Efficacy of cervical-smear collection devices: a systematic review and meta-analysis

Summary

Background

Few randomised controlled trials have sufficient power to show clear advantages of different designs of cervical-smear collection devices. We studied by systematic review whether the design of cervical-smear devices affects rates of inadequate smears and detection of disease and whether the presence of endocervical cells in the smear affects detection of disease.

Methods

We sought relevant randomised controlled trials by computer literature review by MEDLINE backed up by a manual search of 16 journals. Each trial was classified according to methodological quality criteria. Odds ratios were calculated where data allowed.

Findings

34 randomised controlled trials investigating cervical Papanicolaou smear collection devices were identified. All 34 trials compared the ability of devices to collect endocervical cells, and 19 compared the ability of devices to detect dyskaryosis. Meta-analyses showed that compared with other collection devices, the Ayre's spatula is an ineffective device for collecting endocervical cells (for example, odds ratio for comparison of extended-tip spatulas vs Ayre's spatula 2·25 [95% Cl 2·06–2·44]) and also gives a lower yield of dyskaryosis (odds ratio for comparison of extended-tip spatulas vs Ayre's spatula 1·21 [1·20–1·33]). Devices that effectively collect endocervical cells also detect a higher proportion of abnormal cytology than those that do not.

Interpretation

The widely used Ayre's spatula is the least effective device for cervical sampling and should be superseded by extended-tip spatulas for primary screening and investigation of women before and after treatment for cervical intraepithelial neoplasia. The presence of endocervical cells is a valid and convenient surrogate for the ability to detect dyskaryosis.

Introduction

Cervical screening is effective in decreasing the incidence and mortality of invasive disease where comprehensive programmes exist.1, 2 The overall effectiveness of such screening programmes depends on wide coverage of the target population, quality of smear collection, and the management of abnormal cytology. The sensitivity of cervical screening is limited to some degree by sampling error, with reported false-negative rates of 55%³ to 1·5%.⁴ The false-negative rate may be related partly to the adequate sampling of exfoliated cells and their interpretation on smears and partly to variability in study design. The sampling of smears depends on the expertise of the operator and the sampling device used.

There has been a tendency for the adequacy of a cervical smear to be judged by the presence or absence of endocervical cells, because this is a common and easily measured endpoint. The idea that the presence of these cells reflects the adequacy of the smear is anatomically plausible, since it suggests that the transformation zone, from which premalignant change arises, has been sampled. The Working Party of the Royal College of Pathologists, the British Society for Clinical Cytology, and the NHS Cervical Screening Programme⁵ recommended that information on the presence of metaplastic and endocervical cells should be documented since these cells provide evidence of probable transformation-zone sampling. Furthermore, the quality assurance guidelines for the cervical screening programme for the UK⁶ recommend that more than 80% of smears should contain such representative cells.

Direct confirmation that the yield of endocervical cells correlates with the detection of dyskaryosis has, however, been sought only within single and in many cases underpowered studies. We have used both structured review and meta-analysis of randomised controlled trials and large observational studies to examine whether the presence of endocervical cells is a surrogate for the quality of cervical sampling in terms of detection of disease and adequate cervical smears. A previous meta-analysis⁷ focused solely on detection of disease as an endpoint.

Inadequate smears have been characterised as: having insufficient cellularity, being poorly fixed, being contaminated by blood or inflammatory cells; or being spread too thickly. Such cervical smears have to be repeated to avoid false-negative reports.⁸ Many women who develop cervical cancer have had inadequate cytology on previous smears.⁸ If the design of cervical-smear collection devices adversely affects rates of inadequate smears, this factor may reduce not only the effectiveness of screening but also the cost-effectiveness and psychological cost because of the need for repeat smears. Extended-tip spatulas have the theoretical advantage of more efficient cell sampling from the proximal part of the transformation zone.

There are many different designs of cervical-smear collection devices; the commonly used devices are illustrated in figure 1. Spatulas are simple to use and cheap to manufacture. The traditional Ayre's spatula has a broad head, and sampling of the cervical canal is therefore impossible in many cases. Modified spatulas, such as the Aylesbury, Rolon, Rocket, Multispatula, Armocervical, and Cytopick, have a modified head to improve endocervical sampling. Wooden spatulas have the theoretical disadvantage compared with plastic spatulas that cells may be trapped within their fabric. Endocervical sampling devices, such as the cotton swab and Cytobrush, are designed to be used in combination with a spatula. Both have the disadvantage of provoking bleeding from columnar epithelium, which can result in contaminated smears. Several devices, such as the Cervexbrush, and Baynebrush, are designed to improve both ectocervical and endocervical sampling. These devices are more expensive than spatulas.