Objective
The purpose of this study was to assess the effectiveness of electronic fetal monitoring (EFM) alone and with additional ST analysis (EFM + ST) in laboring women with a singleton term pregnancy that is in cephalic presentation in the prevention of metabolic acidosis by the application of individual patient data metaanalysis.
Study Design
We conducted an individual patient data metaanalysis using data from 4 randomized trials, which enabled us to account for missing data and investigate relevant subgroups. The primary outcome was metabolic acidosis, which was defined as an umbilical cord-artery pH <7.05 and a base deficit that had been calculated in the extra cellular fluid compartment >12 mmol/L. We performed 8 explanatory subgroup analyses for 8 different endpoints.
Results
We analyzed data from 12,987 women and their newborn infants. Metabolic acidosis was present in 57 women (0.9%) in the EFM + ST group and 73 women (1.1%) in the EFM alone group (relative risk [RR], 0.76; 95% CI, 0.53–1.10). Compared with EFM alone, the use of EFM + ST resulted in a reduction in the frequency of instrumental vaginal deliveries (RR, 0.90; 95% CI, 0.83–0.99) and fetal blood samples (RR, 0.49; 95% CI, 0.44–0.55). Cesarean delivery rates were comparable between both groups (RR, 0.99; 95% CI, 0.91–1.09). Subgroup analyses showed that EFM + ST resulted in fewer admissions to a neonatal intensive care unit for women with a duration of pregnancy of >41 weeks (RR, 0.61; 95% CI, 0.39–0.95).
Conclusion
EFM + ST does not reduce the risk of metabolic acidosis, but it does reduce the need for instrumental vaginal deliveries and fetal blood sampling.
Perinatal asphyxia is associated with several short- and long-term complications that vary from mild hypoxic ischemic encephalopathy to cerebral palsy and death. Fetal monitoring during delivery helps identify fetuses at risk of asphyxia. A relatively new method for continuous fetal monitoring is the STAN method (Neoventa Medical, Gothenburg, Sweden) in which the classification of the electronic fetal monitor (EFM) is combined with ST analysis of the fetal electrocardiogram. Similar to the postpartum electrocardiogram, information can be evaluated about the amplitude of the T-wave in relation to the QRS-complex (T/QRS ratio) and the conduction in the ST segment. Changes in the fetal electrocardiogram in combination EFM abnormalities could be an indication of fetal hypoxia, as shown in previous animal studies. Westgate et al were the first to conduct a randomized controlled trial (RCT) on the effect of intrapartum fetal electrocardiogram monitoring. Four subsequent RCTs focused on automatically detected T/QRS changes rather than absolute values of T/QRS. All 5 RCTs were inconclusive; 4 of the studies showed no statistically significant effect. In one study, ST analysis significantly reduced the incidence of metabolic acidosis. 7
See related editorial, page 163
To study the effect of ST analysis + EFM compared with EFM alone, metaanalyses were performed with the use of aggregated data (ADMA). These metaanalyses showed a nonsignificant reduction of metabolic acidosis when intrapartum ST analysis was used. These metaanalyses relied on published data. Because not all RCTs reported all endpoints of interest, some endpoints were excluded from the metaanalyses. Another limitation of these metaanalyses was that they did not investigate subgroups. Obviously, more information on relevant endpoints and subgroups was collected in the individual studies than was reported. A metaanalysis that uses individual participant data (IPDMA) allows for a more thorough investigation of endpoints and relevant subgroups by taking all this information into account. Furthermore, in IPDMA, it is possible to account for missing data.
In view of the shortcomings of conventional metaanalyses with the use of ADMA, we performed an IPDMA using data from RCTs to investigate the additional effect of ST analysis in EFM.
Materials and Methods
This study was conducted based on a previously written, but unpublished, protocol. The reporting of the IPDMA was carried out according to the preferred reporting items for systematic reviews and metaanalyses (PRISMA) guidelines.
