Prior to an amniocentesis what action by the client will need to be completed

Please note, all analyses were carried out using a random‐effects model and so the average treatment effect is presented for all results.

(3) Chorionic villus sampling (CVS) versus amniocentesis

Primary outcomes

Total pregnancy loss was higher after transcervical CVS compared with amniocentesis (14.5% versus 11.0%; RR 1.40, 95% CI 1.09 to 1.81; women = 6527; studies = four; Analysis 3.1), but these results should be interpreted cautiously. In the transcervical CVS group, the total pregnancy loss varied from 7.3% in the Ammala 1993 (MRC Finland) trial to 19.5% in the Borrell 1999 trial. It is important to note that this was an intention‐to‐treat analysis, which included post‐randomisation, pre‐procedure pregnancy losses. Unfortuntaely, in the Borrell 1999 trial, these losses were extremely high (10.9%) and unbalanced between the two groups. The overall results for total pregnancy loss changed little without Borrell 1999 (RR 1.40, 95% CI 1.00 to 2.06). Interestingly, the statistical test for heterogeneity was significant, despite the fact that the results looked quite similar in terms of the size and direction of the observed differences in total pregnancy loss. The sensitivity analysis suggested that the heterogeneity was caused by the differences between the two largest trials (Canada 1989; Smidt‐Jensen 1993 (Denmark)). The increase in pregnancy loss after transcervical CVS was clear in the Smidt‐Jensen 1993 (Denmark) trial (RR 1.70, 95% CI 1.30 to 2.22), but not in the Canada 1989 trial (1.10, 95% CI 0.92 to 1.30). The results for spontaneous miscarriages (12.9% versus 9.4%; RR 1.5, 95% CI 1.07 to 2.11; studies = three; women = 5506; Analysis 3.2) were consistent with the results for total pregnancy loss described above.

There was no clear difference between groups for spontaneous miscarriage after test (pregnancy loss in women who had the test actually performed (RR 1.77, 95% CI 0.28 to 11.00; studies = two; women = 1579; Analysis 3.3).

Secondary outcomes

For technical difficulties in sampling, the transcervical CVS group had more multiple insertions (30.8% versus 7.8%; RR 3.93, 95% CI 2.72 to 5.68; studies = one; women = 794; Analysis 3.6), and second tests (6.3% versus 0.2%; RR 19.63, 95% CI 1.24 to 309.9; studies = three; women = 4256; Analysis 3.7), while there were no clear differences between groups for non‐compliance with allocated procedure (RR 0.51, 95% CI 0.14 to 1.87; studies = three; women = 4595; Analysis 3.4), or sampling failure (RR 0.55, 95% CI 0.26 to 1.19; studies = one; women = 797; Analysis 3.5).

For cytogenetic analysis, the transcervical CVS group had more laboratory failure (1.7% versus 0.1%; RR 22.62, 95% CI 3.07 to 166.89; studies = two; women = 2792; Analysis 3.8), and confined mosaics (2.3% versus 0.4%; RR 5.66, 95% CI 1.97 to 16.24; studies = one; women = 1995; Analysis 3.11), and maternal contamination (3.8% versus 0.3%; RR 12.3, 95% CI 3.81 to 39.67; studies = one; women = 1991; Analysis 3.12).

For pregnancy complications, the transcervical CVS group had more vaginal bleeding after test (19.4% versus 2.4%; RR 11.48, 95% CI 2.58 to 51.08; studies = two; women = 3193; Analysis 3.19), and pre‐labour ruptured membranes before 28 weeks (4.1% versus 0.8%; RR 4.97, 95% CI 1.45 to 17.03; studies = one; women = 722; Analysis 3.22).

