- Email: [email protected]
Quantitative cardiotocography to improve fetal assessment during labor: a preliminary randomized controlled trial
Accepted Manuscript Title: Quantitative cardiotocography to improve fetal assessment during labor: A preliminary randomized controlled trial Author: Petar N. Ignatov Jennifer E. Lutomski PII: DOI: Reference:
S0301-2115(16)30871-5 http://dx.doi.org/doi:10.1016/j.ejogrb.2016.08.023 EURO 9567
To appear in:
EURO
Received date: Revised date: Accepted date:
6-1-2016 29-7-2016 1-8-2016
Please cite this article as: Ignatov Petar N, Lutomski Jennifer E.Quantitative cardiotocography to improve fetal assessment during labor: A preliminary randomized controlled trial.European Journal of Obstetrics and Gynecology and Reproductive Biology http://dx.doi.org/10.1016/j.ejogrb.2016.08.023 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Quantitative cardiotocography to improve fetal assessment during labor: A preliminary randomized controlled trial Petar N. Ignatov, MD, PhD1,2; Jennifer E. Lutomski, PhD3,4; 1
Second Municipal Hospital for Obstetrics and Gynecology Sheynovo, Sofia, Bulgaria 2
Nadezhda Women’s Health Hospital, Sofia, Bulgaria
3
National Perinatal Epidemiology Centre, Department of Obstetrics and Gynaecology, University College Cork, Cork, Ireland 4
Radboud Institute for Health Sciences, Radboud University Medical Center, Nijmegen, Netherlands
Correspondence to: Petar Ignatov Department of Perinatal Medicine 41-43 Skobelev blvd, Sofia 1606, Bulgaria E-mail: [email protected] Tel: +359 888 963 189 Country of study: Bulgaria Source of financial support: (1) Bulgarian Christmas 2013-2014 Charity Initiative; (2) Sheynovo - Second Municipal Hospital for Obstetrics and Gynaecology Abstract word count: 261 Manuscript word count: 2,882 Running headline: qCTG to improve fetal assessment Abbreviations: cardiotography (CTG); quantitative cardiotocography (qCTG); neonatal intensive care unit (NICU); relative risk (RR); confidence interval (CI); receiving operator characteristic (ROC); area under the curve (AUC) Conflicts of interest: Petar N. Ignatov assisted in the development of the reported quantitative cardiotocography system and is currently developing a web-based software interface. Jennifer E. Lutomski is also the lead author of a Cochrane Systematic Review, entitled “Expert Systems for Fetal Assessment”.
1
CONDENSATION Observer bias in conventional cardiotocography (CTG) monitoring may be reduced through using computerized decision support. This preliminary trial suggests that CTG with computerized decision support reduces incidence of hypoxia, acidemia, cesarean delivery and admission to the neonatal intensive care unit.
2
ABSTRACT Objective: To evaluate the effectiveness of a computerized decision support system, referred to as “quantitative cardiotocography” (qCTG), to reduce adverse birth outcomes compared to conventional CTG with fetal blood sampling. Study Design: A preliminary parallel randomized control trial in a tertiary maternity hospital (Sofia, Bulgaria) was conducted with a sample size of 360 women per trial arm (N=720). Women in labor were recruited between March 2008 and March 2011. Unadjusted relative risks were derived to assess the effect of qCTG on outcomes of interest. A ROC curve was derived to determine the sensitivity and specificity of qCTG to detect acidemia. (Clinical trial registration: Current Controlled Trials, http://www.controlled-trials.com/, ISRCTN46449237) Main outcome measures: Primary outcomes were hypoxia (cord-artery blood pH <7.20), acidemia (umbilical-artery blood pH <7.05), cesarean delivery, and forceps extraction. Secondary outcomes were Apgar score <7 at five minutes, neonatal seizures, and admission to the neonatal intensive care unit (NICU). Results: Reduced risks were observed for all outcomes of interest in women monitored using qCTG. There was a significant reduction in hypoxia (RR: 0.53; 0.33, 0.84), acidemia (RR: 0.31; 95% CI: 0.12, 0.84), cesarean delivery (95% CI: 0.45, 0.85), and admission to the NICU (RR: 0.33; 95% CI: 0.14, 0.77) in women monitored using qCTG versus conventional CTG. Conclusion: qCTG may reduce risk of adverse birth outcomes; however, the small sample size and long recruitment period in this trial may overstate the benefits of this intervention. Further large-scale randomized control trials with sufficient sample size to detect rare adverse events are required prior to the adoption of qCTG in daily clinical practice.
3
Keywords: cardiotocography; clinical decision support systems; acidemia; fetal hypoxia; cesarean section
4
1
INTRODUCTION
2
Cardiotocography (CTG) is often used for fetal assessment in obstetric practice; yet,
3
interpreting CTG traces can prove challenging. Extensive research has found that
4
the same CTG trace may elicit inconsistent interpretations between maternity care
5
providers [1-5], which is disconcerting given the impact of CTG traces on clinical
6
decision-making [6]. A CTG trace incorrectly identified as normal delays necessary
7
intervention, potentially increasing risk of hypoxia or metabolic acidosis in the infant
8
[6]. Conversely, a trace incorrectly identified as abnormal may result in unnecessary
9
intervention, such as induction of labor or cesarean delivery.
10
Substantial research has been invested into improving CTG interpretation
11
through concurrent assessment of fetal pulse oximetry [7], lactate level
12
measurements [8], and fetal ECG waveform analysis [9] with limited success [7-9].
13
However, the use of computerized decision support systems may be an underutilized
14
alterative to improve CTG interpretation by synthesizing clinical data to generate
15
alerts and/or recommendations on appropriate interventions.
16
The potential for computerized decision support systems to improve CTG
17
interpretation has resulted in on-going trials [10, 11] and a recent Cochrane review
18
[6]. Yet, despite growing interest [12-14], evidence in this area remains sparse.
19
Therefore, the purpose of this trial was to evaluate the effectiveness of a
20
computerized decision support system, referred to as “quantitative cardiotocography”
21
(qCTG), to facilitate CTG interpretation. The hypothesis of the trial was that the
22
incidence of adverse birth outcomes would be reduced in women monitored with
23
qCTG versus conventional CTG with fetal blood sampling.
24
5
25
METHODS AND MATERIALS
26
Setting and participants
27
A preliminary parallel trial was undertaken between March 2008 and March 2011 at
28
Second Municipal Hospital for Obstetrics and Gynecology Sheynovo, Sofia, Bulgaria,
29
a large (~4,000 deliveries per annum) tertiary maternity hospital. In 2011, the Second
30
Municipal Hospital recorded a perinatal mortality rate of 7 per 1,000 deliveries; the
31
incidence of cesarean delivery was 32%.
32
Eligible participants were women >18 years of age with a singleton pregnancy
33
in cephalic position and no known fetal structural abnormalities. Furthermore, women
34
had to be in active labor. Sample size was based on the incidence of acidosis [15]
35
using one-sided testing, with a 90% confidence level with 80% power. Accounting for
36
potential attrition, approximately 720 women would be required to detect a significant
37
decrease in acidosis incidence from 5.0% in women monitored with conventional
38
CTG to 2.0% in women monitored with qCTG. Ethical approval for this trial was
39
obtained from the Second Municipal Hospital for Obstetrics and Gynecology
40
Research Ethical Committee (Reference: 00134/19.02.2008); all participating women
41
provided informed consent. This trial has been retrospectively registered on Current
42
Controlled Trials website (http://www.controlled-trials.com/; Trial registration
43
identification number: ISRCTN46449237; Registered 02/10/2014).
44
The intervention
45
The intervention was a computerized decision support system to facilitate CTG
46
interpretation. This system, referred to as qCTG, uses external monitoring to
47
synthesize the three domains of a CTG: microfluctuations in fetal heart rate, fetal
48
heart rate and decelerations. Notably, the domain “microfluctuations” is distinct from
49
fetal heart rate variability and refers to the number of extrema per minute, the mean
6
50
beat-to-beat variability per minute and the oscillation amplitudes. These three
51
domains are scored on a scale ranging between zero (normal measure) and six
52
(highly abnormal measure) and summated for an overall CTG score. Thus, the
53
overall CTG score ranges between zero (normal trace) and 18 (pre-terminal trace).
54
Using cordocentesis, Roemer and Walden previously demonstrated a strong
55
correlation between the overall CTG score and fetal pH at delivery [16, 17]. Ignatov
56
et al. undertook additional validation work of the qCTG system and modified the
57
system to enhance prognostic ability [18, 19]. In summary, this validation work
58
identified slight measurement error between qCTG predicted pH values and “true”
59
pH values based on blood gas analysis. When averaging the last six measurements
60
taken prior to delivery, qCTG predicted pH values which ranged from -0.037 to
61
+0.046 relative to the “true” pH value [18]. Moreover, the major parameters of a CTG
62
(microfluctuations in fetal heart rate, fetal heart rate and decelerations) were not
63
equal in terms of their prognostic ability of fetal pH, justifying the evaluation of
64
specific subgroups of parameters [19]. To account for these measurement issues,
65
Ignatov et al. developed qCTG guidelines for clinical application [20] (Table 1).