Objective
The main objective of this study was to assess the effectiveness of the combination of ST analysis of the fetal electrocardiogram and EFM compared with EFM alone in laboring women with a term singleton pregnancy that was in cephalic presentation in the prevention of metabolic acidosis by means of an IPDMA.
Search strategy and selection criteria
Trials were identified by a search the following electronic databases for phase III trials of EFM + ST analysis compared with EFM alone, in laboring women with a term singleton pregnancy in cephalic presentation: Cochrane Central Register of Controlled Trials, PubMed, MEDLINE, Embase, ClinicalTrials.gov , and controlled-trials.com , following the search strategy of Becker et al. Two review authors (E.S. and A.K.) independently assessed inclusion criteria, study quality, and risk of bias. Discrepancies were resolved by third author (R.H.H.G.). The risk of bias was assessed by 2 independent reviewers (E.S. and A.K.) who used a modified version of the risk of bias tool that was developed by the Cochrane collaboration that contains specific items that assess adequate sequence generation (ie, computer-generated random number, the use of a random number table, or other truly random process), allocation concealment (ie, web-based or telephone central randomization), incomplete outcome data, and other possible sources of bias. Selective outcome reporting was not considered an issue because IPDMAs rely on IPD rather than reported outcomes. Studies were included if they had a low risk of bias, were focused on T/QRS changes of the fetal electrocardiogram, were completed before Dec. 1, 2011, and the principal investigators had provided the IPD relating EFM + ST analysis vs EFM alone. The relevant baseline characteristics and outcomes of interest, which are described later, were extracted by one of the authors (E.S.). Data quality (eg, discrepancies between published and shared data) was assessed independently by 2 review authors (E.S. and A.K.), and a third author (R.H.H.G.) resolved discrepancies.
Outcomes
The primary outcome was metabolic acidosis BDecf , defined as an umbilical cord-artery pH below 7.05 and a base deficit calculated in the extra cellular fluid compartment (BD ecf ) above 12 mmol/L, calculated with the Sigaard-Andersen algorithm. Secondary outcomes included metabolic acidosis BDblood that is defined as an umbilical cord-artery pH <7.05 and a base deficit calculated in blood (BD blood ) >12 mmol/L. Additional secondary outcomes were cord-artery pH <7.15, cord-artery pH <7.05, cord-artery pH <7.00, BD ecf >12 mmol/L, BD blood >12 mmol/L, 5-minute Apgar score <7, admission to a neonatal intensive care unit (NICU), hypoxic-ischemic encephalopathy, intubation, seizures, perinatal death, frequency of fetal blood samples, cesarean delivery, vaginal instrumental delivery, and the total frequency of operative deliveries. To increase comparability with a currently ongoing RCT that is being conducted by the National Institute of Child Health Development (NICHD) in the United States, we also used their primary outcome as one of our secondary outcomes. This outcome is a composite of intrapartum fetal death, neonatal death, Apgar score of ≤3 at 5 minutes, seizure(s), cord artery pH ≤7.05 and BD ecf ≥12 mmol/L, intubation for ventilation at delivery, or presence of neonatal encephalopathy.
Subgroups
Secondary objectives were to assess the additional effect of ST analysis in different subgroups differentiated by the following: (1) gestational age defined as <37 weeks, 37-40 weeks, 40-41 weeks, or >41 weeks; (2) parity defined as nulli- or multiparous; (3) previous cesarean delivery (yes/no); (4) maternal diabetes mellitus (yes/no); (5) induced onset of labor (yes/no); (6) meconium-stained amniotic fluid (yes/no); (7) epidural anesthesia (yes/no); and (8) birthweight below the tenth percentile (yes/no). The subgroup effects were investigated for the primary outcome, metabolic acidosis BDecf , and the following secondary outcomes: composite neonatal outcome, cesarean delivery, need for intubation, NICU admission, hypoxic-ischemic encephalopathy, instrumental vaginal delivery, and fetal blood sampling.