There were a number of outcomes with potentially clinically important differences, but the number of events were not large enough to precisely estimate the risks. These were: termination of pregnancy (all; 2.4% versus 2.7%; RR 0.88, 95% CI 0.58 to 1.34; studies = two; women = 3454, Analysis 3.26), perinatal mortality (stillbirths and neonatal deaths in the first week of life; 0.5% versus 0.3%; RR 1.79, 95% CI 0.42 to 7.69; studies = three; women = 5521; Analysis 3.27), stillbirths (0.3% versus 0.2%; RR 0.94, 95% CI 0.02 to 45.31; studies = two; women = 3454; Analysis 3.28), neonatal death (death in the first week of life; 0.2% versus 0.1%; RR 1.63, 95% CI 0.38 to 7.05; studies = three; women = 4251; Analysis 3.29), and all recorded deaths after viability (0.5% versus 0.6%; RR 0.78, 95% CI 0.02 to 25.93; studies = two; women = 1579; Analysis 3.30.

The same was true for neonatal complications, such as anomalies (all recorded; 1.0% versus 1.6%; RR 0.62, 95% CI 0.25 to 1.59; studies = two; neonates = 1408; Analysis 3.31), and talipes (clubfoot; RR 0.62, actual numbers not provided; studies = one; neonates = 797).

The following secondary outcomes were not reported.

• All mosaics.

• Reporting time (interval between sampling and result).

• Amniotic leakage after test.

• Delivery before 33 weeks.

• Haemangiomas (localised vascular lesions of the skin and subcutaneous tissue).

• Limb reduction defects.

• Admission to special care baby unit.

• Neonatal respiratory distress syndrome.

• Birthweight below the 10th centile.

• Birthweight below the 5th centile.

3.2. Transabdominal CVS versus second trimester amniocentesis

One trial with 2234 women contributed to this comparison (Smidt‐Jensen 1993 (Denmark)).

Primary outcomes

A subgroup of Smidt‐Jensen 1993 (Denmark) compared transabdominal CVS with second trimester amniocentesis and found no clear difference in the total pregnancy loss between the two procedures (6.3% versus 7%; RR 0.90, 95% CI 0.66 to 1.23; studies = one; women = 2234; Analysis 3.1; low‐quality evidence). The same was true for spontaneous miscarriage (3.0% versus 3.9%; RR 0.77, 95% CI 0.49 to 1.21; studies = one; women = 2069; Analysis 3.2; low‐quality evidence).

Spontaneous miscarriage after test (pregnancy loss in women who had the test actually performed) was not reported.

Secondary outcomes

There were no clinically important differences in perinatal mortality (stillbirths and neonatal deaths in the first week of life; 0.7% versus 0.6%; RR 1.18, 95% CI 0.40 to 3.51; studies = one; women = 2069; Analysis 3.27; low‐quality evidence), but the number of events were not large enough to precisely estimate the risks. There were no clear differences between groups for amniotic leakage after test (1.4% versus 0.6%; RR 2.53, 95% CI 0.81 to 7.92; studies = one; women = 1485; Analysis 3.20).

The following secondary outcomes were not reported.

• Non‐compliance with allocated procedure.

• Sampling failure.

• Multiple insertions.

• Second test performed.

• Laboratory failure.

• All non‐mosaic abnormalities.

• All mosaics.

• True mosaics.

• Confined mosaics.

• Maternal contamination.

• Known false positive after birth.

• Known false negative after birth.

• Reporting time (interval between sampling and result).

• Vaginal bleeding after test.

• Vaginal bleeding after 20 weeks.

• pre‐labour ruptured membranes before 28 weeks.

• Antenatal hospital admission.

• Delivery before 37 weeks.

• Delivery before 33 weeks.

• Termination of pregnancy (all).

• Stillbirths.

• Neonatal death (death in the first week of life).

• All recorded deaths after viability.

• Anomalies (all recorded).

• Talipes (clubfoot).

• Haemangiomas (localised vascular lesions of the skin and subcutaneous tissue).

• Limb reduction defects.

• Admission to special care baby unit.

