Main text
Introduction
The
rationale for the proposed link between excessive uterine activity (UA)
and fetal hypoxia is based on physiological studies of the haemodynamic
changes that occur in the utero-placental and fetal circulations during
a contraction. As a uterine contraction progresses, the amount of
oxygenated blood delivered to the placenta decreases. Thus, the lowest
FSpO2 values are found towards the end of a contraction
and take some time to recover. Increased contraction frequencies may not
allow adequate recovery time and may result in progressive reductions in
fetal oxygen levels. (1-3)
There is evidence that tocolytic medications may improve fetal heart
rate (FHR) abnormalities when used while emergency delivery is pending.
(4) Compared to emergent delivery,
tocolysis for fetal distress may improve umbilical artery (UmA) base
excess (BE) values and reduce neonatal intensive care unit admission at
the expense of increasing caesarean delivery.
(5) A 2018 Cochrane review found several
studies which showed an improvement in fetal wellbeing in response to
tocolysis but concluded that, given the small sample sizes involved,
“the clinical significance is unclear.
(6)
Current guidelines are based on the opinion that when there is an
abnormal FHR pattern and tachysystole (TS), increased UA may cause fetal
hypoxia. (7) The 2014 American College of
Obstetricians and Gynaecologists (ACOG) guidelines, Neonatal
Encephalopathy and Neurologic Outcome, state that TS should be treated
whenever it is associated with recurrent FHR decelerations or if
oxytocin is being administered. (8) The UK
National Institute for Health and Care Excellence guidelines on labour
management state that where the FHR pattern is suspicious or
pathological any uterine hyperstimulation should be corrected.
(9) There are no existing guidelines which
address TS in unaugmented labour without FHR abnormalities.
The aim of this systematic review is to determine if increased UA during
labour is associated with an increased likelihood of either short- or
long-term neurological injury in term neonates, or with proxy measures
of intrapartum hypoxia-ischemia.
Methods
The study protocol was registered with PROSPERO (CRD4201705258). The
study followed the Preferred Reporting Items for Systematic Reviews and
Meta‐Analyses (PRISMA) statement.
Search Strategy
We searched MEDLINE, CINAHL, and ClinicalTrials.gov using terms related
to UA (“uterine activity”, “excessive uterine activity”, “XSUA”,
“uterine hyperstimulation”, and “tachysystole”). The publication
start date was 1st January 1996. Reference lists of
relevant papers were checked to identify further papers for
consideration. The results were de-duplicated using COVIDENCE (Covidence
systematic review software, Veritas Health Innovation Ltd., Melbourne,
Australia). The search was repeated prior to final analysis.
Eligibility Criteria
Eligibility criteria were pre-determined. Based on the anticipated
heterogeneity and small size of the evidence base, we opted for
permissive eligibility criteria. Any study that analysed the
relationship between UA during term labour and neurological outcomes or
selected proxy outcomes was eligible. The proxy outcomes were selected
because they are commonly used indicators of intrapartum
hypoxia-ischemia. (10) Based on
incomplete reporting, conference abstracts were excluded. Studies
retrieved were screened by one author and potentially eligible studies
were then assessed by two team members before a final inclusion
decision.
Data Extraction and Evidence Summary
A standardised form was used to extract data. Missing data was requested
from authors and included if returned. Outcomes from individual studies
were reported in tables and presented descriptively with odds ratios
(OR) and 95% confidence intervals (CI) for dichotomous outcomes and
means with standard deviations (SD) for continuous outcomes where
available. Where group numbers were provided, ORs and CIs were
calculated according to Altman. (11)
Results adjusted based on multivariate analysis were included as
reported. A post-hoc analysis of the relationship between uterine
activity and both FHR patterns and delivery mode was also performed.
Quality Assessment
The risk of bias was independently assessed by both reviewers using The
Scottish Intercollegiate Guideline Network (SIGN) critical appraisal
checklists but was not used to exclude any results.
Funding
Funding was provided by The National Women and Infants Health Programme
(Ireland) and the Rotunda Foundation (Registered Charity Number.
20079529).