Analysis
All analyses were performed on all randomly assigned women in labor with a term singleton in cephalic presentation with an indication for internal EFM. The analyses were conducted on an intention-to-treat basis (ie, according to the treatment assigned by randomization, regardless of treatment actually received).
Descriptive comparisons between studies were conducted to assess between-study differences. Treatment effects on the primary and secondary outcomes were estimated by means of a random effects log-binomial model. The measure of association was the risk ratio (RR), with an RR <1 indicating treatment benefit. Both heterogeneity across studies and dependency between data that originated from the same study were taken into account by fitting a random intercept for each original study by means of a random effects model. The presence of heterogeneity in outcomes across trials was assessed using the I 2 measure, and the values were interpreted as follows: 0% indicates no observed heterogeneity; 25%, 50%, and 75% indicate low, moderate, and high heterogeneity, respectively. If necessary, analyses were adjusted for variables that were used in stratified randomization (eg, center and/or parity) by including them as covariates in the regression model. Additionally, we calculated the number needed to test (NNT) with 95% CI when an association was found to be statistically significant. The NNT is comparable with the numbers needed to treat but refers to the number of tests (in this case the number of laboring women who need to be monitored with EFM + ST analysis) to prevent 1 case of metabolic acidosis BDecf .
To investigate subgroup effects, the treatment effects were investigated with an interaction term between the allocation and the subgroup in the regression model defined earlier. When a significant interaction was present, the treatment effect was then estimated within strata based on that subgrouping variable. For the primary outcome, metabolic acidosis BDecf , which is a stratified analysis across the predefined subgroups, was performed despite the presence of a significant interaction in the regression model to investigate the direction of the additional effect of ST analysis in different strata of the subgroups.
The 4 RCTs had different proportions of missing values for the primary outcome that ranged from 2.4–14.5% ( Table 1 ). Because these missing values are often selectively missing, which was also the case in these RCTs (Appendix; Supplementary Tables 1-3 ; and Appendix 3 of Westerhuis et al ), a complete case analysis is likely to yield biased results. To avoid this bias, we used observed patient characteristics to impute missing data by means of multiple imputations. Missing data were imputed (10 times) with the use of a logistic regression model that included the following variables: center, allocation, parity, neonatal sex, Apgar score at 1 and 5 minutes, arterial pH, arterial BD blood , arterial BD ecf , arterial pCO2, venous pH, venous pCO2, birthweight, and indication for the intervention. The primary outcome was included in the imputation model to improve imputations for missing data on other variables of interest. Missing data were imputed within each individual study before pooling the studies. Analyses were performed individually on each of the 10 imputed data sets and results were pooled using standard methods (Rubin’s rule).
| Characteristic | Amer-Wahlin et al (Sweden) 2001 | Ojala et al (Finland) 2006 | Vayssiere et al (France) 2007 | Westerhuis et al (The Netherlands) 2010 |
|---|---|---|---|---|
| Type of study | Multicenter | Single center | Multicenter | Multicenter |
| n | 5049 | 1472 | 799 | 5667 |
| Inclusion criteria | Laboring women at >36 weeks of gestation; singleton fetus; cephalic position; continuous internal EFM needed | Laboring women at >36 weeks of gestation; singleton fetus; cephalic position; amniotomy decided | Laboring women at >36 weeks of gestation; singleton fetus; cephalic position; abnormal EFM or thick meconium-stained amniotic fluid (7%) during labor | Laboring women at >36 weeks of gestation; singleton fetus; cephalic position; age >18 years; indication for internal EFM |
| Exclusion criteria | None mentioned in article | Contraindication scalp electrode; admittance during second stage of labor | Contraindication scalp electrode; cardiac malformation | None mentioned in article |
| Index test | ST waveform + EFM (S21 a ) | ST waveform + EFM (S21) | ST waveform + EFM (S21) | ST waveform + EFM (S21 or S31 b ) |
| Controls | EFM | EFM | EFM | EFM |
| Allocation concealment | Yes; adequate | Yes; adequate | Yes; adequate | Yes; adequate |
| Sequence generation | Yes; adequate | Yes; adequate | Yes; adequate | Yes; adequate |
| Blinding of participants and medical professionals | No; not possible | No; not possible | No; not possible | No; not possible |
| Blinding of outcome assessors | Yes; adequate | Yes; adequate | Yes; adequate | Yes; adequate |
| Participant with incomplete primary outcome data, n (%) | 731 (14.5) | 36 (2.4) | 34 (4.3) | 549 (9.7) |
a STAN S21 fetal heart monitor that provides EFM + automatic ST analysis of the fetal electrocardiogram;
b STAN S31 fetal heart monitor (modern version of STAN S21) that provides EFM + automatic ST analysis of the fetal electrocardiogram.
