Performance of two colposcopic indices for predicting premalignant cervical lesions

Cervical cancer ranks as the fourth most frequently occurring cancer in women globally and stands as the second most prevalent disease affecting women in Bangladesh [1]. Cervical cancer is often preceded by a lengthy premalignant stage of cervical intraepithelial neoplasia (CIN). This characteristic allows for early detection and prevent [2]. Commonly utilized screening techniques include visual inspection of the cervix with acetic acid (VIA), Pap smear, and human papillomavirus-DNA testing. Confirmation of the diagnosis then entails a biopsy. Colposcopy serves as a valuable tool for triaging and providing guidance for biopsy procedures. Reid and Scalzi introduced the Reid‘s colposcopic index (RCI) as a means to reduce subjectivity in colposcopic diagnosis, and it has become the commonest scoring system [3, 4]. The RCI is determined through an assessment of the margin of the acetowhite lesion, its color, the presence of atypical vessels, and iodine staining. Studies reported RCI’s high sensitivity and specificity [4]. In 2005, Strander et al. introduced a novel colposcopic scoring system known as the Swede score. This system, built upon the four parameters of the RCI, includes lesion size as an additional variable [5]. The specificity of the Swede score was 95% for the detection of CIN2+ lesions [6].

This cross-sectional study was conducted at the colposcopy clinic of Bangabandhu Sheikh Mujib Medical University (currently, Bangladesh Medical University), from August 2020 to September 2021. The sample size was calculated based on an expected sensitivity of 90%, 5% absolute precision, and a 95% confidence interval, resulting in 270 subjects. A final sample size of 300 was chosen to account for possible non-response.

Adult women, exhibiting various indicators such as positive VIA results, atypical squamous cells of undetermined significance or more severe findings on Pap smear, human papillomavirus DNA positivity, cervix abnormalities, and persistent pervaginal discharge, were selected for participation in the study via purposive sampling. The exclusion criteria included women with evident growth, prior cervical procedures (such as cold coagulation, cryotherapy, or conization), pregnancy, and unsatisfactory colposcopy. Prior to inclusion, all participants provided written informed consent, emphasizing the voluntary nature of their involvement.

Two colposcopic (RCI and Swede scores) tests followed by biopsy were taken in every participant. Biopsies were obtained from abnormal areas via Tischler forceps; in cases without evident lesions, a four-quadrant biopsy was taken from the squamocolumnar junction of the cervix. Hemostasis was ensured, and the speculum was then gently removed. Biopsy samples were preserved in 10% formalin and sent to the Department of Pathology for histopathological examination. For the RCI, four features were scored from 0 to 2: acetowhiteness, margins, vascular pattern, and iodine staining. A score ≥5 indicated high-grade CIN. The Swede score evaluates the same parameters and additionally includes lesion size, each graded from 0 to 2. A score ≥5 suggests CIN2+; ≥8 was used for "see and treat".

Analysis of lesion size in relation to histopathology showed a significant trend. Lesions larger than 15 mm were more frequently associated with CIN2+ outcomes (Table 3). This supports the inclusion of lesion size in the Swede scoring system. The agreement between the Reid’s Colposcopic Index and the Swede score was moderate (κ=0.4), indicating some diagnostic overlap, they remain complementary tools rather than interchangeable scoring systems.

Colposcopy remains an essential tool for evaluating cervical premalignant lesions, particularly in low- and middle-income countries, where cytology and HPV testing may have limited availability. The development of colposcopic indices, such as the Reid’s Colposcopic Index (RCI) and Swede score, aims to improve diagnostic objectivity and provide practical guidance for screening and treatment decisions. This study compared the performance of these two indices in predicting high-grade cervical intraepithelial neoplasia (CIN2+) considering histopathology as gold standard. Significant differences were found in their diagnostic profiles.

The four shared features allow for a valid head to head comparison of sensitivity, specificity, predictive value and clinical utility. However, it is acknowledged that the Swede score includes an additional parameter-lesion size-which is a critical predictor of high-grade disease. This makes the Swede score more comprehensive and objective. Moreover, its flexible use of dual thresholds (≥5 for screening and ≥8 for treatment) offers practical advantages, especially in resource-limited settings employing single-visit strategies. Therefore, while the scores are not interchangeable, their comparability remains appropriate for evaluating diagnostic accuracy, as was done in this study.

Our findings of high specificity but low sensitivity at a cutoff value of 5 are consistent with previous studies by Durdi et al. and Mousavi et al., which reported that RCI is effective in ruling out high-grade lesions but may miss a significant proportion of CIN2+ cases [7, 8]. The high specificity of the RCI makes it suitable for confirming high-grade lesions, thereby minimizing overtreatment in “see and treat” programmes. However, its low sensitivity limits its effectiveness as a primary screening tool for this disease. Kushwah et al. and Hong et al. also observed similar trends, emphasizing the risk of underdiagnosis when relying solely on RCI in high-prevalence settings [9, 10].

In contrast, our findings of higher sensitivity but lower specificity of Swede score at a cutoff of 5 aligns with the findings of Strander et al. and Nessa et al., who showed that the Swede score is a better screening tool because of its ability to detect more CIN2+ lesions, reducing the likelihood of missed diagnoses [5, 11]. However, its lower specificity may increase the risk of overtreatment, particularly in resource-limited settings, where follow-up can be challenging. Ranga et al. similarly concluded that while the Swede score improves sensitivity, it must be applied judiciously to avoid unnecessary interventions [12].

When the Swede score cutoff was increased to 8, specificity improved markedly (92.3%) with an associated reduction in sensitivity (11.1%), indicating its value in “see and treat” scenarios. This dual cutoff approach has also been supported by Suwanthananon et al., who recommended using cutoffs of 5 and 8 for screening and treatment, respectively, to balance sensitivity and specificity [13]. Such flexibility offers practical advantages in diverse clinical contexts, particularly in settings where access to repeated follow-up visits is limited.

An important finding of this study was the significant association between lesion size and the likelihood of CIN2+ disease, which supports the inclusion of lesion size in the Swede score. Kierkegaard et al. previously demonstrated that larger lesions were more likely to be associated with high-grade histopathology, and our results are in agreement with this evidence [14]. By incorporating lesion size, the Swede score reduces interobserver variability and enhances diagnostic reproducibility. Similar findings by Ranga et al. suggest that while these two indices share common diagnostic features, they are not interchangeable in clinical practice [12]. Instead, they complement as indicated by a moderate agreement. The RCI may serve as a confirmatory tool owing to its high specificity, whereas the Swede score can be prioritized for initial screening.

The current this study supports the complementary use of these two indices. Although the RCI remains reliable for confirming high-grade disease, the Swede score offers superior versatility for screening and treatment decisions. The implementation of these indices in combination, tailored to local resource availability, may enhance diagnostic precision and improve cervical cancer prevention outcomes. It is important to note the lack of generalizability of our findings because exclusively referred cases were included in this study at Colposcopy Clinic of a tertiary care hospital.

Conclusion

The Swede score, with its dual cutoff approach, provides greater flexibility than the Reid’s Colposcopic Index for both screening and immediate treatment strategies. While RCI remains valuable for its high specificity, the Swede score offers superior adaptability in resource-limited settings.