Professional Documents
Culture Documents
Published online 2 December 2013 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/uog.12566
Correspondence to: Dr W. P. Martins, Department of Obstetrics and Gynecology, Medical School of Ribeirao Preto, University of Sao
Paulo (FMRP-USP), Av. Bandeirantes, 3900 - 8 andar - HCRP - Campus Universitario, Ribeirao Preto, Sao Paulo 14048-900, Brazil
(e-mail: wpmartins@gmail.com)
Accepted: 8 July 2013
Copyright 2013 ISUOG. Published by John Wiley & Sons Ltd. SYSTEMATIC REVIEW
26 Martins et al.
Copyright 2013 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2014; 43: 25–33.
Monitoring controlled ovarian stimulation 27
Copyright 2013 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2014; 43: 25–33.
28 Martins et al.
Figure 1 Flowchart of selection of studies comparing methods for monitoring controlled ovarian stimulation.
Copyright 2013 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2014; 43: 25–33.
Table 1 Characteristics of included randomized controlled trials comparing methods for monitoring controlled ovarian stimulation
*Only provided pregnancy rate per oocyte retrieval and embryo transfer, not per woman allocated. 2D, two-dimensional; 3D, three-dimensional; COS, controlled ovarian stimulation;
GnRH, gonadotropin-releasing hormone; H, hormonal monitoring; hCG, human chorionic gonadotropin; IVF, in-vitro fertilization; OHSS, ovarian hyperstimulation syndrome; SonoAVC,
3D sonography-based automated volume count; US, ultrasound.
Other bias
outcome (RR, 0.95; 95% CI, 0.78–1.16; P = 0.61; four
RCTs, 611 women; I2 = 10%, low-quality evidence)
(Figure 4). The CI was very wide when sensitivity
analysis involved restriction of eligibility criteria to
Aguirre et al. (2010)26 + + + − + +
studies judged not to be at high risk of bias and we were
Golan et al. (1994)24 + + + + not able to conclude whether ultrasonography alone is
better, similar or less effective than ultrasonography and
Lass (2003)29 + + + + + + −
hormonal assessment for this outcome (RR, 0.96; 95%
Raine-Fenning et al. (2010)27 + + + + + + CI, 0.62–1.48; P = 0.84; one RCT, 185 women).
Only one study reported on miscarriage per clinical
Rongieres (2006)28 + + + + + + pregnancy, and the CI was very wide; we were not
+ + + + − + +
able to conclude whether ultrasonography alone is
Wiser et al. (2012)25
better, similar or less effective than ultrasonography
and hormonal assessment for this outcome (RR, 0.37;
Figure 2 Risk of bias summary: review of authors’ judgements 95% CI, 0.07–1.79; P = 0.21; 45 pregnant women,
about each risk of bias for each included study. very-low-quality evidence). Sensitivity analysis was
not consider that blinding was likely to influence the risk not undertaken because the only study that reported
of performance bias for the present review. Two studies miscarriage was considered to be at high risk of bias.
were deemed at high risk of attrition bias because some For total number of oocytes retrieved, the CI was
participants were excluded from analysis25,26 : in one, four somewhat wide and we were not able to conclude
women were excluded from analysis because they did not whether ultrasonography alone is better or similar
follow the study protocol26 ; in the other, two women were to ultrasonography and hormonal assessment for this
excluded after allocation25 (one became pregnant before outcome (mean difference (MD), 0.77 oocytes; 95% CI,
COS and the other left the clinic before treatment). One –0.42 to 1.96; P = 0.20; three RCTs, 474 women; I2 = 10,
study was considered at unclear risk of reporting bias24 ; moderate-quality evidence) (Figure 5). When performing
the authors reported the proportion of clinical pregnan- the sensitivity analysis, the CI was wide and we were not
cies per oocyte retrieval and per embryo transfer but did able to conclude whether ultrasonography alone is better,
not report the number of women submitted to oocyte similar or slightly less effective than ultrasonography and
retrieval and/or embryo transfer in each group; thus, the hormonal assessment for this outcome (MD, 1.70 oocytes;
total number of clinical pregnancies in each group could 95% CI, –1.22 to 4.62; P = 0.25; 1 RCT, 114 women).
