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% CI 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Interpretation: The widely used Ayre's spatula is the least effective device for cervical sampling and should be superseded by extended-tip spatulas for primary screening and investigation of women before and after treatment for cervical intraepithelial neoplasia.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.