• Neonatal respiratory distress syndrome.

• Birthweight below the 10th centile.

• Birthweight below the 5th centile.

3.3. CVS by any route versus second trimester amniocentesis

Primary outcomes

Two trials with 6503 women presented data for the comparison between CVS performed by any route and mid‐trimester amniocentesis (MRC 1991; Smidt‐Jensen 1993 (Denmark)).

Overall pregnancy loss (including termination of pregnancy) was higher after CVS (11.1% versus 8.2%; RR 1.43, 95% CI 1.22 to 1.67; studies = two; women = 6503; Analysis 3.1). Again, an increase in spontaneous miscarriages after CVS was the main contributing factor (RR 3.46, 95% CI 2.21 to 5.42; studies = one; women = 3201; Analysis 3.3. There were more spontaneous miscarriages in the CVS group (7.1% versus 5.0%; RR 1.51, 95% CI 1.23 to 1.85; studies = two; women = 6280; Analysis 3.2).

Secondary outcomes

Overall, the test had to be repeated more commonly after transcervical CVS compared with second trimester amniocentesis. Also, there were more problems in analysing placental tissue obtained from CVS compared with amniotic fluid. In the transcervical CVS group, laboratory failure occurred in 1.7% cases, compared with only 0.07% after amniocentesis; there were also more cytogenetic abnormalities confined to the placenta, and more known false positive after birth and known false negative after birth results. However, cytogenetic results presented here should be interpreted with caution. They probably underestimate the true incidence of inaccurate results in both the CVS and amniocentesis groups, because the majority of fetal losses were not karyotyped post‐mortem, either because of technical difficulties or concerns about medico‐legal implications. The lack of complete cytogenetic follow‐up in all trials made unbiased analyses on all randomised women impossible.

Complications were uncommon after both procedures, and there were no reports that these were ever life‐threatening. Vaginal bleeding following the procedure was more common after transcervical CVS, although there was no difference in the incidence of vaginal bleeding later in pregnancy. There was no significant difference in the amniotic fluid leakage following the procedure, and pre‐labour spontaneous rupture of membranes before 28 weeks in MRC 1991, but this observation should be interpreted cautiously, because data on ruptured membranes were missing for large numbers of women. Interestingly, one participating centre reported a significant increase in ruptured membranes after transcervical CVS (Ammala 1993 (MRC Finland)). No differential effect was detected on antenatal admission to hospital.

In the sub‐project of the Canada 1989 trial, Spencer 1987, Spencer 1988, and Robinson 1988 compared the psychological effects of transcervical CVS and amniocentesis. In mid‐pregnancy, women allocated to amniocentesis were more anxious, and felt less attachment to their babies, although by 22 weeks, these differences seemed to disappear (data were not available in a form suitable to include in a meta‐analysis). Nevertheless, at 22 weeks, there was a suggestion of a persistent differential effect manifested in a decreased desire for another child, associated with amniocentesis (7/26 in the CVS group compared with 13/25 after amniocentesis).

Possible links between CVS, amniocentesis, and congenital anomalies could not be fully explored because of incomplete reporting, and a relatively small number of participants. There have been several reports in the past suggesting the presence of congenital anomalies (limb deformities in particular) in infants exposed to CVS in the first trimester. The available data from included studies did not support this observation. However, it should be remembered that the relationship may be gestation‐dependent. The majority of procedures were carried out after nine weeks' gestation, and therefore, did not address the possibility that CVS carried out very early in pregnancy may increase the risk of congenital abnormalities.