66
Currently, qCTG is available in the NEXUS / OBSTETRICS system, formerly
67
known as the ARGUS system (Nexus GMT, Frankfurt, Germany), which is one of
68
several recognized fetal monitoring systems. In this system, predicted pH values are
69
calculated and updated every five minutes. As seen in Figure 1, the most recently
70
predicted pH value is displayed in red font on the left side of the interface; previous
71
pH values are represented by red points in the white area below the CTG reading.
72
Microfluctuations in fetal heart rate, fetal heart rate and decelerations (abbreviated
73
as OSZ, FRQ, and DEC respectively) are numerically presented on the lower left
74
side of the interface. 7
75
Prior to the onset of the trial, four obstetricians and the head of the delivery
76
ward received on-site specialized training from Nexus GMT representatives. Overall,
77
seven obstetricians assisted in the implementation of the study. Women were
78
enrolled into the trial by an attending obstetrician in the prenatal unit of the clinic.
79
Women who met the inclusion criteria were randomly assigned to receive
80
conventional CTG monitoring with fetal blood sampling (control group) or qCTG
81
monitoring (intervention group) in a 1:1 ratio with randomly varied permuted block
82
sizes of 10 and 20. The randomization was computer generated by an external
83
statistician. Sequentially numbered, sealed, opaque envelopes were selected in
84
consecutive order by a senior obstetrician in the delivery ward to allocate women to
85
the intervention or control groups. Given the nature of the intervention, neither the
86
women nor the attending obstetricians were blinded to the intervention.
87
In the control group, CTG traces were interpreted according to a modified
88
version of the International Federation of Gynecology and Obstetrics (FIGO)
89
guidelines [21]. In the event of an abnormal CTG trace, fetal blood sampling was
90
performed to assist clinical decision making. A persistent abnormal CTG and/or a
91
fetal scalp blood sample with a pH <7.20 resulted in immediate emergency cesarean
92
delivery.
93
In the qCTG arm, management of labor was conducted in accordance with
94
clinical practice guidelines developed by Ignatov et al. [20] According to these
95
guidelines, if the results from the qCTG were normal, labor was monitored
96
intermittently. If the results were suspicious (i.e. predicted umbilical blood pH level
97
between 7.20 and 7.10), labor was monitored continuously with regular re-
98
evaluation. If the qCTG resulted in persistent abnormal readings (i.e. predicted
99
umbilical blood pH level <7.10), an emergency cesarean delivery was performed.
8
100
In both arms of the trial, qCTG or CTG monitoring was discontinued
101
approximately five to ten minutes before the cesarean delivery or vaginal birth.
102
Outcomes of interest
103
There were four primary outcomes of interest: incidence of hypoxia (defined as a
104
cord-artery blood pH <7.20); incidence of acidemia (defined as umbilical-artery blood
105
pH <7.05); cesarean delivery; forceps extraction. Notably, blood gases were
106
measured in all neonates in both study arms. There were three secondary outcomes
107
of interest: Apgar score less than seven at five minutes; incidence of neonatal
108
seizures; admission to the neonatal intensive care unit (NICU). All outcomes were
109
assessed immediately after birth with the exception of neonatal seizures and NICU
110
admission. The incidence of neonatal seizures and NICU admission were assessed
111
within the first 24-hours post-delivery.
112
Statistical analysis
113
To avoid detection bias, the statistician performing the analysis was blinded to
114
whether the women where in the intervention or the control group. Unadjusted
115
relative risks (RR) and corresponding 95% confidence intervals (95% CI) were
116
derived for all outcomes of interest. Furthermore, a Receiver Operating
117
Characteristic (ROC) curve and area under the curve (AUC) were derived to assess
118
the prognostic ability of qCTG and conventional CTG to identify fetal hypoxia. Fetal
119
hypoxia was selected for this analysis since severe cases, i.e. cases of hypoxic
120
ischemic encephalopathy, is a major cause of infant mortality and chronic disability.
121
Values for the AUC range between zero and one, with a value of one representing
122
perfect prediction and a value of 0.5 representing no predictive value. Analyses were
123
conducted using MedCalc (V12.2.1.0. Ostend, Belgium: MedCalc Software) and
124
SPSS (Version 19.0. Armonk, NY: IBM Corp).
9
125
RESULTS
126
There was no attrition or exclusions of women after randomization. Subsequent
127
results are based on 720 women with 360 women in both the intervention and
128
control arms (Figure 2). Overall, nearly two-thirds (61.8%) of women recruited into
129
the trial were nulliparous, and more than one-third (42.5%) received epidural
130
analgesia (Table 2). The majority of labors (58.8%) were augmented with oxytocin.
131
Obstetric characteristics were similar in both trial arms (Table 2). In the control arm,
132
fetal blood sampling was performed in 38 (10.6%) women to guide CTG
133
interpretation. Multiple fetal blood samples were taken in 15 (4.2%) women.
134
A reduced risk for all primary and secondary adverse outcomes was observed
135
among women monitored using qCTG (Table 3). There was a significant reduction in
136
fetal hypoxia (RR: 0.53; 0.33, 0.84) and acidemia (RR: 0.31; 95% CI: 0.12, 0.84).
137
Moreover, relative to women monitored using conventional CTG with fetal blood
138
sampling, women monitored using qCTG had 0.62 (95% CI: 0.45, 0.85) times the
139
risk of a cesarean delivery. Furthermore, the risk of admission to the NICU was 0.33
140
(95% CI: 0.14, 0.77) in women monitored using qCTG versus conventional CTG. In
141
absolute terms, there would be six fewer cases of fetal hypoxia (95% CI: -10, -1),
142
three fewer cases of acidemia (95% CI: -5, -1), eight fewer cesarean deliveries (95%
143
CI: -14, -3) and three fewer NICU admissions (95% CI: -6, -1) per100 women
144
monitored using qCTG.
145
Marked differences were observed in the sensitivity and specificity of the
146
qCTG and CTG with fetal blood sampling in the identification of fetal hypoxia
147
(Figures 3 and 4). The sensitivity in the identification of hypoxia was 97.6% in
148
women monitored using qCTG compared to 87.2% in women monitored using
149
conventional CTG with fetal blood sampling. Moreover, specificity in the identification
10
150
of hypoxia was 89.2% in women monitored with qCTG compared to 59.7% in women
151
monitored with conventional CTG with fetal blood sampling.
152
COMMENT
153
The incidence of fetal hypoxia, acidemia, cesarean delivery and admission to the
154
NICU was significantly lower among women who were monitored using qCTG versus
155
women who were monitored using conventional CTG monitoring with fetal blood
156
sampling. These preliminary trial results are promising and suggest that
157
computerized decision support systems may be able enhance conventional CTG
158
interpretation. If such findings are supported in other clinical trials, a movement
159
towards computerized decision support systems for fetal assessment could have
160
profound effects on obstetric medicine. When an abnormal CTG trace is observed in
161
current practice, fetal blood sampling is often performed to assess fetal pH levels.
162
However, fetal blood sampling can only be performed at intermittent time points, in
163
essence providing cross-sectional data. Time gaps between fetal blood samples may
164
not capture the initial decline in fetal pH, consequently delaying timely diagnosis of
165
hypoxia and appropriate obstetric interventions. Fetal blood sampling also requires a
166
certain degree of cervical effacement, ruptured membranes, absence of vaginal
167
infection, and trained staff to perform the procedure; these characteristics are not
168
present in all deliveries. By providing continuous real-time predicted pH values
169
irrespective of cervical condition and membrane integrity, qCTG circumvents these
170
aforementioned issues.
171
In this study, the sensitivity for the identification of hypoxia in the conventional
172
CTG with fetal blood sampling group was 87.2%, which is in line with previous
173
research [17, 22-26]. To date, conventional CTG monitoring has failed to deliver
174
consistent results in respect to the specificity, with reports varying between 9% and
11
175
63% [27-29]. With a specificity of 59.7%, this study was of no exception. In contrast,
176
whereas both sensitivity and specificity for the identification of hypoxia improved in
177
women monitored using qCTG, specificity was strikingly higher.
178
Given the non-invasive nature of the intervention, no secondary harms were
179
anticipated during the design phase of the trial. Future trials, however, may wish to
180
assess maternal dissatisfaction as a possible source of unintended consequences
181
as well as longitudinal outcomes, such as developmental delay.
182
Several limitations should be noted. This trial was underpowered to examine
183
the breadth of relatively rare (<10% incidence rate) adverse birth outcomes, such as
184
perinatal death. Notably, no perinatal deaths occurred in either trial arm. However, if
185
the trial were prolonged, there remains the possibility that there may be a higher risk
186
for these rarer events. Early-phase studies of experimental interventions, such as
187
qCTG, may justify underpowered trials [30]. Despite the small sample size, there
188
were marked decreases the incidence of several outcomes of interest in women
189
monitored using qCTG, underscoring the potential for this technology. These data
190
will further contribute to systematic reviews in this field of study [6].