not be determined. One study was deemed at high risk
of other bias because the initial intervention proposed Comparison of 3D and 2D ultrasound monitoring
for the ultrasonography/hormonal assessment group was There was only one study included in this comparison,
modified during the trial: 40/148 women did not match which did not report live birth, OHSS or miscarriage. CIs
the criteria for hCG administration using the initial pro- were very wide for the two reported outcomes (clinical
posed criteria but still received hCG according to criteria pregnancy and total number of oocytes retrieved) and we
proposed for the ultrasonography only group29 . No study were not able to conclude whether use of 3D ultrasound is
reported significant differences in baseline age, body mass better, similar, or less effective than use of 2D ultrasound;
index or follicle-stimulating hormone level between the considering the values for clinical pregnancy (RR, 1.00;
groups. The risk of bias summary is presented in Figure 2. 95% CI, 0.58–1.73; P = 1.00; 72 women) and total
Monitoring with ultrasonography only vs number of oocytes retrieved (MD, –0.37 oocytes; 95%
ultrasonography and hormonal assessment CI, –3.63 to 2.89; P = 0.82; 72 women), this study was
not judged to be at high risk of bias.
For OHSS, the CI was very wide and we were not We did not identify any evidence of publication bias
able to conclude whether ultrasonography alone is or selective reporting within studies. Since fewer than
Copyright 2013 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2014; 43: 25–33.
Monitoring controlled ovarian stimulation 31
US alone US + Hormones
Study or Subgroup Events Total Events Total Weight (%) Peto odds ratio Peto odds ratio
Figure 3 Forest plot for ovarian hyperstimulation syndrome: comparison of ultrasound (US) monitoring alone vs US and hormonal
monitoring.
US alone US + Hormones
Study or Subgroup Events Total Events Total Weight (%) Risk ratio Risk ratio
26
Aguirre et al. (2010) 26 35 19 31 17.4 1.21 (0.86, 1.70)
Lass (2003)29 46 149 49 148 42.4 0.93 (0.67, 1.30)
Rongieres (2006)28 26 88 30 97 24.6 0.96 (0.62, 1.48)
Wiser et al. (2012)25 12 30 19 33 15.6 0.69 (0.41, 1.18)
Figure 4 Forest plot for clinical pregnancy: comparison of ultrasound (US) monitoring alone vs US and hormonal monitoring.
10 studies were included, we did not perform funnel 95% CI) and another level because of the high risk of bias
plot analysis. Sensitivity analysis, restricting the inclusion in the included studies. The quality of evidence for clinical
criteria to studies judged not to be at high risk of bias pregnancy was also considered to be low; it was also
was the only additional analysis performed, and this was downgraded one level because of imprecision and another
reported along with the main results. level because of the high risk of bias. The quality of
evidence for miscarriage was considered to be very low; it
was downgraded two levels because of serious imprecision
DISCUSSION
(very wide 95% CI) and another level because of the high
Five studies were included for comparison of ultra- risk of bias. The quality of evidence for the number of
sonography only with ultrasonography and hormonal oocytes retrieved was considered to be moderate; it was
assessment (Table 2). No study evaluated live birth. We downgraded one level because of the high risk of bias
observed moderate-quality evidence that monitoring COS observed in the included studies.
by ultrasonography alone is unlikely to cause substantial In the comparison of 3D and 2D ultrasound (Table 2),
reduction in the number of oocytes retrieved as compared the quality of evidence for the two reported outcomes
with that with ultrasonography and hormonal assessment. (clinical pregnancy and number of oocytes retrieved) was
There is low-quality evidence that monitoring COS only considered to be low; it was downgraded two levels
by ultrasonography probably does not substantially alter because of serious imprecision (very wide 95% CI).
the chance of achieving a clinical pregnancy. We are still Another systematic review published in 200816
uncertain of its effect on OHSS and miscarriage. addressed this issue; at that time only two studies were
Only one study was included for comparison of 3D and included, and both are included also in the present
2D ultrasound monitoring. This study provided data on review24,29 . Those authors concluded that evidence was
the number of oocytes retrieved and on clinical pregnancy; too scarce at that point to draw any conclusion. We
however, the estimates were not sufficiently precise to rule included four more studies: one from 200628 and three
out appreciable harm or benefit. studies that were published after the former review25 – 27 .
In the comparison of ultrasonography only with In addition to the six studies included in the quan-
ultrasonography and hormonal assessment (Table 2), the titative review, three other studies addressed the same
quality of evidence for OHSS was considered to be low; it question but were excluded because they were not
was downgraded one level because of imprecision (wide randomized15,21,22 . Two of these studies21,22 evaluated
Copyright 2013 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2014; 43: 25–33.