Results
Search Results and Study
Selection
Our search retrieved 1,777
citations from MEDLINE, CINAHL. (Table S1) Four additional articles were
identified after review of references. The search result details are
specified in supplementary material. A search of ClinicalTrials.gov did
not reveal any unpublished but otherwise eligible studies. Two
conference abstracts with possibly eligible results were identified but
not included owing to the incomplete nature of the reports.
(12, 13)
After screening of abstracts, 21 articles were selected for full-text
review. Of these, nine were excluded.(3,
14-21) (Figure 1)
Study characteristics
The characteristics of each included study are summarized in Table S2.
Twelve studies met the inclusion criteria.
(22-33) In five
(22, 23,
26, 31,
32), the primary analysis was grouped by
exposure i.e. UA. In three (24,
30, 33),
cases were identified and then selectively matched to controls. For the
other nine studies, all eligible patients were recruited regardless of
outcome (median subject number: 1,433 [interquartile range: 720-8,008,
range: 430-50,335]). In three (23,
25, 27)
data was collected prospectively with the primary purpose being to
compare UA to fetal outcomes. In two (34,35)
(31, 32),
data was collected prospectively but the included studies were secondary
analyses. The data collected pertained to labours from 1993 to 2015.
Subjects were recruited in North America (7 studies), Sweden, Turkey,
Italy, The Netherlands, and Ireland (1 study each).
Included labours were exclusively spontaneous and unaugmented
(26), exclusively induced/augmented
(23, 25,
27, 32),
or mixed (22,
24, 28-31,
33).
One study (29) exclusively used
measurements from intra-uterine pressure catheters (IUPC). Three studies
(27, 32,
33) used external tocography data only.
Two studies (23,
30) reported using IUPC and external
tocography but mainly external measurements. Six studies
(22,
24-26, 28,
31) did not specify the measurement
technology.
Three studies (28-30) used automated
analysis. Four studies (22,
26, 27,
33) used documentation recorded during
labour by the attending midwives The remainder
(23-25,
31, 32)
reported retrospectively analysed recordings, the interpretation of
which was stated to be blinded in all except one
(32).
The period of labour analysed varied significantly between studies.
Three (26,
27, 33)
did not explicitly state the period included. Two
(25, 32)
analysed the entire period from induction to delivery. One
(24) analysed data from the first 4 hours
after induction. The remainder analysed a period leading up to delivery
(Bakker: last hour of 1st stage and total
2nd stage, Hamilton: 4h, Smith: ≤4h, Heuser: ≥2h,
Jonsson: 2h, Palanisamy: 1h).
Eleven studies featured contraction rate as an individual exposure, with
one (27) reporting rate only as part of a
combined exposure with contraction duration. All studies apart from
Bakker et al. reported contraction rate as either a dichotomised or
categorised variable. Two studies (29,
30) included contraction rate as a
continuous variable. All studies reported dichotomised outcomes only
apart from one study (27) which reported
UmA pH and BE as continuous measures.
Six of the ten included studies published after 2008 adhered precisely
or closely to the ACOG 2008 definition of TS i.e. more than five
contractions in 10 minutes, averaged over a 30-minute window. Of the
four that did not follow the ACOG definition: Hayes et al. referred to
15-minute contraction rates >7, Mutli Meydanli et al.
reported a composite exposure which included the ACOG definition along
with any contraction longer than 2-minutes, Bofill et al. required two
consecutive 10-minute windows with ≥6 contractions each, and Stewart et
al. reported the maximum 10-minute contraction rate.
Two studies (23,
29) included non-composite exposures
besides contraction rate. Stewart et al. reported the number of
contractions >120s. Bakker et al. reported on relaxation
time as well as contraction duration, amplitude, and surface (mean and
total), as well as Montevideo units, and active planimeter units (mean).
Risk of bias within studies
Of the twelve included studies, three were classed as at high risk of
bias, six were classed as at medium risk, and three were classed as at
low risk. (Tables S3-S14).
Synthesis of results
The results of the individual studies are summarized in Table S15.
Neurodevelopmental Outcome
No studies were found which reported neurodevelopmental outcomes in
relation to UA.
Hypoxic Ischemic Encephalopathy (HIE)
One study reported HIE as an outcome. In Hayes et al. contraction rates
greater than 7 in 15 minutes were associated with an increased risk of
neonatal encephalopathy (TS: 87/187 [46.5%], No TS: 120/457
[26.3%], OR: 2.44 [95% CI: 1.71-3.48], aOR: 2.07 [95% CI:
1.13 to 3.81]). (33)
Umbilical Artery pH
Seven studies featured UmA pH as an individual outcome. Three rejected
and four retained the null hypothesis that UA is not associated with UmA
pH levels. Jonsson et al. compared cases with UmA pH <7.05 to
controls with pH≥7.05 and 5-minute Apgar scores ≥5.
(24) Tachysystole (TS) in the last two
hours before delivery was found to be associated with the likelihood of
an adverse outcome (TS: 84/127 [66.1%], No TS: 132/469
[28.1%], OR: 4.99 [95% CI: 3.28-7.58]). Mutlu Meydanli et al.
found that TS was associated with a 7.1 (95% CI: 1.3–38.7) times
increase in the relative risk of a UA pH ≤7.15 in women after
misoprostol induction (Bishop’s score ≤4).
(25) Bakker et al. found that UmA pH
levels ≤7.11 were associated with higher contractions per 10-minutes as
measured by intrauterine pressure catheters both in the first (mean: 5.0
[SD: 0.7] versus 4.8 [0.7], p=0.006) and second stages (mean:
5.5 [SD: 0.9] versus 5.2 [0.9], p=0.002).
(29) Bofill et al. reported that TS after
cervical ripening was not associated with mean UmA pH levels (≥3
episodes: 7.22 [SD: 0.08], No episodes: 7.22 [SD: 0.08],
p=0.435), or with the likelihood of a UmA pH <7.0 (TS: 3/131
[2.3%], No TS: 8/631 [1.3%], OR: 1.83 [95% CI:
0.48-6.97]). Stewart et al. reported that the maximum contraction rate
in the first four hours after induction was not associated with an
increased risk of acidosis (≤4: 1/152 [1%], 5: 4/179 [2%], 6:
2/134 [1%], ≥7: 6/119 [5%], p for trend=0.06, OR for
>5: 2.20 [95% CI: 0.71-6.81]). Mussi et al. found that
the presence of TS or prolonged contractions was not linearly correlated
with UmA pH (r=0.006, p=0.8). (27)
Palanisamy et al. found that TS in the last hour of labour was not
associated with UmA pH levels (TS: 9/513 [1.8%], No TS: 140/8067
[1.7%], OR: 1.01 [95% CI: 0.51-2.00]).
(31)
Umbilical Artery Base Excess
Four studies tested whether UmA BE was linked to UA with two rejecting
the null hypothesis. Mussi et al. found that the presence of TS or
prolonged contractions was not linearly correlated with the UmA BE
(r=0.15, p=0.07). Hamilton et al. compared labours with UA BE
<-12 to labours with UA BE >-8. The found that TS
was associated with acidaemia (TS: 138/1215 [11.4%], No TS:
178/2105 [8.5%], OR: 1.39 [95% CI: 1.10-1.75]).
(30) In those with TS, the duration of TS
was not associated with metabolic acidosis. Bofill et al. found that
labours with ≥3 20-minute episodes of TS did not have lower mean BE
values than labours without TS (mean BE: -6.81 [SD: 4.18] versus
-6.57 [4.0], p=0.535). Palanisamy et al. reported that TS in the
last hour before delivery was more common in labours with a BE ≤-8 (TS:
33/513 [6.4%], No TS: 315/8067 [3.9%], OR: 1.69 [95% CI:
1.17-2.45]).
Umbilical Artery Lactate
One study reported UmA lactate values. Palanisamy et al. observed that
UmA lactate values ≥4 were more common when the last hour before
delivery was complicated by TS (TS: 173/513 [33.7%], No TS:
2032/8067 [25.2%], OR: 1.51 [95% CI: 1.25-1.83]).
Apgar Score
Four of nine cohort studies included Apgar score as an individual
outcome. One rejected and three retained the null hypothesis. