Swede Score can predict high-grade cervical intraepithelial neoplasia without previous cytology in colposcopies with transformation zone types 1 or 2

Abstract

Objective

To evaluate the Swede score for the prediction of high-grade cervical lesions in women undergoing colposcopy without cytology results. This simulates the upcoming scenario in Brazil of HPV-based screening for women testing positive for HPV 16/18.

Methods

This was a partial analysis of an ongoing cross-sectional study that prospectively included 86 women referred for colposcopy due to cytological abnormalities to a public referral center in Rio de Janeiro, Brazil, between September 2023 and June 2025. All women went to colposcopy evaluation. Colposcopy was performed by an examiner blinded to the referral cytology. For this analysis, we included all women with a fully visible squamocolumnar junction (transformation zone types 1 or 2). The gold standard was the histopathological report. In cases where no biopsy or excisional treatment was performed, classification relied on final colposcopic impression and cytology. Diagnostic performance (sensitivity, specificity, predictive values, and likelihood ratios) was calculated using different Swede score cutoffs. Post-test probabilities were calculated assuming a 20% pre-test probability of CIN2+.

Results

Prevalence of CIN2 + was 33.7%. Swede score ≥ 8 predicted CIN2 + with a positive predictive value (PPV) of 100% and a likelihood ratio of 55.17. Post-test probability in a HPV 16/18-positive scenario without cytology in this threshold was estimated at 93%.

Conclusion

The Swede score cutoff ≥ 8 in women who tested positive for HPV 16/18 and had fully visible transformation zones can be an accurate method to predict that they have CIN2+. These results could support the decision to proceed with immediate treatment without prior cytology or biopsy.

Trial registration

Not applicable. This study does not report the results of a health care intervention.

Background

Colposcopy is a triage test performed after screening to detect cervical dysplasia and early cancer. In Brazil, it is currently recommended after a positive cytology, but the updated national guidelines foresee its use after detection of high-risk HPV. According to these recommendations, women who test positive for HPV 16/18 will be referred directly to colposcopy, while those positive for other high-risk types will undergo reflex cytology before referral. In this upcoming scenario, it is estimated that approximately 55% of women will be referred to colposcopy without reflex cytology [1], removing this important element that traditionally supports colposcopic decision-making regarding immediate treatment or biopsy.

Colposcopy is known for its subjectivity, with significant intra and interobserver variability [2]. The colposcopic terminology recommended by the International Federation of Cervical Pathology and Colposcopy (IFCPC) provides criteria to differentiate CIN2 + from benign or inflammatory changes [3].

Despite its widespread use, IFCPC classification is not the only one available. In 1985, Reid and Scalzi introduced a colposcopic index based on four lesion characteristics: margin, acetowhiteness, vascular pattern, and iodine staining [4]. Later, Strander et al. added lesion size as a fifth variable and developed the Swede score [5]. Reid index ranges from 0 to 8 points, while the Swede score ranges from 0 to 10. In the original study by Strander et al., a cutoff ≥ 5 identified all high-grade lesion (cervical intraepithelial neoplasia grade 2 or worse – CIN2+) , and a cutoff ≥ 8 resulted in only 10% of overtreatment using immediate treatment strategy. These scoring systems help reduce subjectivity and may improve colposcopic accuracy [6, 7].

Immediate treatment is based on ablative or excisional techniques, with the advantage of reducing clinical visits and diagnostic steps, thereby decreasing the risk of follow-up losses. Costs related to unnecessary biopsies, clinical appointments, and patient travel are also reduced. These factors are especially relevant in remote or underserved areas [8].

Because there is no histological confirmation before treatment, one of the risks associated with this approach is overtreatment, whose consequences vary depending on the therapeutic procedure. Furthermore, when excisional treatment is used, some patients can have excessive bleeding and adverse perinatal outcomes [8].

Overtreatment occurs when histological analysis of the excised specimen fails to reveal a CIN2+. A meta-analysis estimated that 11.6% of women undergoing excisional treatment without prior biopsy had negative histology. In this metanalysis, this proportion of overtreatment was found in women with high-grade squamous intraepithelial lesions (HSIL) or atypical squamous cells that cannot exclude high-grade lesions (ASC-H), and major colposcopic findings [4]. A more recent meta-analysis comparing immediate treatment to biopsy-guided approaches revealed overtreatment rates of 12% among women with HSIL cytology and 24% among those with ASC-H [9].

In the current cytology-based screening program in Brazil, immediate treatment is recommended for women aged 25 years or older who present ASC-H or HSIL cytology, major abnormal colposcopic findings, and a fully visible squamocolumnar junction (SCJ), corresponding to transformation zone (TZ) types 1 or 2 [10]. A previous study conducted at our center reported an overtreatment rate of 8.44% using these criteria [11].

Guidelines from the American Society for Colposcopy and Cervical Pathology (ASCCP) support immediate treatment in HPV-based screening programs when the estimated risk of CIN2 + exceeds 60%. This happens in women who tested positive for HPV 16 or 18 and cytology with HSIL result [2] but cytology will not be available in these women according to the upcoming national guidelines in Brazil [12]. In women positive for HPV 16/18 , the estimated risk of CIN2 + is approximately 20% [13], but may range from 24% to 63% [14] considering only HPV status.

In the upcoming scenario in Brazil, establishing criteria for immediate treatment without cytology is crucial. However, we found no direct evidence to support this practice in the absence of a cytological report. Although no studies have directly reported overtreatment based solely on HPV status and colposcopic impression, the meta-analysis by Silver et al. demonstrated a 76% risk (95% CI: 66–85%) of CIN2 + in women with a high-grade colposcopic impression and HPV16/18 positivity [14]. This finding suggests a potential overtreatment rate ranging from 15% to 44% in this population when treatment is performed without a prior biopsy.

To support the upcoming update of Brazil’s cervical cancer screening guidelines, we analysed the diagnostic performance of the Swede score to detect CIN2 + in women with type 1–2 TZ with colposcopists blinded to previous cytology.

Methods

These are partial results from an ongoing cross-sectional study with prospective recruitment initiated in 2023 to evaluate the diagnostic performance of the Swede Score. The study has been conducted at a public referral center in Rio de Janeiro, Brazil, which has extensive experience in diagnosing and treating precancerous cervical lesions. This was approved by the Research Ethics Committee of the Instituto Nacional de Saúde da Mulher, da Criança e do Adolescente Fernandes Figueira (IFF/Fiocruz), Rio de Janeiro, Brazil (CAAE: 72835523.1.0000.5269; consubstantiated opinion number: 6.316.309).

Eligible participants were women referred for colposcopy because of abnormal cytology, in accordance with the current Brazilian screening program [10]. No HPV testing was performed, as the study was designed to model the future scenario in which women testing positive for HPV 16/18 will be directly referred to colposcopy without reflex cytology. Thus, HPV positivity in this analysis was assumed for modeling purposes, not measured. All participants provided written informed consent; however, for this analysis, we included only those with fully visible SCJ (TZ types 1 or 2) and had confirmed final diagnoses until June 2025. We also excluded pregnant women, previously biopsied or treated women, and those with clinically suspected invasive cancer (presence of gross cervical tumors, irregular friable lesions, necrosis, spontaneous or contact bleeding, or distortion of cervical anatomy).

Colposcopic examinations were performed by experienced examiners who were blinded to the referral cytology results. All colposcopists are from the same specialized unit, both trained under identical institutional protocols and using standardized evaluation criteria. The Swede score was calculated using translated parameters from the original study [5].

After that, a second examiner, who was aware of the cytology results, recorded the colposcopic impression in accordance with IFCPC classification [3]. After reaching consensus, biopsies, treatment, or follow-up were recommended accordingly.

Histology, obtained via colposcopic guided biopsy and/or excisional procedure, was the gold standard for diagnosis. Not all participants had histology results available; for those who did not undergo biopsy or excisional treatment, the final diagnosis relied on findings from colposcopy and referral cytology. For glandular atypia, endocervical cytology was also considered. All cytopathological and histopathological evaluations were conducted by expert pathologists in cervical pathology who were aware of the previous tests. The presence of disease was defined by a histological diagnosis of CIN2+; absence of disease was established by a histological diagnosis of CIN1 (cervical intraepithelial neoplasia grade 1) or less significant findings (such as cervicitis, endocervical polyps, or squamous metaplasia), or by the presence of minor or no abnormalities according to the IFCPC classification in patients with LSIL/ASC-US cytology and without biopsy.

Sensitivity, specificity, and positive and negative predictive values (PPV/NPV) were calculated for Swede score thresholds ≥ 5, which correspond to the threshold associated with CIN2+ as identified in the Strander et al. study [5].