191
Secondly, this trial compared a novel fetal monitoring technology
192
(intervention) to a well-established fetal monitoring approach (conventional CTG with
193
fetal blood pH assessment, control). Reproducibility in other countries may be
194
difficult if their respective ethical committees deemed the unknown risk of this new
195
technology to outweigh the potential benefits. In the context of this preliminary trial,
196
careful monitoring of participants in conjunction with an interim analysis based on
197
220 women [15] was undertaken to avoid detrimental outcomes.
198
Whereas indication for cesarean delivery was typically fetal distress, there
199
were medical exceptions to ensure the health of the woman and infant. In several
12
200
cases of caesarean delivery, fetal blood samples were insufficient for analysis, and
201
given the presence of abnormal tracings, there was no time to repeat the procedure.
202
Furthermore, in some cases effacement was not sufficient, membranes were not
203
ruptured or vaginal infection was possible. Due to the presence of one or more of
204
these circumstances, blood sampling could not always be performed. However,
205
these cases were not omitted from the analysis because they represent realistic
206
limitations.
207
Furthermore, fetal blood sampling was only made available in the control arm
208
of the trial. Previous work had confirmed a strong correlation between actual and
209
predicted fetal pH levels [18]. Thus, this procedure was not performed in the
210
intervention arm.
211
Neither the women nor the obstetricians were blinded to the intervention,
212
potentially introducing performance bias. However, given the nature of the
213
intervention, blinding was not feasible.
214
The study had a relatively long recruitment period which may have resulted in
215
a sampling bias. The recruitment period was due to funding issues largely related to
216
all neonates requiring measurements for blood gases. Sheynovo is a municipality
217
hospital, and each year there was a three to six month wait for the funding body to
218
allocate resources. Despite the long recruitment period, arguably this is a minimal
219
source of bias.
220
Lastly, it is important to acknowledge that this trial was retrospectively
221
registered, which is often viewed as a potential source of reporting bias. This work
222
was originally carried out to improve obstetric outcomes in Bulgaria and, given the
223
target audience, trial registration was not originally sought. However, this research
224
has already led to changes in local obstetric practice and in light of the All Trials
13
225
campaign, which promotes the registration and publication of all performed trials
226
(www.alltrials.net), this trial was registered to increase the transparency and
227
accessibility of this study. Although all outcomes of interest in this trial were reported,
228
to avoid the risk of reporting bias, future trials should ideally always be prospectively
229
registered – irrespective of the trial size, geographic location or intended audience.
230
One major strength of this trial is that it is among one of the first publications
231
investigating the utility of computerized decision support systems for fetal monitoring
232
in labor and will complement future trials in this area for numerous reasons [15].
233
Firstly, not all computerized decision support systems will be based on identical
234
criteria. As more research develops in this area [10, 11], comparisons between
235
different systems will be necessary to identify the best performing computerized
236
decision support system. Secondly, the clinical practice guidelines used in this trial
237
include several improvements which enhance the correlation between prognostic
238
and actual fetal pH values [18, 19]. This allows for immediate recognition of any
239
significant changes in fetal pH, resulting in better timing of needed procedures.
240
Lastly, a potential bias often cited in research is that randomized control trials are
241
geographically restricted to certain regions, such as Western Europe or North
242
America. This study, performed in Eastern Europe (Bulgaria), represents output from
243
a historically under-represented country with a distinct maternity profile relative to
244
many of its Western counterparts. For instance, the national rate of cesarean
245
delivery is approximately 40%, with some private hospitals reporting rates of nearly
246
60% [31]. A recent report identified that two out of five women reported a clear
247
preference for cesarean delivery [32]. Moreover, based on local hospital observation,
248
the incidence of acidemia was notably higher than other settings. This has been
249
attributed, in part, to the large number of high risk referrals. Such underlying
14
250
conditions can influence new obstetric interventions, underscoring the need for
251
globally representative research.
252
Improving CTG monitoring during labor is a well-recognized goal in obstetric
253
medicine. This trial supports that qCTG has the potential to significantly reduce the
254
incidence of fetal hypoxia, acidemia, cesarean delivery and admission to the NICU in
255
comparison to conventional CTG monitoring with fetal blood sampling. Prior to the
256
adoption of qCTG in daily clinical practice, further large-scale randomized control
257
trials are justified as well as evaluation of different computerized decision support
258
systems for fetal monitoring.
15
REFERENCES (1) Chauhan SP, Klauser CK, Woodring TC, Sanderson M, Magann EF, Morrison JC. Intrapartum nonreassuring fetal heart rate tracing and prediction of adverse outcomes: interobserver variability. Am J Obstet Gynecol. 2008;199:623 e1-5. (2) Devane D, Lalor J. Midwives' visual interpretation of intrapartum cardiotocographs: intra- and inter-observer agreement. J Adv Nurs. 2005;52:133-41. (3) Figueras F, Albela S, Bonino S, Palacio M, Barrau E, Hernandez S, et al. Visual analysis of antepartum fetal heart rate tracings: inter- and intra-observer agreement and impact of knowledge of neonatal outcome. J Perinat Med. 2005;33:241-5. (4) Blix E, Sviggum O, Koss KS, Oian P. Inter-observer variation in assessment of 845 labour admission tests: comparison between midwives and obstetricians in the clinical setting and two experts. BJOG. 2003;110:1-5. (5) Ayres-de-Campos D, Bernardes J, Costa-Pereira A, Pereira-Leite L. Inconsistencies in classification by experts of cardiotocograms and subsequent clinical decision. Br J Obstet Gynaecol. 1999;106:1307-10. (6) Lutomski JE, Meaney S, Greene RA, Ryan AC, Devane D. Expert systems for fetal assessment in labour. Cochrane Database of Systematic Reviews 2015, Issue 4. Art. No.: CD010708. DOI: 10.1002/14651858.CD010708.pub2. (7) East CE, Chan FY, Colditz PB, Begg LM. Fetal pulse oximetry for fetal assessment in labour. Cochrane Database Syst Rev. 2007:CD004075. (8) East CE, Leader LR, Sheehan P, Henshall NE, Colditz PB. Intrapartum fetal scalp lactate sampling for fetal assessment in the presence of a non-reassuring fetal heart rate trace. Cochrane Database Syst Rev. 2010:CD006174. (9) Neilson JP. Fetal electrocardiogram (ECG) for fetal monitoring during labour. Cochrane Database Syst Rev. 2012;4:CD000116.
16
(10) Brocklehurst PA. A multicentre randomised controlled trial of an intelligent system to support decision making in the management of labour using the cardiotocogram (INFANT). Current Controlled Trials (www.controlledtrials. com/mrct) [accessed 6 September 2014]. 2013. (11) Ayres-de-Campos D, Ugwumadu A, Banfield P, Lynch P, Amin P, Horwell D, et al. A randomised clinical trial of intrapartum fetal monitoring with computer analysis and alerts versus previously available monitoring. BMC Pregnancy Childbirth. 2010;10:71. (12) Schiermeier S, Pildner von Steinburg S, Thieme A, Reinhard J, Daumer M, Scholz M, et al. Sensitivity and specificity of intrapartum computerised FIGO criteria for cardiotocography and fetal scalp pH during labour: multicentre, observational study. BJOG. 2008;115:1557-63. (13) Costa A, Santos C, Ayres-de-Campos D, Costa C, Bernardes J. Access to computerised analysis of intrapartum cardiotocographs improves clinicians' prediction of newborn umbilical artery blood pH. BJOG. 2010;117:1288-93. (14) Greene KR. Intelligent fetal heart rate computer systems in intrapartum surveillance. Curr Opin Obstet Gynecol. 1996;8:123-7. (15) Ignatov P, Atanasov B. [Indirect standard cardiotocography plus fetal blood sampling versus indirect quantitative cardiotocography--a randomized comparative study in intrapartum monitoring]. Akush Ginekol (Sofiia). 2012;51:3-10. (16) Roemer VM. [CTG: microfluctuation]. Z Geburtshilfe Neonatol. 2004;208:210-9. (17) Roemer VM, Walden R. [Quantitative Cardiotocography--what does it look like and what can we expect]. Z Geburtshilfe Neonatol. 2006;210:77-91. (18) Ignatov P, Atanasov B, Kostov I, Velev R, Kovacheva A, Dobreva A. [A comparative study of the prognostic pH values of the fetus in utero, generated by the
17
"quantitative cardiotocography" computer method, and the actual pH values measured immediately after delivery]. Akush Ginekol (Sofiia). 2010;49:3-11. (19) Ignatov P, Atanasov B. [Structure and function of the cardiotocographic score (CTG-score) calculated by the "quantitative cardiotocography" computer method. Determining the significance of its components for the accuracy of the estimates for the ph of the fetus]. Akush Ginekol (Sofiia). 2011;50:13-20. (20) Ignatov P, Atanasov B. ["Quantitative cardiotocography"--clinical practice guideline]. Akush Ginekol (Sofiia). 2011;50:3-9. (21) Su LL, Chong YS, Biswas A. Use of fetal electrocardiogram for intrapartum monitoring. Ann Acad Med Singapore. 2007;36:416-20. (22) International Federation of Gynecology and Obstetrics. Intrapartum surveillance: recommendations on current practice and overview of new developments. FIGO Study Group on the Assessment of New Technology. Int J Gynaecol Obstet. 1995;49:213-21. (23) Brown VA, Sawers RS, Parsons RJ, Duncan SL, Cooke ID. The value of antenatal cardiotocography in the management of high-risk pregnancy: a randomized controlled trial. Br J Obstet Gynaecol. 1982;89:716-22. (24) Goeschen K. Derzeitiger Stand der intrapartalen U berwachung des Kindes. Gynakologe. 1997;30. (25) Schneider H. Evaluation des CTG. Kritische Evaluation des CTG 1996;29:3-11. German. (26) Seelbach-Göbel B, Huch R, Luttkus A, Saling E, Vetter K. Ist die Pulsoxymetrie ein Gewinn für die überwachung des Feten sub partu? Perinatalmedizin. 1998;10:77-80. German.
