Question

What is the reproducibility of the Nugent’s Gram stain scoring system for the diagnosis of bacterial vaginosis (BV) at three different centers in Turkey? 

Design
Gram stained vaginal smears prepared and evaluated using the Nugent scoring system at a laboratory in Turkey were scored blindly by two other independent evaluators and the results from each center were compared.

Participants
Specimens were collected from 372 family planning clients who attended two family planning clinics in Trabzon, Turkey.  Slides that were reported as unsatisfactory by any of the observers were excluded.  Three hundred one (81%) of 372 slides were judged satisfactory by all three evaluators.  

Description of Tests and Diagnostic Standard
One smear taken from the vaginal wall using a cotton swab was prepared from each client by rolling the swab over a clean glass slide, air-drying it, and fixing with methanol.  The slides were transported to the microbiology laboratory, Gram stained, and evaluated by a microbiologist with previous experience in using the Nugent scoring system.  The same slides were transported to two other laboratories and evaluated by microbiologists who were experienced in the assay.  A score of 0 to 3 was considered normal, 4 to 6 was intermediate, and 7 to 10 was BV positive.  

Main Outcome Measures
Nugent scores reported by each evaluator were compared by Spearman correlation co-efficients.  In addition, Kappa statistics were computed on results categorized as BV-negative for scores between 0 and 6 and BV-positive for scores between 7 and 10. 

Main Results
The Nugent Gram stain score results for the 301 slides obtained by each evaluator are shown in the table.   The Spearman rank correlation coefficients for the results of each evaluator compared to the other were 0.75 for evaluators 1 and 2, 0.76 for evaluators 2 and 3, and 0.84 for evaluators 1 and 3.  All correlation coefficients were statistically significant.  When scores of 0 to 6 were classified as BV-negative and 7 to 10 as BV-positive, the percent agreement between the evaluators was 81.1% for evaluators 1 and 2, 79.5% for evaluators 2 and 3, and 86.9% for evaluators 1 and 3.  When the group of slides with scores of 4 to 6 (intermediate) was omitted and analyses conducted again, the calculated coefficients did not differ significantly.  

Nugent Gram stain score results obtained by 3 independent evaluators for the same 301 vaginal smears.

Authors’ Conclusions
The interpretation of Gram-stained vaginal smears according to the Nugent scoring system had high reliability in the study group.  However, it is possible that inclusion of lower-quality slides might have altered the results, and could have decreased the degree of agreement.  

Source of funding:  The German Technical Collaboration Agency (GTZ).

For correspondence:  Pinar Zarakolu, Hacettepe University School of Medicine, Department of Internal Medicine, Section of Infectious Diseases, Ankara, Turkey.  E-mail address:  zarakolu@hacettepe.edu.tr.