For technical difficulties in sampling, the CVS group had more sampling failure (4.8% versus 1.6%; RR 3.09, 95% CI 1.98 to 4.82; studies = one; women = 3201; Analysis 3.5), multiple insertions (30.7% versus 6.3%; RR 4.85, 95% CI 3.92 to 6.01; studies = one; women = 2917; Analysis 3.6), and second tests performed (6.2% versus 2.1%; RR 2.83, 95% CI 1.94 to 4.13; studies = one; women = 3201; Analysis 3.7). There were no clear differences between the groups for numbers of laboratory failures (RR 0.77, 95% CI 0.29 to 2.06; studies = one; women = 3201; Analysis 3.8), or known false positive after birth results (RR 0.99, 95% CI 0.06 to 15.80; studies = one; women = 3201; Analysis 3.13).

Reporting time for cytogenetic analyses was analysed in weekly tranches, although we did not pre‐specify these particular time intervals. The CVS group had more results given in less than seven days (15.2% versus 0.6%; RR 23.52, 95% CI 12.54 to 44.10; studies = one; women = 3099; Analysis 3.15), less than 14 days (22.5% versus 5.7%; RR 3.96, 95% CI 3.17 to 4.95; studies = one; women = 3099; Analysis 3.16), and fewer results given at more than 21 days (10.8% versus 32.6%; RR 0.33, 95% CI 0.28 to 0.39; studies = one; women = 3099; Analysis 3.18).

For pregnancy complications, the CVS group had more deliveries at less than 37 weeks (18.3% versus 13.7%; RR 1.33, 95% CI 1.13 to 1.57; studies = one; women = 3189; Analysis 3.24).

There were a number of outcomes with potentially clinically important differences, but the number of events was not large enough to precisely estimate the risks. These were: termination of pregnancy (all; 3.7% versus 2.6%; RR 1.42, 95% CI 0.96 to 2.11; studies = one; women = 3201; Analysis 3.26), perinatal mortality (stillbirths and neonatal deaths in the first week of life; 0.7% versus 0.6%; RR 1.20, 95% CI 0.64 to 2.24; studies = two; neonates = 6280; Analysis 3.27), stillbirths (0.4% versus 0.4%; RR 0.99, 95% CI 0.35 to 2.81; studies = one; neonates = 3201; Analysis 3.28), neonatal death (death in the first week of life; 0.5% versus 0.2%; RR 2.64, 95% CI 0.7 to 9.93; number of studies = one; neonates = 3201; Analysis 3.29), all recorded deaths after viability (1.0% versus 0.7%; RR 1.44; 95% CI 0.67 to 3.09; studies = one; neonates = 3201; Analysis 3.30), anomalies (all recorded; 5.9% versus 7.2%; RR 0.77, 95% CI 0.66 to 0.89; studies = two; number of neonates = 3338; Analysis 3.31), haemangiomas (localised vascular lesions of the skin and subcutaneous tissue; 29.4% versus 21.8%; RR 1.35, 95% CI 0.81 to 2.24; studies = one; neonates = 182; Analysis 3.32), and limb reduction defects (0.1% versus 0.0%; RR 4.95, 95% CI 0.24 to 102.97; studies = one; neonates = 3201; Analysis 3.33).

The following secondary outcomes were not reported.

• All non‐mosaic abnormalities.

• All mosaics.

• True mosaics.

• Confined mosaics.

• Reporting time (interval between sampling and result).

• Vaginal bleeding after test.

• Delivery before 33 weeks.

• Talipes (clubfoot).

• Admission to special care baby unit.

• Neonatal respiratory distress syndrome.

• Birthweight below the 10th centile.

• Birthweight below the 5th centile.

