HPV self-sampling as an alternative strategy in non-attenders for cervical screening - a randomised controlled trial.

backgroundA randomised trial to ascertain whether women who do not attend for cervical screening are more likely to respond to the opportunity to collect a self-sample for human papillomavirus (HPV) testing, or to a further invitation to attend for cervical screening.

methodsThe study was carried out in a Primary Care Trust (PCT) in London between June 2009 and December 2009. In total, 3000 women were randomly selected from persistent non-responders (i.e., who had not responded to at least two invitations to attend for screening). The women were randomised on a 1 : 1 basis to either receive an HPV self-sampling kit or a further invitation to attend for cervical cytology. The main outcome measures were (1) percentage of women attending for cervical cytology compared with those returning a self-sample HPV test or attending for cytology subsequent to receiving the kit and (2) percentage of those testing positive for HPV who attended further investigation.

resultsThe total response in the self-sampling group for screening was 10.2%. Of the 1500 women in the control group sent a further invitation for cervical screening, 4.5% attended for cytology screening. This difference is highly statistically significant (P<0.0001). Of the 8 women who tested positive for HPV, 7 attended for a cervical smear and had a concurrent colposcopy. Three of these (43%) had high-grade disease (defined as CIN 2+), with one found to have an invasive cancer (stage 1b) and one CIN 3.

conclusionsThe value of this intervention relies on the detection of high-grade CIN and early stage cancer with a good prognosis. The relatively high yield of abnormalities found is consistent with that expected among a hard to reach and relatively high-risk group of women. Our study suggests that self-sampling could increase participation among non-responders in England, but further work is needed to ascertain whether the low response rate seen here is likely to be representative of the rest of the country. Other studies are needed to investigate the response to self-sampling in different demographic and geographic settings.