18
(27) Steer PJ, Eigbe F, Lissauer TJ, Beard RW. Interrelationships among abnormal cardiotocograms in labor, meconium staining of the amniotic fluid, arterial cord blood pH, and Apgar scores. Obstet Gynecol. 1989;74:715-21. (28) Murphy KW, Johnson P, Moorcraft J, Pattinson R, Russell V, Turnbull A. Birth asphyxia and the intrapartum cardiotocograph. Br J Obstet Gynaecol. 1990;97:4709. (29) Axt R. Das intrapartale CTG. Gynakologe. 1997;30:577-80. German. (30) Halpern SD, Karlawish JH, Berlin JA. The continuing unethical conduct of underpowered clinical trials. JAMA. 2002;288:358-62. (31) Konstantinov S, Zlatkov V. [Types of hospital property and the relative rate of cesarean section occurrence]. Akush Ginekol (Sofiia). 2015;54:8-13. (32) Dimitrov A, Tsankova M, Krusteva K, Nikolov A. [Study of women preference regarding mode of delivery]. Akush Ginekol (Sofiia). 2004;43:13-7.
19
ACKNOWLEDGEMENTS PNI contributed to the conception, design and interpretation of data, drafting of the manuscript and critically revised the manuscript for intellectual content; PNI is guarantor. JEL contributed to the interpretation of data, drafting of the manuscript and critically revised the manuscript for intellectual content.
FUNDING This trial was funded by the Bulgarian Christmas 2013-2014 Charity Initiative and Sheynovo - Second Municipal Hospital for Obstetrics and Gynaecology. CONFLICTS OF INTEREST PNI has no conflicts of interest to declare. Petar N. Ignatov assisted in the development of the reported quantitative cardiotocography system and is currently developing a web-based software interface. JEL is the lead author of a Cochrane Systematic Review, entitled “Expert Systems for Fetal Assessment”.
20
Table 1: Clinical application guidelines for quantitative cardiotocography (qCTG) CTG Classification
Predicted pH values based on qCTGa
CTG-score componentsb
Management
Normal
7.350 – 7.237
All
Expectant Intermittent monitoring
OSZ OSZ+FRQ
Expectant Continuous monitoring
FRQ Suspicious
7.237 – 7.137
OSZ+DEC DEC FRQ+DEC OSZ+FRQ+DEC OSZ OSZ+FRQ FRQ
Abnormal
≤ 7.137
<30 mins Continuous monitoring >30 mins Urgent delivery <30 mins Continuous monitoring >30 mins Urgent delivery
OSZ+DEC DEC FRQ+DEC
Urgent delivery
OSZ+FRQ+DEC Abbreviations: CTG, cardiotocography; OSZ, microfluctuation; FRQ, fetal heart rate; DEC, decelerations a Although obstetric guidelines typically defined hypoxia as cord-artery blood pH <7.20 and acidemia as umbilical-artery blood pH <7.05, these thresholds were modified in this guideline to account for potential measurement error in qCTG. The degree of measurement error has been found to range between -0.037 and +0.046 (when compared to the reference standard of blood gas analyses). b OSZ, OSZ+FRQ and FRQ subgroups of the CGT-score were found to be less accurate in predicting pH levels compared to the remaining subgroups [19]. Therefore, no immediate actions are advised when these are present. Expectant management is advocated if suspicious findings are observed based on these less accurate subgroups of the CGT-score. If abnormal prognostic pH readings are observed based on OSZ, OSZ+FRQ and FRQ and they persist longer than 30 minutes, urgent delivery is recommended.
21
Table 2: Obstetric characteristics of included women qCTG CTG + FBS (N=360) (N=360) N (%) n (%) Age (range) 18-42 18-44 Nulliparous 211 (58.6) 234 (65.0) Gestational weeks 37 to 40 346 (96.1) 351 (97.5) +1 40 to 42 14 (3.9) 9 (2.5) Epidural analgesia 145 (40.3) 161 (44.7) Oxytocin augmentation 201 (55.8) 222 (61.7) Birth weight <2500g 21 (5.8) 17 (4.7)
Overall trial (N=720) n (%) 18-44 445 (61.8) 697 (96.8) 23 (3.2) 306 (42.5) 423 (58.8) 38 (5.3)
Abbreviations: qCTG, quantitative cardiotocography with decision support system; CTG + FBS, conventional cardiotocography with fetal blood sampling
22
Table 3: Risk estimates of primary and secondary outcomes qCTG CTG + FBS (N=360) (N=360) n (%) n (%) Primary outcomes Hypoxia 25 (6.9) 47 (13.1) Acidemia 5 (1.4) 16 (4.4) Cesarean delivery 50 (13.9) 81 (22.5) Forceps extraction 2 (0.6) 4 (1.1) Secondary outcomes Apgar score <7 at 5 minutes 9 (2.5) 18 (5.0) Neonatal seizures 3 (0.8) 5 (1.4) Admission to NICU 7 (1.9) 21 (5.8)
Relative Risk (95% CI)
0.53 (0.33, 0.84) 0.31 (0.12, 0.84) 0.62 (0.45, 0.85) 0.50 (0.09, 2.71) 0.50 (0.23, 1.10) 0.60 (0.14, 2.49) 0.33 (0.14, 0.77)
Abbreviations: qCTG, quantitative cardiotocography with decision support system; CTG + FBS, conventional cardiotocography with fetal blood sampling; CI, confidence interval; NICU, neonatal intensive care unit Note: Hypoxia was defined as a cord-artery blood pH <7.20. Acidemia was defined as umbilical-artery blood pH <7.05.
259
23
Figure 1: Illustrative example of quantitative cardiotocography (qCTG) interface 260 261
Please find attached separately
262
24
Figure 2: Participant flow chart
Figure 3: Receiving Operator Characteristic curve for identification of hypoxia using cardiotocography with computerized decision support system (qCTG) 266
Please find attached separately Note: Graph reflects predicted pH values generated by qCTG versus pH value determined by
blood gas analyses (reference standard). Data are based on a continuous scale. In both arms of the trial, qCTG or CTG monitoring was discontinued approximately five to ten minutes before the cesarean delivery or vaginal birth.
26
Figure 4: Receiving Operator Characteristic curve for identification of hypoxia using conventional cardiotocography (CTG) 267
Please find attached separately Note: Graph reflects likely hypoxia based on conventional CTG traces (yes/no)
versus hypoxia determined by blood gas analyses (reference standard). Data are dichotomous. In both arms of the trial, qCTG or CTG monitoring was discontinued approximately five to ten minutes before the cesarean delivery or vaginal birth.
27
Figure 1: Illustrative example of quantitative cardiotocography (qCTG) interface
260 261
24
Figure 2: Participant flow chart 262 263 264
Enrollment
Assessed for eligibility (n=765)
Excluded (n= 45) Did not meet inclusion criteria (n=35) Declined to participate (n=3) Other reasons (n=7)
Randomized (n=720)
Allocation Allocated to intervention (n=360) Received allocated intervention (n=360)
Allocated to intervention (n=360) Received allocated intervention (n=360)
Did not receive allocated intervention (n=0)
Did not receive allocated intervention (n=0)
Follow-Up Lost to follow-up (Non-applicable due to short time frame) Discontinued intervention (n=0)
Lost to follow-up (Non-applicable due to short time frame) Discontinued intervention (n=0)
Analysis Analysed (n=360) Excluded from analysis (n=0)
Analysed (n=360) Excluded from analysis (n=0)
25
Figure 3: Receiving Operator Characteristic curve for identification of hypoxia using cardiotocography with computerized decision support system (qCTG)
Note: Graph reflects predicted pH values generated by qCTG versus pH value determined by
blood gas analyses (reference standard). Data are based on a continuous scale. In both arms of the trial, qCTG or CTG monitoring was discontinued approximately five to ten minutes before the cesarean delivery or vaginal birth.
26
Figure 4: Receiving Operator Characteristic curve for identification of hypoxia using conventional cardiotocography (CTG)
Note: Graph reflects likely hypoxia based on conventional CTG traces (yes/no) versus
hypoxia determined by blood gas analyses (reference standard). Data are dichotomous. In both arms of the trial, qCTG or CTG monitoring was discontinued approximately five to ten minutes before the cesarean delivery or vaginal birth.