Primary outcomes

Compared with transabdominal CVS, total pregnancy loss and spontaneous miscarriages were higher after transcervical CVS, but this was due to the excess loss in the transcervical arm of the Smidt‐Jensen 1993 (Denmark) trial. This trial reported total pregnancy loss after transcervical CVS of 12.4% compared with 7.4% after transabdominal CVS. Corresponding figures for spontaneous pregnancy loss were 8.2% and 3%. However, there was no clear difference between groups for total pregnancy loss and miscarriage rate in the other four trials (Bovicelli 1986; Brambati 1991; Tomassini 1988; Jackson 1992). Because of these differences, there was statistical heterogeneity for these two outcomes (I² = 72.3%). When the fixed‐effect model was used to summarise the results for these two outcomes, transcervical CVS was associated with an increase in total pregnancy loss (RR 1.23, 95% CI 1.06 to 1.42) and spontaneous miscarriage (RR 1.75, 95% CI 1.33 to 2.29). However, in the presence of heterogeneity, it was prudent to apply a more conservative random‐effects model. When we applied this statistical model, the differences in pregnancy loss and miscarriage between transabdominal and transcervical CVS were no longer clear (RR 1.68, 95% CI 0.79 to 3.58; Analysis 4.2 for spontaneous miscarriage, and RR 1.16, 95% CI 0.81 to 1.56; Analysis 4.1 for total pregnancy loss; very low‐quality evidence).

There were many outcomes with potentially clinically important differences, but the number of events was not large enough to precisely estimate the risks. These were: all known pregnancy loss (including termination of pregnancy; 9.0% versus 7.4%; RR 1.16, 95% CI 0.81 to 1.65; studies = five; women = 7978; Analysis 4.1; very low‐quality evidence), spontaneous miscarriage; 7.9% versus 4.5%; RR 1.68, 95% CI 0.79 to 3.58; studies = four; women = 3384; Analysis 4.2; very low‐quality evidence), spontaneous miscarriage after test (pregnancy loss in women who had the test actually performed; 4.9% versus 3.9%; RR 1.23, 95% CI 0.75 to 2.04; studies = three; women = 1347; Analysis 4.3).

Secondary outcomes

Congenital anomalies were reported only in two studies with 1314 women, but the numbers were too small for meaningful comparisons (Brambati 1991; Smidt‐Jensen 1993 (Denmark)).

Transcervical CVS was more likely to fail, although there was a disproportionate contribution of the data from Jackson 1992 (weight 91%). Transcervical CVS appeared to be more technically demanding, requiring more multiple insertions and causing more vaginal bleeding. As far as cytogenetic analysis was concerned, both procedures were comparable.

For technical difficulties in sampling, the transcervical CVS group had more multiple insertions (11.2% versus 4.1%; RR 2.54, 95% CI 1.47 to 4.42; studies = two; women = 1314; Analysis 4.6), and more sampling failures (RR 1.79, 95% CI 1.13 to 2.82; studies = four; women = 5231; Analysis 4.5; moderate‐quality evidence). There were too few events to show a difference in laboratory failures between the two groups (1.5% versus 0.6%; RR 2.23, 95% CI 0.69 to 7.22; studies = one; women = 1194; low‐quality evidence; Analysis 4.8).

There were many outcomes with potentially clinically important differences, but the number of events were not large enough to precisely estimate the risks. These were: all non‐mosaic abnormalities (4.8% versus 3.9%; RR 1.23, 95% CI 0.87 to 1.75; studies = one; neonates = 2862; Analysis 4.9), true mosaics (0.7% versus 0.8%; RR 0.92, 95% CI 0.39 to 2.17; studies = one; neonates = 2862; Analysis 4.10), and confined mosaics (0.3% versus 0.4%; RR 0.85, 95% CI 0.26 to 2.77; studies = one; neonates = 2862; Analysis 4.11), amniotic leakage after test (0% versus 5%; RR 0.28, 95% CI 0.01 to 6.52; studies = one; women = 44; Analysis 4.12), vaginal bleeding after test (10.0% versus 1.6%; RR 6.93, 95% CI 0.77 to 62.83; studies = three; women = 1358; Analysis 4.13), termination of pregnancy (all; 6.6% versus 8.0%; RR 0.83, 95% CI 0.56 to 1.22; studies = two; women = 1303; Analysis 4.14), perinatal mortality (stillbirths and neonatal deaths in the first week of life; 0.3% versus 0.7%; RR 0.44, 95% CI 0.11 to 1.68; studies = one; neonates = 2037; Analysis 4.15), stillbirths (0.6% versus 0.3%; RR 1.36, 95% CI 0.11 to 17.53; number of studies = two; unborn neonates = 1227; Analysis 4.16), neonatal death (death in the first week of life; 0.1% versus 0.2%; RR 0.60, 95% CI 0.14 to 2.55; studies = two; neonates = 4845; Analysis 4.17), anomalies (all recorded; 1.4% versus 2.0%; RR 0.68, 95% CI 0.41 to 1.12; studies = two; neonates = 3622; Analysis 4.18; low‐quality evidence), and talipes (clubfoot; 0.2% versus 0.07%; RR 3.21, 95% CI 0.33 to 30.80; studies = one; neonates = 2622; Analysis 4.19).

