Six studies, involving a total of 28,420 assessments, were included in the review.
The pooled sensitivity estimate for the ASCUS+ threshold was 64.9% (95% confidence interval, CI: 50.7, 79.1). Five studies contributed to the LSIL+ and HSIL+ estimates, which were 72.6% (95% CI: 60.0, 85.2) and 85.7% (95% CI: 77.8, 93.6), respectively. A regression analysis using data from 4 studies showed a positive relationship between sensitivity and duration of RPS, with an increase of 0.26% (95% CI: 0.09, 0.42) for every second spent screening above 30 seconds. Sensitivity was also influenced by the severity of cytologic abnormality, with a 23.2% (CI: 11.1, 35.2) increase in sensitivity at the HSIL+ threshold compared with the ASCUS+ threshold. Workload had a negative effect on sensitivity: each additional slide per session decreased sensitivity by 0.68%. The mode of slide movement did not affect sensitivity. Data from one study suggested that the level of experience of the cytotechnologists had no effect on sensitivity at the HSIL+ threshold, but that at the ASCUS+ threshold, greater experience led to greater sensitivity.
The pooled specificity estimate (at the ASCUS+ threshold) was 95.0% (95% CI: 92.8, 97.3). However, the false-positive rate of one study was considered to be too high, and with the exclusion of this study specificity was 96.8% (95% CI: 95.8, 97.8).
Five studies were pooled to give a PPV of 60.4% (95% CI: 49.6, 71.2). One study was excluded from this meta-analysis as it had a very low PPV of 3.0% and a prevalence of cytologic abnormality of only 0.4% (compared with a range of 2.7 to 4.7% in the other studies). The pooled NPV was 97.4% (95% CI: 96.2, 98.5).
Additional positives detected by RPS.
Half of the studies had abnormal smears that were detected at RPS, but not at full screening. The pooled proportion of additional positives detected by RPS was 2.9% (95% CI: 0.0, 5.8).