27
S0301-2115(16)30871-5 http://dx.doi.org/doi:10.1016/j.ejogrb.2016.08.023 EURO 9567
To appear in:
EURO
Received date: Revised date: Accepted date:
6-1-2016 29-7-2016 1-8-2016
Please cite this article as: Ignatov Petar N, Lutomski Jennifer E.Quantitative cardiotocography to improve fetal assessment during labor: A preliminary randomized controlled trial.European Journal of Obstetrics and Gynecology and Reproductive Biology http://dx.doi.org/10.1016/j.ejogrb.2016.08.023 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Quantitative cardiotocography to improve fetal assessment during labor: A preliminary randomized controlled trial Petar N. Ignatov, MD, PhD1,2; Jennifer E. Lutomski, PhD3,4; 1
Second Municipal Hospital for Obstetrics and Gynecology Sheynovo, Sofia, Bulgaria 2
Nadezhda Women’s Health Hospital, Sofia, Bulgaria
3
National Perinatal Epidemiology Centre, Department of Obstetrics and Gynaecology, University College Cork, Cork, Ireland 4
Radboud Institute for Health Sciences, Radboud University Medical Center, Nijmegen, Netherlands
Correspondence to: Petar Ignatov Department of Perinatal Medicine 41-43 Skobelev blvd, Sofia 1606, Bulgaria E-mail: [email protected] Tel: +359 888 963 189 Country of study: Bulgaria Source of financial support: (1) Bulgarian Christmas 2013-2014 Charity Initiative; (2) Sheynovo - Second Municipal Hospital for Obstetrics and Gynaecology Abstract word count: 261 Manuscript word count: 2,882 Running headline: qCTG to improve fetal assessment Abbreviations: cardiotography (CTG); quantitative cardiotocography (qCTG); neonatal intensive care unit (NICU); relative risk (RR); confidence interval (CI); receiving operator characteristic (ROC); area under the curve (AUC) Conflicts of interest: Petar N. Ignatov assisted in the development of the reported quantitative cardiotocography system and is currently developing a web-based software interface. Jennifer E. Lutomski is also the lead author of a Cochrane Systematic Review, entitled “Expert Systems for Fetal Assessment”.
1
CONDENSATION Observer bias in conventional cardiotocography (CTG) monitoring may be reduced through using computerized decision support. This preliminary trial suggests that CTG with computerized decision support reduces incidence of hypoxia, acidemia, cesarean delivery and admission to the neonatal intensive care unit.
2
ABSTRACT Objective: To evaluate the effectiveness of a computerized decision support system, referred to as “quantitative cardiotocography” (qCTG), to reduce adverse birth outcomes compared to conventional CTG with fetal blood sampling. Study Design: A preliminary parallel randomized control trial in a tertiary maternity hospital (Sofia, Bulgaria) was conducted with a sample size of 360 women per trial arm (N=720). Women in labor were recruited between March 2008 and March 2011. Unadjusted relative risks were derived to assess the effect of qCTG on outcomes of interest. A ROC curve was derived to determine the sensitivity and specificity of qCTG to detect acidemia. (Clinical trial registration: Current Controlled Trials, http://www.controlled-trials.com/, ISRCTN46449237) Main outcome measures: Primary outcomes were hypoxia (cord-artery blood pH <7.20), acidemia (umbilical-artery blood pH <7.05), cesarean delivery, and forceps extraction. Secondary outcomes were Apgar score <7 at five minutes, neonatal seizures, and admission to the neonatal intensive care unit (NICU). Results: Reduced risks were observed for all outcomes of interest in women monitored using qCTG. There was a significant reduction in hypoxia (RR: 0.53; 0.33, 0.84), acidemia (RR: 0.31; 95% CI: 0.12, 0.84), cesarean delivery (95% CI: 0.45, 0.85), and admission to the NICU (RR: 0.33; 95% CI: 0.14, 0.77) in women monitored using qCTG versus conventional CTG. Conclusion: qCTG may reduce risk of adverse birth outcomes; however, the small sample size and long recruitment period in this trial may overstate the benefits of this intervention. Further large-scale randomized control trials with sufficient sample size to detect rare adverse events are required prior to the adoption of qCTG in daily clinical practice.
3
Keywords: cardiotocography; clinical decision support systems; acidemia; fetal hypoxia; cesarean section
4
1
INTRODUCTION
2
Cardiotocography (CTG) is often used for fetal assessment in obstetric practice; yet,
3
interpreting CTG traces can prove challenging. Extensive research has found that
4
the same CTG trace may elicit inconsistent interpretations between maternity care
5
providers [1-5], which is disconcerting given the impact of CTG traces on clinical
6
decision-making [6]. A CTG trace incorrectly identified as normal delays necessary
7
intervention, potentially increasing risk of hypoxia or metabolic acidosis in the infant
8
[6]. Conversely, a trace incorrectly identified as abnormal may result in unnecessary
9
intervention, such as induction of labor or cesarean delivery.
10
Substantial research has been invested into improving CTG interpretation
11
through concurrent assessment of fetal pulse oximetry [7], lactate level
12
measurements [8], and fetal ECG waveform analysis [9] with limited success [7-9].
13
However, the use of computerized decision support systems may be an underutilized
14
alterative to improve CTG interpretation by synthesizing clinical data to generate
15
alerts and/or recommendations on appropriate interventions.
16
The potential for computerized decision support systems to improve CTG
17
interpretation has resulted in on-going trials [10, 11] and a recent Cochrane review
18
[6]. Yet, despite growing interest [12-14], evidence in this area remains sparse.
19
Therefore, the purpose of this trial was to evaluate the effectiveness of a
20
computerized decision support system, referred to as “quantitative cardiotocography”
21
(qCTG), to facilitate CTG interpretation. The hypothesis of the trial was that the
22
incidence of adverse birth outcomes would be reduced in women monitored with
23
qCTG versus conventional CTG with fetal blood sampling.
24
5
25
METHODS AND MATERIALS
26
Setting and participants
27
A preliminary parallel trial was undertaken between March 2008 and March 2011 at
28
Second Municipal Hospital for Obstetrics and Gynecology Sheynovo, Sofia, Bulgaria,
29
a large (~4,000 deliveries per annum) tertiary maternity hospital. In 2011, the Second
30
Municipal Hospital recorded a perinatal mortality rate of 7 per 1,000 deliveries; the
31
incidence of cesarean delivery was 32%.
32
Eligible participants were women >18 years of age with a singleton pregnancy
33
in cephalic position and no known fetal structural abnormalities. Furthermore, women
34
had to be in active labor. Sample size was based on the incidence of acidosis [15]
35
using one-sided testing, with a 90% confidence level with 80% power. Accounting for
36
potential attrition, approximately 720 women would be required to detect a significant
37
decrease in acidosis incidence from 5.0% in women monitored with conventional
38
CTG to 2.0% in women monitored with qCTG. Ethical approval for this trial was
39
obtained from the Second Municipal Hospital for Obstetrics and Gynecology
40
Research Ethical Committee (Reference: 00134/19.02.2008); all participating women
41
provided informed consent. This trial has been retrospectively registered on Current
42
Controlled Trials website (http://www.controlled-trials.com/; Trial registration
43
identification number: ISRCTN46449237; Registered 02/10/2014).
44
The intervention
45
The intervention was a computerized decision support system to facilitate CTG
46
interpretation. This system, referred to as qCTG, uses external monitoring to
47
synthesize the three domains of a CTG: microfluctuations in fetal heart rate, fetal
48
heart rate and decelerations. Notably, the domain “microfluctuations” is distinct from
49
fetal heart rate variability and refers to the number of extrema per minute, the mean
6
50
beat-to-beat variability per minute and the oscillation amplitudes. These three
51
domains are scored on a scale ranging between zero (normal measure) and six
52
(highly abnormal measure) and summated for an overall CTG score. Thus, the
53
overall CTG score ranges between zero (normal trace) and 18 (pre-terminal trace).
54
Using cordocentesis, Roemer and Walden previously demonstrated a strong
55
correlation between the overall CTG score and fetal pH at delivery [16, 17]. Ignatov
56
et al. undertook additional validation work of the qCTG system and modified the
57
system to enhance prognostic ability [18, 19]. In summary, this validation work
58
identified slight measurement error between qCTG predicted pH values and “true”
59
pH values based on blood gas analysis. When averaging the last six measurements
60
taken prior to delivery, qCTG predicted pH values which ranged from -0.037 to
61
+0.046 relative to the “true” pH value [18]. Moreover, the major parameters of a CTG
62
(microfluctuations in fetal heart rate, fetal heart rate and decelerations) were not
63
equal in terms of their prognostic ability of fetal pH, justifying the evaluation of
64
specific subgroups of parameters [19]. To account for these measurement issues,
65
Ignatov et al. developed qCTG guidelines for clinical application [20] (Table 1).
