, Shott S, Tangora J, Aroutchea A, et al. The role of bacterial vaginosis in infection after major gynecologic surgery. Infect Dis Obstet Gynecol 1999;7:169–74.
14. Persson E, Bergstrom M, Larsson PG, Moberg P, Platz-Christensen JJ, Schedvins K, et al. Infections after hysterectomy: a prospective nation-wide Swedish study. The Study Group on Infectious Diseases in Obstetrics and Gynecology within the Swedish Society of Obstetrics and Gynecology. Acta Obstet Gynecol Scand 1996;75:757–61.
15. Amsel R, Totten PA, Spiegel CA, Chen KC, Eschenbach D, Holmes KK. Nonspecific vaginitis: diagnostic criteria and microbial and epidemiologic associations. Am J Med 1983;74:14–22.
16. Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing bacterial vaginosis is improving by a standardized method of Gram stain interpretation. J Clin Microbiol 1991;29:297–301.
17. Yen S, Shafer MA, Moncada J, Campbell CJ, Flinn SD, Boyer CB. Bacterial vaginosis in sexually experienced and non-sexually experienced young women entering the military. Obstet Gynecol 2003;102:927–33.
18. Eschenbach DA, Hillier S, Critchlow C, Stevens C, DeRouen T, Holmes KK. Diagnosis and clinical manifestations of bacterial vaginosis. Am J Obstet Gynecol 1988;158:819–28.
19. Larsson PG, Platz-Christensen JJ, Sundstrom E. Is bacterial vaginosis a sexually transmitted disease? Int J STD AIDS 1991;2:362–4.
20. Brizzolara S, Killeen J, Severino R. Vaginal pH and parabasal cells in postmenopausal women. Obstet Gynecol 1999;94:700–3.
21. Bump RC, Zuspan FP, Buesching WJ 3rd, Ayers LW, Stephens TJ. The prevalence, six-month persistence, and predictive values of laboratory indicators of bacterial vaginosis (nonspecific vaginitis) in asymptomatic women. Am J Obstet Gynecol 1984;150:917–24.
© 2005 The American College of Obstetricians and Gynecologists
22. Sonnex C. The amine test: a simple, rapid, inexpensive method for diagnosing bacterial vaginosis. Br J Obstet Gynaecol 1995;102:160–1.