The following secondary outcomes were not reported.

• All mosaics.

• Maternal contamination.

• Known false positive after birth.

• Known false negative after birth.

• Vaginal bleeding after test.

• Antenatal hospital admission.

• Delivery before 33 weeks.

• Delivery before 37 weeks.

• All recorded deaths after viability.

• Haemangiomas (localised vascular lesions of the skin and subcutaneous tissue).

• Limb reduction defects.

• Admission to special care baby unit.

• Neonatal respiratory distress syndrome.

• Birthweight below the 10th centile.

• Birthweight below the 5th centile.

Primary outcomes

In the early amniocentesis group, there were more spontaneous miscarriages; 2.2% versus 1.3%; RR 1.73, 95% CI 1.15 to 2.60; studies = four; women = 5491; Analysis 5.2; moderate‐quality evidence), and more spontaneous miscarriages after test (pregnancy loss in women who had the test actually performed; RR 1.71, 95% CI 1.12 to 2.61; studies = four; women = 5489; Analysis 5.3).

There was no clear difference in all known pregnancy loss (including termination of pregnancy; 3.5% versus 3.0%; RR 1.15, 95% CI 0.89 to 1.54; studies = four; women = 5491; Analysis 5.1; low‐quality evidence).

Secondary outcomes

There was no clear difference in the overall incidence of anomalies in the newborn infants (RR 1.14, 95% CI 0.57 to 2.30; Analysis 5.27; very low‐quality evidence). However, inter‐study heterogeneity was significant for this outcome, with no obvious explanation for the observed differences between Sundberg 1997 (Copenhagen) and Leiden 1998. Both groups had specifically highlighted two types of anomalies, talipes equinovarus and haemangiomas. The incidence of talipes in the EA group was 0.9% compared with 0.1% in the CVS group (RR 3.75, 95% CI 1.42 to 9.88).

An increased number of haemangiomas after CVS, seen in Leiden 1998, was not seen in the other two studies (RR 0.75, 95% CI 0.26 to 2.20). Only Leiden 1998 reported long‐term follow‐up of randomised infants, and none of them had abnormal results on the Dutch version of the Denver Developmental Screening Test, when visited at home between six and nine months of age.

Transabdominal CVS appeared to be more technically demanding, with more technical difficulties during the procedure, i.e. sampling failure, multiple insertions, and the need for a second test. However, the overall incidence of these complications was low. There were no clear differences in the rate of laboratory failures or the number of women with various chromosomal abnormalities. However, the numbers were too small for any meaningful comparison between the two methods.

In Sundberg 1997 (Copenhagen), the EA samples required a mean of 9.5 days (range 5 to 19) for culturing, compared to 6.1 days (range 4 to 14) for the CVS samples. In Leiden 1998, the mean culture time in the EA group was 13.8 days for the Amniomax culture and 15.6 for the Chang culture, compared to eight days in the CVS group. In the Philip 2004 (NICHD EATA) EA, 10.3 days were needed to obtain the result (standard deviation (SD) 2.5), compared with 6.3 days (SD 3). We did not pool these results because they were not normally distributed.