66
Currently, qCTG is available in the NEXUS / OBSTETRICS system, formerly
67
known as the ARGUS system (Nexus GMT, Frankfurt, Germany), which is one of
68
several recognized fetal monitoring systems. In this system, predicted pH values are
69
calculated and updated every five minutes. As seen in Figure 1, the most recently
70
predicted pH value is displayed in red font on the left side of the interface; previous
71
pH values are represented by red points in the white area below the CTG reading.
72
Microfluctuations in fetal heart rate, fetal heart rate and decelerations (abbreviated
73
as OSZ, FRQ, and DEC respectively) are numerically presented on the lower left
74
side of the interface. 7
75
Prior to the onset of the trial, four obstetricians and the head of the delivery
76
ward received on-site specialized training from Nexus GMT representatives. Overall,
77
seven obstetricians assisted in the implementation of the study. Women were
78
enrolled into the trial by an attending obstetrician in the prenatal unit of the clinic.
79
Women who met the inclusion criteria were randomly assigned to receive
80
conventional CTG monitoring with fetal blood sampling (control group) or qCTG
81
monitoring (intervention group) in a 1:1 ratio with randomly varied permuted block
82
sizes of 10 and 20. The randomization was computer generated by an external
83
statistician. Sequentially numbered, sealed, opaque envelopes were selected in
84
consecutive order by a senior obstetrician in the delivery ward to allocate women to
85
the intervention or control groups. Given the nature of the intervention, neither the
86
women nor the attending obstetricians were blinded to the intervention.
87
In the control group, CTG traces were interpreted according to a modified
88
version of the International Federation of Gynecology and Obstetrics (FIGO)
89
guidelines [21]. In the event of an abnormal CTG trace, fetal blood sampling was
90
performed to assist clinical decision making. A persistent abnormal CTG and/or a
91
fetal scalp blood sample with a pH <7.20 resulted in immediate emergency cesarean
92
delivery.
93
In the qCTG arm, management of labor was conducted in accordance with
94
clinical practice guidelines developed by Ignatov et al. [20] According to these
95
guidelines, if the results from the qCTG were normal, labor was monitored
96
intermittently. If the results were suspicious (i.e. predicted umbilical blood pH level
97
between 7.20 and 7.10), labor was monitored continuously with regular re-
98
evaluation. If the qCTG resulted in persistent abnormal readings (i.e. predicted
99
umbilical blood pH level <7.10), an emergency cesarean delivery was performed.
8
100
In both arms of the trial, qCTG or CTG monitoring was discontinued
101
approximately five to ten minutes before the cesarean delivery or vaginal birth.
102
Outcomes of interest
103
There were four primary outcomes of interest: incidence of hypoxia (defined as a
104
cord-artery blood pH <7.20); incidence of acidemia (defined as umbilical-artery blood
105
pH <7.05); cesarean delivery; forceps extraction. Notably, blood gases were
106
measured in all neonates in both study arms. There were three secondary outcomes
107
of interest: Apgar score less than seven at five minutes; incidence of neonatal
108
seizures; admission to the neonatal intensive care unit (NICU). All outcomes were
109
assessed immediately after birth with the exception of neonatal seizures and NICU
110
admission. The incidence of neonatal seizures and NICU admission were assessed
111
within the first 24-hours post-delivery.
112
Statistical analysis
113
To avoid detection bias, the statistician performing the analysis was blinded to
114
whether the women where in the intervention or the control group. Unadjusted
115
relative risks (RR) and corresponding 95% confidence intervals (95% CI) were
116
derived for all outcomes of interest. Furthermore, a Receiver Operating
117
Characteristic (ROC) curve and area under the curve (AUC) were derived to assess
118
the prognostic ability of qCTG and conventional CTG to identify fetal hypoxia. Fetal
119
hypoxia was selected for this analysis since severe cases, i.e. cases of hypoxic
120
ischemic encephalopathy, is a major cause of infant mortality and chronic disability.
121
Values for the AUC range between zero and one, with a value of one representing
122
perfect prediction and a value of 0.5 representing no predictive value. Analyses were
123
conducted using MedCalc (V12.2.1.0. Ostend, Belgium: MedCalc Software) and
124
SPSS (Version 19.0. Armonk, NY: IBM Corp).
9
125
RESULTS
126
There was no attrition or exclusions of women after randomization. Subsequent
127
results are based on 720 women with 360 women in both the intervention and
128
control arms (Figure 2). Overall, nearly two-thirds (61.8%) of women recruited into
129
the trial were nulliparous, and more than one-third (42.5%) received epidural
130
analgesia (Table 2). The majority of labors (58.8%) were augmented with oxytocin.
131
Obstetric characteristics were similar in both trial arms (Table 2). In the control arm,
132
fetal blood sampling was performed in 38 (10.6%) women to guide CTG
133
interpretation. Multiple fetal blood samples were taken in 15 (4.2%) women.
134
A reduced risk for all primary and secondary adverse outcomes was observed
135
among women monitored using qCTG (Table 3). There was a significant reduction in
136
fetal hypoxia (RR: 0.53; 0.33, 0.84) and acidemia (RR: 0.31; 95% CI: 0.12, 0.84).
137
Moreover, relative to women monitored using conventional CTG with fetal blood
138
sampling, women monitored using qCTG had 0.62 (95% CI: 0.45, 0.85) times the
139
risk of a cesarean delivery. Furthermore, the risk of admission to the NICU was 0.33
140
(95% CI: 0.14, 0.77) in women monitored using qCTG versus conventional CTG. In
141
absolute terms, there would be six fewer cases of fetal hypoxia (95% CI: -10, -1),
142
three fewer cases of acidemia (95% CI: -5, -1), eight fewer cesarean deliveries (95%
143
CI: -14, -3) and three fewer NICU admissions (95% CI: -6, -1) per100 women
144
monitored using qCTG.
145
Marked differences were observed in the sensitivity and specificity of the
146
qCTG and CTG with fetal blood sampling in the identification of fetal hypoxia
147
(Figures 3 and 4). The sensitivity in the identification of hypoxia was 97.6% in
148
women monitored using qCTG compared to 87.2% in women monitored using
149
conventional CTG with fetal blood sampling. Moreover, specificity in the identification
10
150
of hypoxia was 89.2% in women monitored with qCTG compared to 59.7% in women
151
monitored with conventional CTG with fetal blood sampling.
152
COMMENT
153
The incidence of fetal hypoxia, acidemia, cesarean delivery and admission to the
154
NICU was significantly lower among women who were monitored using qCTG versus
155
women who were monitored using conventional CTG monitoring with fetal blood
156
sampling. These preliminary trial results are promising and suggest that
157
computerized decision support systems may be able enhance conventional CTG
158
interpretation. If such findings are supported in other clinical trials, a movement
159
towards computerized decision support systems for fetal assessment could have
160
profound effects on obstetric medicine. When an abnormal CTG trace is observed in
161
current practice, fetal blood sampling is often performed to assess fetal pH levels.
162
However, fetal blood sampling can only be performed at intermittent time points, in
163
essence providing cross-sectional data. Time gaps between fetal blood samples may
164
not capture the initial decline in fetal pH, consequently delaying timely diagnosis of
165
hypoxia and appropriate obstetric interventions. Fetal blood sampling also requires a
166
certain degree of cervical effacement, ruptured membranes, absence of vaginal
167
infection, and trained staff to perform the procedure; these characteristics are not
168
present in all deliveries. By providing continuous real-time predicted pH values
169
irrespective of cervical condition and membrane integrity, qCTG circumvents these
170
aforementioned issues.
171
In this study, the sensitivity for the identification of hypoxia in the conventional
172
CTG with fetal blood sampling group was 87.2%, which is in line with previous
173
research [17, 22-26]. To date, conventional CTG monitoring has failed to deliver
174
consistent results in respect to the specificity, with reports varying between 9% and
11
175
63% [27-29]. With a specificity of 59.7%, this study was of no exception. In contrast,
176
whereas both sensitivity and specificity for the identification of hypoxia improved in
177
women monitored using qCTG, specificity was strikingly higher.
178
Given the non-invasive nature of the intervention, no secondary harms were
179
anticipated during the design phase of the trial. Future trials, however, may wish to
180
assess maternal dissatisfaction as a possible source of unintended consequences
181
as well as longitudinal outcomes, such as developmental delay.
182
Several limitations should be noted. This trial was underpowered to examine
183
the breadth of relatively rare (<10% incidence rate) adverse birth outcomes, such as
184
perinatal death. Notably, no perinatal deaths occurred in either trial arm. However, if
185
the trial were prolonged, there remains the possibility that there may be a higher risk
186
for these rarer events. Early-phase studies of experimental interventions, such as
187
qCTG, may justify underpowered trials [30]. Despite the small sample size, there
188
were marked decreases the incidence of several outcomes of interest in women
189
monitored using qCTG, underscoring the potential for this technology. These data
190
will further contribute to systematic reviews in this field of study [6].
