CRD summary

This review concluded that thin liquid-based cytology was promising for primary cervical cancer screening in resourced regions, and visual inspection with Lugol iodine could be used in resource-constrained regions. Caution is necessary when comparing the performance estimates for different screening tests, as these were derived from different study populations and the authors excluded many randomised controlled trials.

Authors' objectives

To compare the accuracy of six common cervical screening strategies: visual inspection with acetic acid, with a magnifying device, or with Lugol iodine; human papillomavirus testing with Hybrid Capture 2 assay; conventional Papanicolaou smear; and thin liquid-based cytology.

Searching

PubMed, EMBASE and The Cochrane library were searched to August 2011, without language restrictions, and search terms were reported. The bibliographies of retrieved articles were screened for additional studies.

Study selection

Studies were eligible for inclusion if they assessed visual inspection with acetic acid, with a magnifying device or with Lugol iodine, human papillomavirus testing with Hybrid Capture 2 assay, conventional Papanicolaou smear,or thin liquid-based cytology, as screening tools for cervical cancer, in apparently healthy women with no gynaecological signs or symptoms. Included studies were required to use histology, or a combination of colposcopy and histology (where abnormal colposcopy was confirmed by histology) as the reference standard. The threshold used to define disease was grade CIN2 or worse.

Randomised controlled trials (RCTs) of screening were excluded.

Most of the included studies were conducted in developing countries. Studies were conducted between 1995 and 2007. The age of study participants ranged from 16 to 70 years. Methods of sample collection and transport media varied across studies (details were reported in the article). Almost all included studies used a combination of colposcopy and histology as the reference standard.

Two reviewers independently assessed studies for inclusion.

Assessment of study quality

The methodological quality of included studies was assessed using the 14-item QUADAS tool and the 25-item STARD reporting guideline for diagnostic studies. Studies with a QUADAS score of below 11 or a STARD score of below 18 were excluded.

Two reviewers independently assessed study quality.

Data extraction

Data were extracted on the numbers of true positive, false positive, false negative and true negative test results. These data were used to calculate estimates of sensitivity, specificity and diagnostic odds ratio, with 95% confidence intervals (CIs).

Two reviewers independently extracted data and any disagreements were resolved by consensus. Authors were contacted for missing data as necessary.

Methods of synthesis

For each screening method, a summary receiver operating characteristic (SROC) curve was generated using the Moses and Littenberg model and area under the curve (AUC) was estimated as a measure of overall test performance. Pooled estimates of sensitivity, specificity and diagnostic odds ratio, with 95% confidence intervals, were also calculated. In the absence of evidence of statistical heterogeneity a fixed-effect model was used, otherwise a random-effects model was used.

Between-study heterogeneity was assessed using Χ² and Ι². Where there was high inconsistency (Ι² greater than 75%), meta-regression was used to explore potential sources of heterogeneity. Threshold effect was assessed when necessary.

Results of the review

Fifteen articles (22 studies, 101,299 participants) were included in the review. QUADAS scores ranged from 11 to 13 and STARD scores ranged from 18 to 24.

For visual inspection with acetic acid (20 studies, 99,972 participants), pooled estimate of sensitivity was 77% (95% CI 75 to 78%; Ι²=91.1%) and specificity was 87% (95% CI 83 to 87%; Ι²=99.7%). The pooled DOR was 27.63 (95% CI 13.93 to 37.45; Ι²=95.9%). Regression analysis indicated that the region in which the study was conducted and the sample size significantly affected estimates of test performance.

For magnified visual inspection with acetic acid (three studies, 26,751 participants), pooled estimate of sensitivity was 64% (95% CI 59 to 68%), Ι²=0% and specificity was 85% (95% CI 85 to 86%: Ι²=94.6%). The pooled DOR was 10.35 (95% CI 7.73 to 13.87; Ι²=49.3%). No regression analysis was conducted.