There were many outcomes with potentially clinically important differences, but the number of events was not large enough to precisely estimate the risks. These were: non‐compliance with allocated procedure (0.1% versus 0.7%; RR 0.25, 95% CI 0.09 to 0.72; studies = four; women = 5566; Analysis 5.4), sampling failure (0.5% versus 0.8%; RR 0.58, 95% CI 0.24 to 1.38; studies = four; women = 5566; Analysis 5.5; low‐quality evidence), multiple insertions (1.1% versus 2.5%; RR 0.45, 95% CI 0.21 to 0.95; studies = three; women = 4445; Analysis 5.6), second test performed (0.8% versus 1.4%; RR 0.63, 95% CI 0.28 to 1.43; studies = four; women = 5566; Analysis 5.7), laboratory failure (0.4% versus 0.6%; RR 0.74, 95% CI 0.34 to 1.63; studies = four; women = 5566; Analysis 5.8; low‐quality evidence), all non‐mosaic abnormalities (1.7% versus 1.7%; RR 0.95, 95% CI 0.47 to 1.90; studies = four; neonates = 5566; Analysis 5.9), true mosaics (0.1% versus 0.2%; RR 0.47, 95% CI 0.10 to 2.20; studies = three; neonates = 5451; Analysis 5.10), maternal contamination (0.4% versus 0.1%; RR 1.92, 95% CI 0.02 to 162.80; studies = two; women = 4330; Analysis 5.12), known false positive after birth (0.0% versus 0.3%; RR 0.36, 95% CI 0.02 to 8.73; studies = two; neonates = 670; Analysis 5.13), known false negative after birth (no events reported; studies = one; neonates = 555; Analysis 5.14; low‐quality evidence), vaginal bleeding after test (1.5% versus 2.2%; RR 0.69, 95% CI 0.42 to 1.12; studies = three; women = 4934; Analysis 5.18), amniotic leakage after test (9.0% versus 3.6%; RR 3.35, 95% CI 0.37 to 30.09; studies = three; women = 4934; Analysis 5.17), vaginal bleeding after 20 weeks (0.7% versus 1.0%; RR 0.71, 95% CI 0.35 to 1.43; studies = one; women = 3698; Analysis 5.16), pre‐labour ruptured membranes before 28 weeks (0.8% versus 1.7%; RR 0.50, 95% CI 0.27 to 0.92; studies = one; women = 3698; Analysis 5.19), delivery before 37 weeks (5.5% versus 4.8%; RR 1.16, 95% CI 0.78 to 1.74; studies = three; women = 1755; Analysis 5.20), delivery before 33 weeks (0.4% versus 0.7%; RR 0.50, 95% CI 0.09 to 2.73; studies = one; women = 1121; Analysis 5.21; low‐quality evidence), termination of pregnancy (all; 0.9% versus 1.2%; RR 0.74, 95% CI 0.45 to 1.25; studies = four; women = 5489; Analysis 5.22), perinatal mortality (stillbirths and neonatal deaths in the first week of life; 0.6% versus 0.5%; RR 1.10, 95% CI 0.53 to 2.28; studies = four; neonates = 5428; Analysis 5.23), stillbirths (0.5% versus 0.5%; RR 1.11, 95% CI 0.52 to 2.36; studies = four; neonates = 5428; Analysis 5.24), neonatal death (death in the first week of life; 0.04% versus 0.1%; RR 0.41, 95% CI 0.05 to 3.11; studies = four; neonates = 5455; Analysis 5.25), all recorded deaths after viability (0.3% versus 0.3%; RR 1.18, 95% CI 0.43 to 3.24; studies = four; neonates = 5453; Analysis 5.26), anomalies (all recorded; 3.0% versus 2.5%; RR 1.14, 95% CI 0.57 to 2.30; studies = four; neonates = 5305; Analysis 5.27; very low‐quality evidence), talipes (clubfoot; 0.9% versus 0.2%; RR 3.75, 95% CI 1.42 to 9.88; studies = four; neonates = 5305; Analysis 5.28), haemangiomas (localised vascular lesions of the skin and subcutaneous tissue; 4.5% versus 5.2%; RR 0.75, 95% CI 0.26 to 2.20; studies = four; neonates = 5305; Analysis 5.29), neonatal respiratory distress syndrome (defined by authors; 0.4% versus 0.4%; RR 0.91, 95% CI 0.21 to 3.98; studies = four; neonates = 4725; Analysis 5.30), birthweight below the 10th centile (6.6% versus 7.9%; RR 0.84, 95% CI 0.66 to 1.06; studies = one; neonates = 3618; Analysis 5.31), birthweight below the 5th centile (2.9% versus 2.8%; RR 0.66, 95% CI 0.05 to 9.38; studies = two; neonates = 629; Analysis 5.32).