191
Secondly, this trial compared a novel fetal monitoring technology
192
(intervention) to a well-established fetal monitoring approach (conventional CTG with
193
fetal blood pH assessment, control). Reproducibility in other countries may be
194
difficult if their respective ethical committees deemed the unknown risk of this new
195
technology to outweigh the potential benefits. In the context of this preliminary trial,
196
careful monitoring of participants in conjunction with an interim analysis based on
197
220 women [15] was undertaken to avoid detrimental outcomes.
198
Whereas indication for cesarean delivery was typically fetal distress, there
199
were medical exceptions to ensure the health of the woman and infant. In several
12
200
cases of caesarean delivery, fetal blood samples were insufficient for analysis, and
201
given the presence of abnormal tracings, there was no time to repeat the procedure.
202
Furthermore, in some cases effacement was not sufficient, membranes were not
203
ruptured or vaginal infection was possible. Due to the presence of one or more of
204
these circumstances, blood sampling could not always be performed. However,
205
these cases were not omitted from the analysis because they represent realistic
206
limitations.
207
Furthermore, fetal blood sampling was only made available in the control arm
208
of the trial. Previous work had confirmed a strong correlation between actual and
209
predicted fetal pH levels [18]. Thus, this procedure was not performed in the
210
intervention arm.
211
Neither the women nor the obstetricians were blinded to the intervention,
212
potentially introducing performance bias. However, given the nature of the
213
intervention, blinding was not feasible.
214
The study had a relatively long recruitment period which may have resulted in
215
a sampling bias. The recruitment period was due to funding issues largely related to
216
all neonates requiring measurements for blood gases. Sheynovo is a municipality
217
hospital, and each year there was a three to six month wait for the funding body to
218
allocate resources. Despite the long recruitment period, arguably this is a minimal
219
source of bias.
220
Lastly, it is important to acknowledge that this trial was retrospectively
221
registered, which is often viewed as a potential source of reporting bias. This work
222
was originally carried out to improve obstetric outcomes in Bulgaria and, given the
223
target audience, trial registration was not originally sought. However, this research
224
has already led to changes in local obstetric practice and in light of the All Trials
13
225
campaign, which promotes the registration and publication of all performed trials
226
(www.alltrials.net), this trial was registered to increase the transparency and
227
accessibility of this study. Although all outcomes of interest in this trial were reported,
228
to avoid the risk of reporting bias, future trials should ideally always be prospectively
229
registered – irrespective of the trial size, geographic location or intended audience.
230
One major strength of this trial is that it is among one of the first publications
231
investigating the utility of computerized decision support systems for fetal monitoring
232
in labor and will complement future trials in this area for numerous reasons [15].
233
Firstly, not all computerized decision support systems will be based on identical
234
criteria. As more research develops in this area [10, 11], comparisons between
235
different systems will be necessary to identify the best performing computerized
236
decision support system. Secondly, the clinical practice guidelines used in this trial
237
include several improvements which enhance the correlation between prognostic
238
and actual fetal pH values [18, 19]. This allows for immediate recognition of any
239
significant changes in fetal pH, resulting in better timing of needed procedures.
240
Lastly, a potential bias often cited in research is that randomized control trials are
241
geographically restricted to certain regions, such as Western Europe or North
242
America. This study, performed in Eastern Europe (Bulgaria), represents output from
243
a historically under-represented country with a distinct maternity profile relative to
244
many of its Western counterparts. For instance, the national rate of cesarean
245
delivery is approximately 40%, with some private hospitals reporting rates of nearly
246
60% [31]. A recent report identified that two out of five women reported a clear
247
preference for cesarean delivery [32]. Moreover, based on local hospital observation,
248
the incidence of acidemia was notably higher than other settings. This has been
249
attributed, in part, to the large number of high risk referrals. Such underlying
14
250
conditions can influence new obstetric interventions, underscoring the need for
251
globally representative research.
252
Improving CTG monitoring during labor is a well-recognized goal in obstetric
253
medicine. This trial supports that qCTG has the potential to significantly reduce the
254
incidence of fetal hypoxia, acidemia, cesarean delivery and admission to the NICU in
255
comparison to conventional CTG monitoring with fetal blood sampling. Prior to the
256
adoption of qCTG in daily clinical practice, further large-scale randomized control
257
trials are justified as well as evaluation of different computerized decision support
258
systems for fetal monitoring.
15
REFERENCES (1) Chauhan SP, Klauser CK, Woodring TC, Sanderson M, Magann EF, Morrison JC. Intrapartum nonreassuring fetal heart rate tracing and prediction of adverse outcomes: interobserver variability. Am J Obstet Gynecol. 2008;199:623 e1-5. (2) Devane D, Lalor J. Midwives' visual interpretation of intrapartum cardiotocographs: intra- and inter-observer agreement. J Adv Nurs. 2005;52:133-41. (3) Figueras F, Albela S, Bonino S, Palacio M, Barrau E, Hernandez S, et al. Visual analysis of antepartum fetal heart rate tracings: inter- and intra-observer agreement and impact of knowledge of neonatal outcome. J Perinat Med. 2005;33:241-5. (4) Blix E, Sviggum O, Koss KS, Oian P. Inter-observer variation in assessment of 845 labour admission tests: comparison between midwives and obstetricians in the clinical setting and two experts. BJOG. 2003;110:1-5. (5) Ayres-de-Campos D, Bernardes J, Costa-Pereira A, Pereira-Leite L. Inconsistencies in classification by experts of cardiotocograms and subsequent clinical decision. Br J Obstet Gynaecol. 1999;106:1307-10. (6) Lutomski JE, Meaney S, Greene RA, Ryan AC, Devane D. Expert systems for fetal assessment in labour. Cochrane Database of Systematic Reviews 2015, Issue 4. Art. No.: CD010708. DOI: 10.1002/14651858.CD010708.pub2. (7) East CE, Chan FY, Colditz PB, Begg LM. Fetal pulse oximetry for fetal assessment in labour. Cochrane Database Syst Rev. 2007:CD004075. (8) East CE, Leader LR, Sheehan P, Henshall NE, Colditz PB. Intrapartum fetal scalp lactate sampling for fetal assessment in the presence of a non-reassuring fetal heart rate trace. Cochrane Database Syst Rev. 2010:CD006174. (9) Neilson JP. Fetal electrocardiogram (ECG) for fetal monitoring during labour. Cochrane Database Syst Rev. 2012;4:CD000116.
16
(10) Brocklehurst PA. A multicentre randomised controlled trial of an intelligent system to support decision making in the management of labour using the cardiotocogram (INFANT). Current Controlled Trials (www.controlledtrials. com/mrct) [accessed 6 September 2014]. 2013. (11) Ayres-de-Campos D, Ugwumadu A, Banfield P, Lynch P, Amin P, Horwell D, et al. A randomised clinical trial of intrapartum fetal monitoring with computer analysis and alerts versus previously available monitoring. BMC Pregnancy Childbirth. 2010;10:71. (12) Schiermeier S, Pildner von Steinburg S, Thieme A, Reinhard J, Daumer M, Scholz M, et al. Sensitivity and specificity of intrapartum computerised FIGO criteria for cardiotocography and fetal scalp pH during labour: multicentre, observational study. BJOG. 2008;115:1557-63. (13) Costa A, Santos C, Ayres-de-Campos D, Costa C, Bernardes J. Access to computerised analysis of intrapartum cardiotocographs improves clinicians' prediction of newborn umbilical artery blood pH. BJOG. 2010;117:1288-93. (14) Greene KR. Intelligent fetal heart rate computer systems in intrapartum surveillance. Curr Opin Obstet Gynecol. 1996;8:123-7. (15) Ignatov P, Atanasov B. [Indirect standard cardiotocography plus fetal blood sampling versus indirect quantitative cardiotocography--a randomized comparative study in intrapartum monitoring]. Akush Ginekol (Sofiia). 2012;51:3-10. (16) Roemer VM. [CTG: microfluctuation]. Z Geburtshilfe Neonatol. 2004;208:210-9. (17) Roemer VM, Walden R. [Quantitative Cardiotocography--what does it look like and what can we expect]. Z Geburtshilfe Neonatol. 2006;210:77-91. (18) Ignatov P, Atanasov B, Kostov I, Velev R, Kovacheva A, Dobreva A. [A comparative study of the prognostic pH values of the fetus in utero, generated by the
17
"quantitative cardiotocography" computer method, and the actual pH values measured immediately after delivery]. Akush Ginekol (Sofiia). 2010;49:3-11. (19) Ignatov P, Atanasov B. [Structure and function of the cardiotocographic score (CTG-score) calculated by the "quantitative cardiotocography" computer method. Determining the significance of its components for the accuracy of the estimates for the ph of the fetus]. Akush Ginekol (Sofiia). 2011;50:13-20. (20) Ignatov P, Atanasov B. ["Quantitative cardiotocography"--clinical practice guideline]. Akush Ginekol (Sofiia). 2011;50:3-9. (21) Su LL, Chong YS, Biswas A. Use of fetal electrocardiogram for intrapartum monitoring. Ann Acad Med Singapore. 2007;36:416-20. (22) International Federation of Gynecology and Obstetrics. Intrapartum surveillance: recommendations on current practice and overview of new developments. FIGO Study Group on the Assessment of New Technology. Int J Gynaecol Obstet. 1995;49:213-21. (23) Brown VA, Sawers RS, Parsons RJ, Duncan SL, Cooke ID. The value of antenatal cardiotocography in the management of high-risk pregnancy: a randomized controlled trial. Br J Obstet Gynaecol. 1982;89:716-22. (24) Goeschen K. Derzeitiger Stand der intrapartalen U berwachung des Kindes. Gynakologe. 1997;30. (25) Schneider H. Evaluation des CTG. Kritische Evaluation des CTG 1996;29:3-11. German. (26) Seelbach-Göbel B, Huch R, Luttkus A, Saling E, Vetter K. Ist die Pulsoxymetrie ein Gewinn für die überwachung des Feten sub partu? Perinatalmedizin. 1998;10:77-80. German.