Statistical analyses and multiple imputations were performed with the use of R software (version 2.15.0; The R Foundation for Statistical Computing, 2012).
Results
Included studies
Six studies on ST analysis in laboring women with a term singleton pregnancy that was in cephalic presentation were identified, of which 4 women met the inclusion criteria ( Figure ; Table 1 ). The study of Strachan et al was excluded because it studied the PR waveform of the fetal electrocardiogram rather than the ST segment. Even though the study of Westgate et al focused on T/QRS changes, the study was excluded because the ST-analysis method that was used was different than the methods used in more recent studies. In the study by Westgate et al, the STAN 8801 recorder (Neoventa Medical) was used; the other studies used the STAN S21 and/or S31. Although investigating T/QRS changes, the threshold for performance of an intervention was based on the absolute T/QRS ratio and not a change in T/QRS ratio. Furthermore, biphasic ST changes were not incorporated in the guideline. Another important difference was that ST changes were identified by visual analysis. The STAN S21 and S31 monitors provide an automatic assessment of the ST changes and give an automatic warning in case of significant changes. Datasets that contained IPD were obtained for 4 RCTs: Amer-Wahlin et al , Ojala et al , Vayssière et al, and Westerhuis et al.
The characteristics of the included studies are shown in Table 1 . In general, all studies had similar inclusion and exclusion criteria. The only exception was the study of Vayssière et al, which included only women who had an abnormal EFM or thick meconium-stained amniotic fluid during labor. Because all studies also used similar interventions and controls, these studies can be considered to have a high degree of homogeneity. The study of Westerhuis et al stratified the randomization of participants to EFM + ST analysis or EFM alone by center and parity (nulli- vs multiparous).
All trials used adequate methods to generate allocation sequences and adequate methods for allocation concealment ( Table 1 ). Because of the nature of the intervention, the blinding of participants and medical professionals was not possible. Blinding the assessors to the outcome was adequate in all trials. The number of women with incomplete primary outcome data differed per study but could be accounted for with multiple imputations. No other problems were found that could lead to bias.
Individual data from 6524 participants who were allocated to EFM + ST analysis of the fetal electrocardiogram and 6463 participants who were allocated to EFM alone were included in this IPDMA. The baseline characteristics of combined participants by treatment groups were similar ( Table 2 ).