32 Martins et al.
US alone US + Hormones
Study or Subgroup Mean SD Total Mean SD Total Weight (%) Mean difference Mean difference
Golan et al., (1994)24 13.4 7.5 57 11.7 8.4 57 16.6 1.70 (−1.22, 4.62)
Lass (2003)29 11.4 6.1 149 11 6.3 148 71.3 0.40 (−1.01, 1.81)
Wiser et al., (2012)25 11.7 8 30 10 5.5 33 12.1 1.70 (−1.72, 5.12)
Figure 5 Forest plot for oocytes retrieved: comparison of ultrasound (US) monitoring alone vs US and hormonal monitoring.
Number of
participants Quality
Variable Absolute risk (95% CI) Difference (95% CI) (studies) of evidence
*Downgraded one level because of imprecision and an additional level because of the quality of the included studies (3/5 studies were at high
risk of bias). †Downgraded one level because of imprecision and an additional level because of the quality of the included studies (3/4 studies
were at high risk of bias). ‡Downgraded two levels because of serious imprecision and an additional level because of the quality of the
included studies (the only included study was at high risk of bias). §Downgraded one level because of the quality of the included studies (2/3
studies were at high risk of bias). ¶Downgraded two levels because of serious imprecision. H, hormonal assessment; MD, mean difference;
OHSS, ovarian hyperstimulation syndrome; OR, odds ratio; RR, risk ratio; US, ultrasonography.
the comparison of ultrasonography alone with ultra- alone should be considered a reasonable alternative for
sonography and hormonal assessment. One study includ- monitoring COS.
ing 206 women22 reported live birth (RR, 1.05; 95% Although the included studies were consistent in their
CI, 0.47–2.32; P = 0.91), OHSS (OR, 0.87; 95% CI, estimate of effect, the quality of evidence compiled in this
0.05–14.12; P = 0.92) and clinical pregnancy per woman review was impaired by the quality of the studies. Three
allocated (RR, 1.04; 95% CI, 0.57–1.90; P = 0.91). The out of six studies were deemed to be at high risk of bias.
other study21 included 120 women and reported clin- Beyond the core reasons for eliminating the monitoring
ical pregnancy per allocated woman (RR, 0.83; 95% of some hormones and inclusion of semiautomated
CI, 0.25–2.80; P = 0.77) and total number of oocytes techniques, the reduction in costs and/or stress and
retrieved (MD, –0.30; 95% CI, –1.63 to 1.03; P = 0.66). the improvement of logistics should be considered; no
The third study15 evaluated use of the software SonoAVC trial has addressed cost-effectiveness or quality of life
for comparison of 3D and 2D ultrasound in 40 women as related to these interventions. Additionally, the small
and reported on clinical pregnancy per allocated woman number of RCTs makes the assessment of publication and
(RR, 0.90; 95% CI, 0.47–1.73; P = 0.75) and total num- reporting bias suboptimal, because of the impossibility of
ber of oocytes retrieved (MD, 3.20 oocytes; 95% CI, performing a funnel plot analysis.
–0.89 to 7.29; P = 0.12). None of these three studies Current evidence suggests that monitoring COS by
reported a significant difference between groups, which is ultrasound alone is unlikely to alter substantially the
in agreement with the meta-analysis of studies included in chance of achieving a clinical pregnancy and the number
the quantitative review. of oocytes retrieved is at least similar when compared
Monitoring COS by both ultrasonography and hor- to that retrieved using ultrasonography and hormonal
monal assessment leads to increased costs as well as assessment. For the other outcomes and comparisons, the
discomfort and stress related to the process of blood col- available data remain inconclusive. We believe that more
lection. Although the effect on other important outcomes studies evaluating the optimal procedure for monitoring
is uncertain, monitoring COS by ultrasonography alone COS are needed; these studies should report on both live
is not likely to reduce the number of oocytes retrieved birth and OHSS and should include at least 300 women
or to alter substantially the chance of achieving a clinical per group to ensure sufficient power to detect an absolute
pregnancy. Therefore, we believe that ultrasonography difference greater than 10% for live birth.
Copyright 2013 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2014; 43: 25–33.
Monitoring controlled ovarian stimulation 33
Copyright 2013 ISUOG. Published by John Wiley & Sons Ltd. Ultrasound Obstet Gynecol 2014; 43: 25–33.