Because predictive values depend on disease prevalence [15] we also calculated positive and negative likelihood ratios (LR + and LR−) and estimated post-test probabilities of CIN2 + based on a conservatively assumed pre-test probability of 20% in women with HPV 16/18.

All data were managed using Microsoft Access 365 and analyzed with EpiInfo v.7.2.6.0. Diagnostic performance was calculated using R Studio.

Results

This analysis included 86 women out of the 180 women included so far, as shown in Figure 1. Baseline characteristics and final diagnoses are presented in Table 1. The median age was 36 years, most of the women had two or fewer children, and had a previous cytology of HSIL or ASC-H or glandular atypia. The prevalence of CIN2 + in this sample was 33.7%.

Fig. 1.

Flowchart of participant

Table 1.

Baseline characteristics and final diagnoses of women included in the partial analysis - women with fully visible squamocolumnar junction (SCJ, transformation zone types 1 or 2)(National Institute of Women`s, Child`s and Adolescent`s Health, of Oswaldo Cruz foundation − 09/21/2023 to 06/10/2025; n = 86)

Characteristics
Age (median; min-max)	36	20–64
Immunosuppression or HIV+ (n; %)	4	4.6
Parity (n; %)
0–2	56	65.1
> 2	30	34.9
Total	86	100.0
Reference Cytological Diagnoses (n; %)
Atypical squamous cells of undetermined significance (ASC-US)	6	6.9
Low-grade squamous intraepithelial lesion (LSIL)	6	6.9
Atypical glandular cells of undetermined significance (AGC-NOS)	1	1.2
Atypical glandular cells of undetermined significance cannot exclude high-grade disease (AGC-H)	1	1.2
Atypical glandular cells of undetermined significance (AGC-US)	16	18.6
Atypical squamous cells of undetermined significance, cannot exclude high-grade disease (ASC-H)	26	30.2
High-grade squamous intraepithelial lesion (HSIL)	28	32.6
HSIL with suspected microinvasion	1	1.2
No result	1	1.2
Total	86	100.0
Colposcopic findings (IFCPC*) (n; %)
Normal findings	35	40.7
Minor abnormalities	21	24.4
Major abnormalities	27	31.4
Miscellanea (endocervical polyps)	3	3.5
Total	86	100.0
Transformation Zone (TZ) type (n; %)
Type 1	38	44.2
Type 2	48	55.8
Total	86	100.0
Final Diagnoses* (n; %)
Microinvasive squamous carcinoma (FIGO IA1 stage)	3	3.5
High-grade squamous intraepithelial lesion (CIN 2 or 3)	26	30.2
Low-grade squamous intraepithelial lesion (CIN 1)	17	19.8
Negative for intraepithelial or malignancy (includes endocervical polyps)	40	46.5
Total	86	100.0

*IFCPC International Federation of Cervical Pathology and Colposcopy, Final Diagnosis based on histology or, when biopsy or excisional procedures were not performed, on colposcopic assessment and, in cases of glandular atypia, also taking endocervical cytology into account, FIGO International Federation of Gynecology and Obstetrics, CIN cervical intraepithelial neoplasia

Table 2 shows the number of CIN2 + or < CIN 2 for each Swede score cutoff ≥ 5 or more.

Table 2.

Tabulation of Swede score 5 or more by final diagnosis (reference standard)

Swede score/Colposcopic Findings	Reference standard		Total
	CIN2+	< CIN2
Swede ≥ 5	28	7	35
Swede < 5	1	50	51
Swede ≥ 6	25	4	29
Swede < 6	4	53	57
Swede ≥ 7	22	2	24
Swede < 7	7	55	62
Swede ≥ 8	16	0	16
Swede < 8	13	57	70
Swede ≥ 9	10	0	10
Swede < 9	19	57	76

CIN2+ Cervical Intraepithelial Neoplasia grade 2 or worse

Diagnostic performance was assessed using the Swede score cutoff ≥ 5 or more as reported in Table3. To prevent division by zero during likelihood ratio calculations, any cells with a value of zero in the contingency table (Table 2) were substituted with a value of 1. This adjustment was applied solely to enable likehood ratio calculations and did not affect the overall number of cases without disease or the clinical interpretation of results.

Table 3.