| Characteristic | Study | Combined treatment groups | ||||
|---|---|---|---|---|---|---|
| Amer-Wahlin et al (Sweden) 2001 | Ojala et al (Finland) 2006 | Vayssiere et al (France) 2007 | Westerhuis et al (The Netherlands) 2010 | ST analysis + EFM | EFM alone | |
| n | 5049 | 1472 | 799 | 5667 | 6524 | 6463 |
| Mean maternal age, y a | NA | 28.0 ± 5.5 | 30.0 ± 5.7 | 32.0 ± 4.8 | 31.0 ± 5.3 | 31.0 ± 5.3 |
| Nulliparous, n (%) | 3105 (61) | 757 (51) | 575 (72) | 3236 (57) | 3851 (59) | 3823 (59) |
| Previous cesarean delivery, n (%) | NA | NA | 49 (6) | 716 (13) | 370 (11) b | 395 (12) b |
| Diabetes mellitus, n (%) | 104 (2) | 115 (8) | 41 (5) | 169 (3) | 261 (4) | 168 (3) |
| Female sex of the newborn infant, n (%) | 2388 (47) | 733 (50) | NA | 2668 (47) | 2860 (47) c | 2929 (48) c |
| Gestational age, wk a | 39.6 ± 1.6 | 40.1 ± 1.3 | 40.0 ± 1.9 | 40.2 ± 1.4 | 39.9 ± 1.5 | 40.0 ± 1.6 |
| Induced onset of labor, n (%) | 866 (17) | 277 (19) | 257 (36) | 2341 (41) | 1879 (29) | 1862 (29) |
| Meconium-stained amniotic fluid, n (%) | 1143 (23) | 260 (18) | 121 (15) | 1471 (26) | 1476 (23) | 1519 (24) |
| Epidural anesthesia, n (%) | 1957 (39) | 793 (54) | 725 (91) | 2389 (42) | 2898 (44) | 2966 (46) |
| Birthweight (g) a | 3567 ± 531 | 3605 ± 503 | 3243 ± 500 | 3544 ± 518 | 3546 ± 527 | 3536 ± 525 |
a Data are presented as mean ± SD;
b Percentage based on studies of Vayssiere et al and Westerhuis et al ;
c Percentage is based on studies of Amer-Wahlin et al, Ojala et al, and Westerhuis et al.
Overall effects of ST analysis of the fetal electrocardiogram
Table 3 shows the effect of ST analysis + EFM compared with EFM alone for the primary and secondary outcomes. The primary outcome, metabolic acidosis BDecf , was present in 57 women (0.9%) in the EFM with additional ST-analysis group and 73 women (1.1%) in the EFM alone group (RR, 0.76; 95% confidence interval [CI], 0.53–1.10). Using a 2-step approach (ie, analysis like an ADMA), we found a moderate amount of heterogeneity for the primary outcome between the studies (I 2 = 42%; 95% CI, 0–81%; Tau = 0.09).
| Outcome | Study sample size, n (%) | Combined treatment groups, n (%) | Relative risk (95% CI) a | P value | No. needed to test (95% CI) | I 2 : percentage (95% CI) | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| Amer-Wahlin et al | Ojala et al | Vayssiere et al | Westerhuis et al | ST analysis + EFM | EFM alone | |||||
| n | 5049 | 1472 | 799 | 5667 | 6524 | 6463 | — | — | — | — |
| Primary outcome | ||||||||||
| Metabolic acidosis BD ecf (pH<7.05; BD ecf >12 mmol/L) | 54 (1) | 10 (1) | 19 (2) | 46 (1) | 57 (0.9) | 73 (1.1) | 0.76 (0.53–1.10) | .13 | NC | 42 (0–81) |
| Secondary outcomes | ||||||||||
| Metabolic acidosis BD blood (pH<7.05 & BD blood >12 mmol/L) | NA | 23 (2) | NA | 107 (2) | 58 (1.6) a | 72 (2.0) a | 0.82 (0.58–1.16) | .25 | NC | 83 (56–93) |
