3.1 The key clinical outcomes for Vision Amniotic Leak Detector (ALD) presented in the decision problem were:
incidence of speculum examinations
diagnostic accuracy for the detection of amniotic fluid leak and premature rupture of membranes
identification of vaginal infection
incidence of speculum-associated cross-infection
bed utilisation and staff time
device-related adverse events.
3.2 The clinical evidence for the Vision ALD was based on 3 comparative diagnostic studies, evaluating the non-inferiority of the Vision ALD to speculum examination. In 2 of the studies (Bornstein et al. 2006 and Bornstein et al. 2009), the comparator was speculum examination or a positive nitrazine (pH) and fern test. In addition, the External Assessment Centre extrapolated unpublished data from these studies (supplied by the sponsor) to provide a comparison against speculum examination alone. As part of this analysis, the External Assessment Centre redesignated false positives as true positives if infection was detected or if ruptured membranes were subsequently diagnosed. Published data from a study by Mulhair et al. (2009) were also included to calculate overall mean sensitivity, specificity and predictive values for the Vision ALD.
3.3 Bornstein et al. (2006) carried out a prospective, single-centre study evaluating the diagnostic efficacy of the Vision ALD (under an alternative brand name, AL-SENSE) involving 103 pregnant women attending a labour and delivery ward at a hospital in Israel. Women were enrolled in 3 groups: a negative control group of women showing no signs of fluid leakage (n=27), a positive control group (n=42) including 32 women with overt spontaneous prelabour rupture of membranes [PROM]) and 10 with artificially ruptured membranes and an intervention group of women with vaginal wetness of unknown cause (n=34). Women were included in the study if they were aged between 18 and 45 years with pregnancies of at least 20 weeks' gestation. Women were given an AL-SENSE liner to wear for 1 hour and asked to record any colour change. After 10 minutes drying time, the colour was read by the attending midwife or obstetrician and the study coordinator, who were all blinded to the participant-recorded result. Another clinician, blinded to the results of the AL-SENSE test, subsequently made a clinical diagnosis based on a speculum examination to look for pooling of amniotic fluid, or by a nitrazine and fern test if no pooling was visible. The sensitivity and specificity of the AL-SENSE liner were calculated by comparing the clinical diagnosis with the final reading of the liner colour. The sensitivity of the device was 100% in the intervention group (10/10, 95% confidence interval [CI] 69.15 to 100%) and the specificity was 75% (18/24, 95% CI 53.29 to 90.23%). It is important to note that device colour change caused by vaginal infection was classed as a false positive; 4 of the 6 false positives were subsequently found to be caused by bacterial vaginosis. The device correctly diagnosed ruptured membranes in all 42 women in the positive control group. In the negative control group, 23 women were confirmed as having intact membranes and 2 women did in fact have ruptured membranes that were identified by the device. The remaining 2 results in this group were false positives.
3.4 Bornstein et al. (2009) carried out a prospective multicentre trial evaluating the non-inferiority of the Vision ALD (under an alternative brand name, AmniScreen) to a nitrazine swab test (AmnioTest, Pro-Lab Diagnostics), which had a reported sensitivity of 91% and specificity of 73% in detecting amniotic fluid. Women aged between 18 and 45, with pregnancies of 16 weeks' gestation or more, presenting at labour and delivery triage with unexplained vaginal wetness were included in the study. Women were recruited from 2 centres in Israel (n=110, n=179) and 1 in the USA (n=50) making a total of 339 women. Data from 309 women were analysed; 10 women were excluded for protocol violations and 20 did not have a complete set of tests. Women were given an AmniScreen liner to wear until wetness was felt and asked to read the result after 30 minutes drying time. The result was subsequently read by a healthcare provider, blinded to the participant-recorded result. Another healthcare provider, blinded to the AmniScreen results, made a clinical diagnosis using a speculum examination to look for pooling of amniotic fluid, or by a nitrazine and fern test if no pooling was visible. If the AmniScreen result was positive and clinical diagnosis negative, testing for vaginal infection was carried out, followed by further clinical assessment within 48 hours to obtain a final clinical diagnosis. Analysis was carried out by comparing the woman's reading of the device with the clinical diagnosis. The healthcare provider's reading of the device was also compared with the clinical diagnosis. The sensitivity of the test was found to be 95.8% (158/165) and the specificity 86.8% (125/144) based on the comparison between women's readings and final clinical diagnosis. The authors concluded that the sensitivity and specificity of the AmniScreen test were significantly higher (p=0.02 and p<0.0001 respectively) than that of the AmnioTest nitrazine swab. Comparison of the healthcare provider's device readings and the final clinical diagnosis indicated that the sensitivity of AmniScreen was 95.8% and the specificity was 88.1%. Comparison of the healthcare provider's and the women's device readings led the authors to conclude that overall agreement between them was above 90%.
3.5 Mulhair et al. (2009) carried out a prospective single-centre study comparing the diagnostic performance of the Vision ALD (under an alternative brand name, AmnioSense) with speculum examination in identifying amniotic fluid leak. The study involved 157 women, with pregnancies between 18 and 42 weeks' gestation, presenting to an antenatal day unit in the UK with suspected rupture of membranes. Women were given AmnioSense to wear for 20 minutes or until wetness was felt. A midwife read the result of the test after 10 minutes drying time. Another midwife, blinded to the AmnioSense results, carried out a speculum examination to obtain a clinical diagnosis. Data from 139 women were analysed after exclusions for protocol violations or clinical circumstances. A high vaginal swab was taken in 103 of these women to test for infection. Results of the AmnioSense test were compared with the clinical diagnosis, revealing a sensitivity of 98.0% (58/59, 95% CI 91 to 100%) and a specificity of 65.0% (52/80, 95% CI 54 to 75%). Of 28 AmnioSense results deemed to be false positives by clinical diagnosis, 7 were associated with infection determined by the vaginal swab. A further 2 women were subsequently identified as having ruptured membranes on further clinical examination, and 7 gave birth within 3 days of having the test. The positive predictive value was 67.4% (95% CI 56.5 to 77.2%) and the negative predictive value was 98.1% (95% CI 89.9 to 100.0%).
3.6 The Committee considered that the studies described above indicate that the Vision ALD is sufficiently accurate to reliably exclude amniotic fluid leak as a cause of vaginal wetness in pregnancy.
3.7 The Committee considered that use of the Vision ALD would avoid unnecessary speculum examinations, which may be intrusive and uncomfortable.
3.8 The Committee noted that 2 studies included additional comparators that were outside the scope, but the External Assessment Centre reanalysed the study data to address this issue and to provide clear comparative data for the Vision ALD against NHS standard practice.
3.9 The Committee noted that the clinical outcome evidence was limited, because all women participating in the studies received a separate clinical diagnosis and their vaginal wetness was not managed solely on the basis of the Vision ALD result.