Diagnostic performance of the Swede score for CIN2+

Swede score or Colposcopic Findings	Sens. (CI95%)	Spec. (CI95%)	PPV (CI95%)	NPV (CI95%)	LR+ (CI95%)	LR− (CI95%)
Swede ≥ 5	0.97 (0.82, 1.00)	0.88 (0.76, 0.95)	0.80 (0.63, 0.92)	0.98 (0.90, 1.00)	7.86 (3.92, 15.79)	0.04 (0.01, 0.27)
Swede ≥ 6	0.86 (0.68, 0.96)	0.93 (0.83, 0.98)	0.86 (0.68, 0.96)	0.93 (0.83, 0.98)	12.28 (4.72, 31.96)	0.15 (0.06, 0.37)
Swede ≥ 7	0.76 (0.56, 0.90)	0.96 (0.88, 1.00)	0.92 (0.73, 0.99)	0.89 (0.78, 0.95)	21.62 (5.46, 85.66)	0.25 (0.13, 0.48)
Swede ≥ 8	0.55 (0.36, 0.74)	1.00 (0.94, 1.00)	1.00 (0.79, 1.00)	0.81 (0.70, 0.90)	31.45 (4.38, 225.55)*	0.46 (0.30, 0.68)*
Swede ≥ 9	0.34 (0.18, 0.54)	1.00 (0.94, 1.00)	1.00 (0.69, 1.00)	0.75 (0.64, 0.84)	19.66 (2.64, 146.17)*	0.67 (0.51, 0.87)*

Sens. sensitivity, Spec. specificity, PPV positive predictive value, NPV negative predictive value, LR+ positive likelihood ratio, LR- negative likelihood ratio, CI95% Exact 95% confidence interval; all calculations performed by R Studio

*For LR calculation, zero cells in Table 2 were set to 1 without changing the total for disease absent

As shown in Table 3, the Swede score demonstrated higher sensitivity and lower specificity at lower cutoff points. As expected, sensitivity decreased as the cutoff score increased, while specificity increased accordingly. Notably, with a cutoff ≥ 7, the PPV for CIN 2+ exceeded 90% (95%CI 81–100).

The prevalence of CIN2 + in this sample was higher than the assumed 20% in women testing positive for HPV 16/18. Thus, Table 4 presents the estimated post-test probabilities using positive likelihood ratios (LR+) from Table 3, assuming a pre-test probability of 20% and a Swede score ≥ 7.

Table 4.

Estimated probability of CIN2+ calculation using the Swede score ≥7

Swede score/Colposcopic Findings	LR+	Pre-test Prob.	Pre-test Odds	Post-test Odds	Post-test Prob.
Swede ≥7	21.61	0.20	0.25	5.40	84%
Swede ≥8	31.45	0.20	0.25	7,86	89%
Swede ≥9	19.66	0.20	0.25	4,915	83%

LR+: positive likelihood ratio from Table 3; Prob.: probability; CIN2+: Cervical Intraepithelial Neoplasia grade 2 or worse

In this hypothetical analysis, a Swede score of ≥ 8 was associated with a probability of CIN2 + near 90% in women with a 20% pre-test probability (i.e., HPV16/18-positive).

Discussion

A Swede score ≥ 7 predicted CIN2 + in more than 90% (95% CI: 81–100%) of women in this sample, even without cytological information. However, predictive values depend on disease prevalence and may not be stable across populations. For this reason, likelihood ratios (LRs), which are based on sensitivity and specificity, were used as a more robust metric for diagnostic interpretation.

In our modeled scenario, assuming a pre-test probability of 20%, the nearest Swede score cutoff for predicting CIN2 + with ≥ 90% certainty was 8 (Table 4). This 20% probability was defined according to Hashim et al. (2020), who reported CIN2 + prevalence between 18% and 25% among HPV 16/18-positive women, supporting its use as a realistic baseline estimate for this subgroup.

Strander et al. (2005) had previously shown that a Swede score ≥ 8 predicted CIN2 + in more than 90% of women referred for colposcopy with fully visible SCJ, although their inclusion of only histologically confirmed cases likely inflated disease prevalence (44%) and diagnostic accuracy. Similarly, Silver et al. (2020) demonstrated high CIN2 + risks in biopsy-based cohorts, which may also have overestimated accuracy. In contrast, our study included women without histological confirmation when the SCJ was fully visible and colposcopy appeared reliable, which minimizes selection bias. Although some false negatives are possible, this risk is low since all exams were performed by two experienced colposcopists using standardized criteria. Clinical decisions were made only after cytology became available, improving the diagnostic accuracy of the final classification.