| Arterial pH <7.15 | 997 (20) | 218 (15) | 159 (20) | 861 (15) | 1118 (17) | 1117 (17) | 0.99 (0.91–1.08) | .79 | NC | 0 (0–84) |
| Arterial pH <7.05 | 178 (4) | 28 (2) | 29 (2) | 117 (2) | 165 (2.5) | 187 (2.9) | 0.87 (0.70–1.09) | .20 | NC | 68 (6–89) |
| Arterial pH <7.00 | 67 (1) | 7 (0) | 14 (2) | 50 (1) | 65 (1.0) | 72 (1.1) | 0.89 (0.62–1.26) | .48 | NC | 57 (0–86) |
| BD ecf >12 mmol/L | 150 (3) | 34 (2) | 123 (15) | 204 (4) | 266 (4) | 246 (4) | 1.07 (0.90–1.29) | .42 | NC | 0 (0–27) |
| BD blood >12 mmol/L | NA | 83 (6) | NA | 413 (7) | 244 (7) a | 251 (7) a | 0.98 (0.82–1.16) | .80 | NC | 0 (0–44) |
| Apgar at 5 minutes <7 | 61 (1) | 17 (1) | 13 (2) | 76 (1) | 89 (1.4) | 78 (1.2) | 1.14 (0.84–1.54) | .41 | NC | 0 (0–0) |
| Admitted to a neonatal intensive care unit | 387 (8) | 49 (3) | 10 (1) | 86 (2) | 258 (4) | 274 (4) | 0.92 (0.78–1.09) | .32 | NC | 0 (0–0) |
| Hypoxic-ischemic encephalopathy | 7 (0) | 1 (0) | NA | 2 (0) | 3 (0.1) | 7 (0.2) | 0.42 (0.11–1.64) | .21 | NC | 0 (0–68) |
| Need for intubation | NA | 12 (1) | 8 (1) | 22 (3) | 16 (1.1) | 26 (1.7) | 0.64 (0.35–1.20) | .16 | NC | 18 (0–92) |
| Seizures | NA | 2 (0) | 2 (0) | 9 (1) | 4 (0.3) | 9 (0.6) | 0.46 (0.14–1.51) | .20 | NC | 0 (0–65) |
| Perinatal death | 3 (0) | 0 (0) | 1 (0) | 5 (0.1) | 5 (0.1) | 4 (0.1) | 1.24 (0.33–4.61) | .75 | NC | 0 (0–75) |
| Composite perinatal outcome b c | 75 (1) | 28 (2) | 33 (4) | 91 (2) | 101 (1.6) | 125 (1.9) | 0.80 (0.62–1.05) | .10 | NC | 0 (0–82) |
| Fetal blood sampling | NA | 166 (11) | 356 (45) | 879 (16) | 460 (12) | 941 (24) | 0.49 (0.44–0.55) | < .0001 | 13 (12–16) | 9 (0–91) |
| Cesarean delivery | 447 (9) | 82 (6) | 209 (26) | 796 (14) | 768 (12) | 766 (12) | 0.99 (0.91–1.09) | .91 | NC | 18 (0–87) |
| Fetal distress | 194 (4) | 30 (2) | 119 (15) | 164 (3) | 253 (4) | 254 (4) | 0.99 (0.83–1.17) | .87 | NC | 42 (0–81) |
| Failure to progress | 217 (4) | 37 (3) | NA | 509 (9) | 388 (6) | 375 (6) | 1.03 (0.90–1.18) | .70 | NC | 55 (0–87) |
| Instrumental vaginal delivery | 542 (11) | 149 (10) | 226 (28) | 815 (14) | 823 (13) | 909 (14) | 0.90 (0.83–0.99) | .02 | 69 (38–357) | 0 (0–77) |
| Fetal distress | 239 (5) | 84 (6) | 161 (20) | 337 (6) | 393 (6) | 428 (7) | 0.91 (0.80–1.05) | .19 | NC | 1 (0–85) |
| Failure to progress | 261 (5) | 30 (2) | NA | 361 (6) | 305 (5) | 347 (6) | 0.87 (0.75–1.01) | .07 | NC | 0 (0–87) |
| Operative delivery | 989 (20) | 231 (16) | 435 (55) | 1611 (28) | 1591 (24) | 1675 (26) | 0.94 (0.88–1.01) | .10 | NC | 0 (0–79) |
| Because of fetal distress | 433 (9) | 114 (8) | 280 (35) | 501 (9) | 646 (10) | 682 (11) | 0.94 (0.84–1.05) | .26 | NC | 50 (0–84) |
| Because of failure to progress | 478 (9) | 67 (5) | NA | 870 (15) | 694 (11) | 722 (12) | 0.95 (0.86–1.05) | .31 | NC | 65 (0–90) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