18
(27) Steer PJ, Eigbe F, Lissauer TJ, Beard RW. Interrelationships among abnormal cardiotocograms in labor, meconium staining of the amniotic fluid, arterial cord blood pH, and Apgar scores. Obstet Gynecol. 1989;74:715-21. (28) Murphy KW, Johnson P, Moorcraft J, Pattinson R, Russell V, Turnbull A. Birth asphyxia and the intrapartum cardiotocograph. Br J Obstet Gynaecol. 1990;97:4709. (29) Axt R. Das intrapartale CTG. Gynakologe. 1997;30:577-80. German. (30) Halpern SD, Karlawish JH, Berlin JA. The continuing unethical conduct of underpowered clinical trials. JAMA. 2002;288:358-62. (31) Konstantinov S, Zlatkov V. [Types of hospital property and the relative rate of cesarean section occurrence]. Akush Ginekol (Sofiia). 2015;54:8-13. (32) Dimitrov A, Tsankova M, Krusteva K, Nikolov A. [Study of women preference regarding mode of delivery]. Akush Ginekol (Sofiia). 2004;43:13-7.
19
ACKNOWLEDGEMENTS PNI contributed to the conception, design and interpretation of data, drafting of the manuscript and critically revised the manuscript for intellectual content; PNI is guarantor. JEL contributed to the interpretation of data, drafting of the manuscript and critically revised the manuscript for intellectual content.
FUNDING This trial was funded by the Bulgarian Christmas 2013-2014 Charity Initiative and Sheynovo - Second Municipal Hospital for Obstetrics and Gynaecology. CONFLICTS OF INTEREST PNI has no conflicts of interest to declare. Petar N. Ignatov assisted in the development of the reported quantitative cardiotocography system and is currently developing a web-based software interface. JEL is the lead author of a Cochrane Systematic Review, entitled “Expert Systems for Fetal Assessment”.
20
Table 1: Clinical application guidelines for quantitative cardiotocography (qCTG) CTG Classification
Predicted pH values based on qCTGa
CTG-score componentsb
Management
Normal
7.350 – 7.237
All
Expectant Intermittent monitoring
OSZ OSZ+FRQ
Expectant Continuous monitoring
FRQ Suspicious
7.237 – 7.137
OSZ+DEC DEC FRQ+DEC OSZ+FRQ+DEC OSZ OSZ+FRQ FRQ
Abnormal
≤ 7.137
<30 mins Continuous monitoring >30 mins Urgent delivery <30 mins Continuous monitoring >30 mins Urgent delivery
OSZ+DEC DEC FRQ+DEC
Urgent delivery
OSZ+FRQ+DEC Abbreviations: CTG, cardiotocography; OSZ, microfluctuation; FRQ, fetal heart rate; DEC, decelerations a Although obstetric guidelines typically defined hypoxia as cord-artery blood pH <7.20 and acidemia as umbilical-artery blood pH <7.05, these thresholds were modified in this guideline to account for potential measurement error in qCTG. The degree of measurement error has been found to range between -0.037 and +0.046 (when compared to the reference standard of blood gas analyses). b OSZ, OSZ+FRQ and FRQ subgroups of the CGT-score were found to be less accurate in predicting pH levels compared to the remaining subgroups [19]. Therefore, no immediate actions are advised when these are present. Expectant management is advocated if suspicious findings are observed based on these less accurate subgroups of the CGT-score. If abnormal prognostic pH readings are observed based on OSZ, OSZ+FRQ and FRQ and they persist longer than 30 minutes, urgent delivery is recommended.
21
Table 2: Obstetric characteristics of included women qCTG CTG + FBS (N=360) (N=360) N (%) n (%) Age (range) 18-42 18-44 Nulliparous 211 (58.6) 234 (65.0) Gestational weeks 37 to 40 346 (96.1) 351 (97.5) +1 40 to 42 14 (3.9) 9 (2.5) Epidural analgesia 145 (40.3) 161 (44.7) Oxytocin augmentation 201 (55.8) 222 (61.7) Birth weight <2500g 21 (5.8) 17 (4.7)
Overall trial (N=720) n (%) 18-44 445 (61.8) 697 (96.8) 23 (3.2) 306 (42.5) 423 (58.8) 38 (5.3)
Abbreviations: qCTG, quantitative cardiotocography with decision support system; CTG + FBS, conventional cardiotocography with fetal blood sampling
22
Table 3: Risk estimates of primary and secondary outcomes qCTG CTG + FBS (N=360) (N=360) n (%) n (%) Primary outcomes Hypoxia 25 (6.9) 47 (13.1) Acidemia 5 (1.4) 16 (4.4) Cesarean delivery 50 (13.9) 81 (22.5) Forceps extraction 2 (0.6) 4 (1.1) Secondary outcomes Apgar score <7 at 5 minutes 9 (2.5) 18 (5.0) Neonatal seizures 3 (0.8) 5 (1.4) Admission to NICU 7 (1.9) 21 (5.8)
Relative Risk (95% CI)
0.53 (0.33, 0.84) 0.31 (0.12, 0.84) 0.62 (0.45, 0.85) 0.50 (0.09, 2.71) 0.50 (0.23, 1.10) 0.60 (0.14, 2.49) 0.33 (0.14, 0.77)
Abbreviations: qCTG, quantitative cardiotocography with decision support system; CTG + FBS, conventional cardiotocography with fetal blood sampling; CI, confidence interval; NICU, neonatal intensive care unit Note: Hypoxia was defined as a cord-artery blood pH <7.20. Acidemia was defined as umbilical-artery blood pH <7.05.
259
23
Figure 1: Illustrative example of quantitative cardiotocography (qCTG) interface 260 261
Please find attached separately
262
24
Figure 2: Participant flow chart
Figure 3: Receiving Operator Characteristic curve for identification of hypoxia using cardiotocography with computerized decision support system (qCTG) 266
Please find attached separately Note: Graph reflects predicted pH values generated by qCTG versus pH value determined by
blood gas analyses (reference standard). Data are based on a continuous scale. In both arms of the trial, qCTG or CTG monitoring was discontinued approximately five to ten minutes before the cesarean delivery or vaginal birth.
26
Figure 4: Receiving Operator Characteristic curve for identification of hypoxia using conventional cardiotocography (CTG) 267
Please find attached separately Note: Graph reflects likely hypoxia based on conventional CTG traces (yes/no)
versus hypoxia determined by blood gas analyses (reference standard). Data are dichotomous. In both arms of the trial, qCTG or CTG monitoring was discontinued approximately five to ten minutes before the cesarean delivery or vaginal birth.
27
Figure 1: Illustrative example of quantitative cardiotocography (qCTG) interface
260 261
24
Figure 2: Participant flow chart 262 263 264
Enrollment
Assessed for eligibility (n=765)
Excluded (n= 45) Did not meet inclusion criteria (n=35) Declined to participate (n=3) Other reasons (n=7)
Randomized (n=720)
Allocation Allocated to intervention (n=360) Received allocated intervention (n=360)
Allocated to intervention (n=360) Received allocated intervention (n=360)
Did not receive allocated intervention (n=0)
Did not receive allocated intervention (n=0)
Follow-Up Lost to follow-up (Non-applicable due to short time frame) Discontinued intervention (n=0)
Lost to follow-up (Non-applicable due to short time frame) Discontinued intervention (n=0)
Analysis Analysed (n=360) Excluded from analysis (n=0)
Analysed (n=360) Excluded from analysis (n=0)
25
Figure 3: Receiving Operator Characteristic curve for identification of hypoxia using cardiotocography with computerized decision support system (qCTG)
Note: Graph reflects predicted pH values generated by qCTG versus pH value determined by
blood gas analyses (reference standard). Data are based on a continuous scale. In both arms of the trial, qCTG or CTG monitoring was discontinued approximately five to ten minutes before the cesarean delivery or vaginal birth.
26
Figure 4: Receiving Operator Characteristic curve for identification of hypoxia using conventional cardiotocography (CTG)
Note: Graph reflects likely hypoxia based on conventional CTG traces (yes/no) versus
hypoxia determined by blood gas analyses (reference standard). Data are dichotomous. In both arms of the trial, qCTG or CTG monitoring was discontinued approximately five to ten minutes before the cesarean delivery or vaginal birth.
27