The following secondary outcomes were not reported.

• All mosaics.

• Confined mosaics.

• Antenatal hospital admission.

• Limb reduction defects.

• Admission to special care baby unit.

(6) Ultrasound‐guided amniocentesis

Nolan 1981 (223 women) evaluated the type of ultrasound‐assisted procedure that is currently considered obsolete (i.e. this was not an ultrasound‐guided procedure in the true meaning of this term). There were no clear differences in the reported outcomes, but the study was too small to assess the true impact of the placental localisation by ultrasound before the needle insertion.

Primary outcomes

There were potentially clinically important differences in all known pregnancy loss (including termination of pregnancy; 0.0% versus 0.9%; RR 0.33, 95% CI 0.01 to 8.02; studies = one; women = 223; Analysis 6.1), spontaneous miscarriage; 0.0% versus 0.9%; RR 0.33, 95% CI 0.01 to 8.02; studies = one; women = 223; Analysis 6.2), spontaneous miscarriage after test (pregnancy loss in women who had the test actually performed; 0.0% versus 0.9%; RR 0.33, 95% CI 0.01 to 8.02; studies = one; women = 223; Analysis 6.3), but the number of events were not large enough to precisely estimate the risks.

Secondary outcomes

For outcomes related to technical difficulties in sampling, there was more sampling failure (4.5% versus 0.0%; RR 10.90, 95% CI 4.5 to 0.0; studies = one; women = 223; Analysis 6.4) and fewer multiple insertions (18.8% versus 27.9%; RR 0.67, 95% CI 0.41 to 1.09; studies = one; women = 223; Analysis 6.5) in the ultrasound group.

The following secondary outcomes were not reported.

• Non‐compliance with allocated procedure.

• Second test performed.

• Laboratory failure.

• All non‐mosaic abnormalities.

• All mosaics.

• True mosaics.

• Confined mosaics.

• Maternal contamination.

• Known false positive after birth.

• Known false negative after birth.

• Vaginal bleeding after test.

• Amniotic leakage after test.

• Vaginal bleeding after 20 weeks.

• pre‐labour ruptured membranes before 28 weeks.

• Antenatal hospital admission.

• Delivery before 37 weeks.

• Delivery before 33 weeks.

• Termination of pregnancy (all) 'included in all known pregnancy loss'.

• Perinatal mortality (stillbirths and neonatal deaths in the first week of life).

• Stillbirths.

• Neonatal death (death in the first week of life).

• All recorded deaths after viability.

• Anomalies (all recorded).

• Talipes (clubfoot).

• Haemangiomas (localised vascular lesions of the skin and subcutaneous tissue).

• Limb reduction defects.

• Admission to special care baby unit.

• Neonatal respiratory distress syndrome.

• Birthweight below the 10th centile.

• Birthweight below the 5th centile.

Page 2

Second trimester amniocentesis compared to control for prenatal diagnosis