For visual inspection with Lugol iodine (15 studies, 81,706 participants), pooled estimate of sensitivity was 91% (95% CI 89 to 92%; Ι²=90.5%) and specificity was 85% (95% CI 85 to 86%; Ι²=99.7%). The pooled DOR was 60.83 (95% CI 32.85 to 112.67; Ι²=92.3%). Regression analysis indicated that the region in which the study was conducted significantly affected estimates of test performance.

For human papillomavirus testing with Hybrid Capture 2 assay (seven studies, 23,628 participants), pooled estimate of sensitivity was 74% (95% CI 69 to 78%; Ι²=87.7%) and specificity was 92% (95% CI 92 to 93%; Ι²=98.2%). The pooled DOR was 33.26 (95% CI 19.38 to 57.08; Ι²=81.9%). No regression analysis was conducted.

For conventional Papanicolaou smear (seven studies, 43,993 participants), pooled estimate of sensitivity was 59% (95% CI 56 to 62%; Ι²=94.9%) and specificity was 94% (95% CI 94 to 94%; Ι²=94.9%). The pooled DOR was 22.49 (95% CI 11.16 to 45.31; Ι²=92.5%). No regression analysis was conducted.

For thin liquid-based cytology (three studies, 4,719 participants), pooled estimate of sensitivity was 88% (95% CI 83 to 91%; Ι²=67.1%) and specificity was 88% (95% CI 87 to 89%; Ι²=99.5%). The pooled DOR was 51.56 (95% CI 11.30 to 235.18; Ι²=92.5%). No regression analysis was conducted.

Authors' conclusions

This meta-analysis suggested that thin liquid-based cytology was a promising method for primary cervical cancer screening in resourced regions, and visual inspection with Lugol iodine might be a good choice in resource-constrained regions. All methods were assessed as single test and combining tests may improve performance.

CRD commentary

The review addressed a clearly stated research objective and defined appropriate inclusion criteria. Several sources were searched for relevant studies and no language restrictions were applied, which decreased the likelihood of missed studies. Measures to minimise error and bias were applied throughout the review process. The methodological quality of included studies was assessed, but this assessment included the use of a reporting guideline as a method to assess quality and reported only summary scores. It was not possible to adequately assess the potential effects of components of methodological quality upon estimates of test accuracy, but this may be less of an issue in this review as most studies met almost all quality criteria. A minimum quality score was used as part of the inclusion criteria for the review and resulted in the exclusion of five studies; inclusion of all relevant studies and full exploration of the effects of study quality may have been a more informative approach.

The meta-analytic methods used were not those recommended and a bivariate or hierarchical modelling approach may have been more appropriate for this data set. The authors' conclusions appear reasonable, but caution is necessary when comparing the performance estimates for different screening tests, as these were derived from different study populations (no study compared more than one screening test). It should be noted that the lack of comparative evidence arose primarily from the authors' decision to exclude randomised controlled trials (166 potentially relevant studies).

Implications of the review for practice and research

The authors did not specify any recommendations for research or practice.

Funding

Not stated.

Bibliographic details

Chen C, Yang Z, Li Z, Li L. Accuracy of several cervical screening strategies for early detection of cervical cancer: a meta-analysis. International Journal of Gynecological Cancer 2012; 22(6): 908-921. [PubMed: 22672987]

Original Paper URL

http://journals.lww.com/ijgc/Fulltext/2012/07000/Accuracy_of_Several_Cervical_Screening_Strategies.3.aspx

Indexing Status

Subject indexing assigned by NLM

MeSH

Acetic Acid /diagnostic use; Cervical Intraepithelial Neoplasia /diagnosis; Coloring Agents /diagnostic use; Cross-Sectional Studies; Female; Humans; Iodides /diagnostic use; Mass Screening; Papanicolaou Test; Papillomaviridae /isolation & purification; ROC Curve; Regression Analysis; Uterine Cervical Neoplasms /diagnosis; Vaginal Smears

AccessionNumber

12012035875

Database entry date

12/02/2013

Record Status

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.