Among all participants, 49 (56%) had histological confirmation, and 37 (43%) were classified based on colposcopic and cytologic findings, as recommended by national guidelines. This method can introduce incorporation bias because colposcopy informs both the index test and reference standard. However, this choice prevented selection bias by allowing the inclusion of women with minor or negative findings, who are often excluded in accuracy studies.

Unlike previous studies, including those by Strander and Silver, we ensured that colposcopists were blinded to cytology when applying the Swede score. This design better reflects the upcoming HPV-based screening scenario in Brazil, where women positive for HPV 16/18 will undergo colposcopy without reflex cytology. In such settings, immediate treatment may be considered the most practical approach.

While the IFCPC classification is qualitative and descriptive, the Swede score offers an objective numeric system with defined cut-offs. This enhances reproducibility, reduces subjectivity, and may guide immediate treatment (“see-and-treat”) decisions for HPV 16/18-positive women.

Although our sample size is not yet complete, resulting in wide confidence intervals, our findings are in accordance with those reported by Silver et al., showing a 76% risk of CIN2 + in HPV 16/18-positive women with major colposcopic findings. These results suggest that treatment without biopsy could be justified when the transformation zone is fully visible and the Swede score is ≥ 8, even in the absence of cytology.

The use of colposcopy to define the reference standard may introduce incorporation bias, potentially leading to an overestimation of accuracy measures. However, in this study, the reference standard was established after the Swede score had been assigned, by a second colposcopist who reviewed the findings together with the first examiner and considered the cytology report. These procedures help ensure that any incorporation bias was minimal and that the results reflect a real-world scenario for colposcopic diagnosis.

Another limitation is that, to allow the calculation of likelihood ratios, zero counts in the contingency table (Table 2) were substituted with a value of 1 to prevent division-by-zero errors. Post-test probabilities derived from these adjusted ratios should be interpreted with caution, as they are matematically influenced by the correction applied to zero cells.

These limitations limit the generalizability of our findings. The interpretation of our findings should not be considered definitive. Nevertheless, the likelihood ratios allow the use of our results to contexts with different prevalences. As Brazil transitions to HPV-based screening, new national data will soon refine prevalence estimates of HPV 16/18-positive women and provide stronger evidence for validating the Swede score in true screening populations.

Conclusion

In this study, a Swede score ≥ 8 predicted CIN2 + in over 90% of women with TZ 1–2, even when cytology results were unknown. This finding corresponds to an overtreatment risk of less than 10%. These results could support immediate treatment in HPV 16/18–positive women without the need for prior cytology or biopsy.

Abbreviations

CIN2+

Cervical intraepithelial neoplasia grade 2 or worse

WHO

World Health Organization

HSIL

High-grade squamous intraepithelial lesions

ASC

Atypical squamous cells that cannot exclude high-grade lesions

SCJ

Squamocolumnar junction

ASCCP

American Society for Colposcopy and Cervical Pathology

CIN1

Cervical intraepithelial neoplasia grade 1

PPV

Positive predictive value

NPV

Negative predictive value

LR +

Positive likelihood ratio

LR -

Negative likelihood ratio

CI

Confidence interval

Authors’ contributions

BSC conceived the study, performed colposcopies, interpreted the data, and drafted the manuscript. PMJ contributed to data collection, performed colposcopic evaluations, and participated in manuscript revision. CVA supported data management and statistical analysis. ACCC performed and interpreted the histopathological analyses and contributed to the interpretation of diagnostic findings. FBR contributed to data collection, supervised the study, performed colposcopic evaluations, supported data management and statistical analysis, participated in its design and coordination, and revised the final version of the manuscript.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. The study was conducted using the authors and institutional resources.

All authors declare that there are no conflicts of interest.

Data availability

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The original study is registered at the institutional IFF/Fiocruz Ethics Committee (CAAE: 72835523.1.0000.5269) and all procedures were performed under the routine care of women received for diagnostic and treatment of CIN in this institution. All participants provided written informed consent, which included authorization for publication.

Consent for publication

Not applicable. This manuscript does not contain any individual person’s data in any form (including images, videos, or identifiable personal details).

Competing interests

The authors declare no competing interests.