Discussion Questions

• ■
  

  How does the study design affect the researchers’ results?

  
  
• ■
  

  How might you explain the superior predictivity of the Nugent score?

  
  
• ■
  

  Do you agree with the researchers’ stated study limitations?

  
  
• ■
  

  Why might bacterial vaginosis be associated with STIs?

  
  
• ■
  

  How will this study influence the way you practice?

Introduction

Bacterial vaginosis (BV) is a common condition associated with a shift in the normal flora of the vagina. Sometimes this shift is accompanied by a symptomatic vaginal discharge that prompts the patient to seek medical care. Additionally, several studies have linked BV with an increased incidence of sexually transmitted infections (STIs). Notable new research by Allsworth and Peipert analyzes diagnostic strategies for the assessment of BV in an effort to identify patients at increased risk for STIs. Journal Club members were eager to determine whether categorization of BV severity would provide a marker that enhanced detection of undiagnosed STIs.

Kimberly Hoover, MD, Todd R. Jenkins, MD, and George A. Macones, MD, MSCE, Associate Editor

Study Design

Jenkins: This study was a secondary analysis of data obtained in the Project PROTECT study, which examined 2 computerized programs for enhancing the use of dual contraception. How does that affect the results obtained by Allsworth and Peipert?

Holley: It provides incomplete data in the sense that if part of the group used condoms, that could definitely have an effect on their incident STIs.

Jenkins: Also, I think knowing whether or not participants were treated for bacterial vaginosis as a part of the initial study would be valuable information. Another limitation is that we are not aware of what particular type of contraception subjects were using. Certainly, a method that increases cervical progesterone levels would result in thickening of the cervical mucus, which would reduce the risk of pelvic inflammatory disease more. I also think patients in the control arm would use condoms less and would have higher vaginal pH and a higher chance of exposure to STIs, and therefore, would be at higher risk for STIs.

Joiner: I think that the authors expect readers to make an assumption regarding the type of patient population involved in the study; that it is a population marked by high-risk sexual activity. You can extrapolate from the goals of the original study that they used some kind of contraception with a barrier method as STI protection. However, as stated, they don’t clarify this in the paper.

Jenkins: In this study, the Nugent score, at least among those patients with a Nugent score of 9 or 10, was more predictive of STI risk than Amsel’s criteria. Why might that have occurred?

Joiner: Well, they did not specify which of the 4 Amsel’s criteria were found on the exams. There is a significantly higher incidence of BV when you specifically look at only the pH component as opposed to 1 of the other 3 criteria. The authors do not delineate this in the study.

Jenkins: My impression is that Amsel’s is more nonspecific as opposed to Nugent being more specific for overgrowth of anaerobic bacteria. However, it was interesting to me that it required such a high Nugent score before it became significant.

Gleason: I think that people with more distorted vaginal flora at the initiation of the study ended up with more problems. In my mind, these patients had more problems going into the study, whereas the other patients were perhaps more on the normal side.

Greer: I think that it was also beneficial that you had 1 physician looking at the Gram stains as opposed to multiple providers giving a more subjective view on Amsel’s criteria.

Joiner: Thus the Nugent scores are a lot more consistent across the board because they are more standardized, whereas the Amsel’s criteria would be more subjective.

Holley: There is compelling evidence that the Nugent criteria have a higher sensitivity and specificity than the Amsel’s criteria. One reason the Nugent score is more precise is because it requires a certified reader. They mentioned in the study that all of these Nugent slides were evaluated in Sharon Hillier’s lab at McGee-Womens Hospital of the University of Pittsburgh Medical Center. In other words, the basic clinician cannot even grade by the Nugent Score because, first, it requires a Gram stain and we don’t use that in the clinic, and second, you have to be certified to grade these.

Hoover: So, that brings up the cost effectiveness of this model for clinical use.

Results

Jenkins: The researchers listed limitations, including limited power for subgroup comparisons, especially regarding race, and a single timepoint assessment for BV. They also describe some more specific limitations regarding the Nugent score and different levels of Lactobacillus based on ethnicity. Do we agree with their list of limitations?

Greer: In the abstract, they talked about the fact that this was a study performed over time. However, in the limitations, they say that they only addressed one specific exam.

Jenkins: I agree, I think a significant limitation of the trial is that we do not know how the patient’s BV was managed. Did patients with BV at initiation of the study receive treatment? Were they reassessed at discreet time points over the life of the study? Did they have BV present concomitantly a year later when they developed their STI? We were not provided with any evidence as to whether patients had recurrent or continuous BV versus a single episode. I think that these limitations make the study more of an assessment of high-risk sexual behavior as opposed to an assessment of anaerobic flora in the vagina (BV) as a predisposing factor for STIs.

Joiner: Again, the authors do comment that there is a significant minority of patients whose BV will self-resolve in time without treatment. They mention this but don’t address it any further.

Jenkins: I guess one thing that would be helpful for me is at the time that they were diagnosed with their STI, did they have BV? Did they have abnormal flora at the time of STI exposure and infection? To me, not knowing that was a limitation.

Gleason: Another limitation is the fact that we don’t receive information regarding whether or not patients were symptomatic and treated at any time during the study. They don’t address this situation at all, do they?

Jenkins: It does mention in the materials and methods of the study that “participants were encouraged to contact the study team if they experience any gynecologic problems,” but they don’t provide us with any data on how many people were treated during the course of the trial. Another limitation mentioned in their initial discussion is that the outcome of interest was described as time from baseline BV to the first positive STI result. The authors described how they were testing for herpes simplex virus (HSV) infection at initiation of the study, 12 months, and 24 months, but I did not see that they assessed other STIs on a regular schedule.

Jenkins: They did not describe performing any testing other than HSV at these intervals. Even if they did perform yearly testing, it was yearly testing. Who knows what happened in the intervening time? The point I was making was that the authors didn’t comment on time to incident infection at all. They just said that the people with the highest Nugent score were 2.7 times more likely to get an STI. We do not know if this was in 3 months or 22 months.

Hoover: Another limitation is the lack of power to perform a subgroup analysis. Their power was 84%. I think they were looking for higher numbers than that. That’s another small limitation.

Holley: I think to reiterate what Dr Jenkins said, the only thing that we know is that the patients that had the highest incident STI had the highest Nugent scores at the initiation of the study. We have no idea what their vaginal microflora was at the incident STI. So, there are virtually no conclusions that you can draw from the study in terms of the risk of the STI correlating with their Nugent scores at the time of the STI.