Objective: To demonstrate the diagnostic accuracy for bacterial vaginosis (BV) by using the sole detection of clue cells

Materials and Methods: All pregnant women who received antenatal care (ANC) at Siriraj Hospital between October 2017 and March 2018 were invited. Eligibility criteria included age of 18 years or older and gestational age (GA) of 12 weeks or more, absence of blood or amniotic fluid, and no antibiotics taken in prior two weeks. Vaginal discharge was obtained, airdried, gram-stained, and sent to two microbiologists. Nugent’s scoring system, which was done by one microbiologist, was used as the diagnostic gold standard. Percentage of clue cells was estimated by evaluating 100 consecutive squamous cells by another microbiologist and reported in the interval of ten. Diagnostic accuracy was derived by 2×2 tables using different cut-off percentage of clue cells.

Results: Of 748 participants, the average age was 29 years with GA of 28 weeks. Around 10% had leukorrhea or pruritus. The prevalence of BV was 18.7% (140/748). Clue cells were detected in 746 slides, which were just presence (n=204), 10% (n=151), 20% (n=142), 30% (n=99), 40% (n=62), 50% (n=30), 60% (n=31), 70% (n=15), and 80% (n=12). Using less than 10% of clue cells as the reference, the detection of clue cells at 20% or more appeared to be the most promising BV diagnostic tool with a sensitivity at 87.1%, specificity at 55.8%, positive predictive value at 31.2%, negative predictive value at 95.0%, diagnostic odd ratio at 8.57, accuracy at 61.6, and Youden index at 0.43.

Conclusion: Detection of clue cells at 20% or more can be an alternative diagnostic tool of BV, especially when other diagnostic methods are not applicable.

Keywords: Bacterial vaginosis, Clue cell, Pregnancy

Received 22 Feb 2020 | Revised 6 Mar 2020 | Accepted 10 Mar 2020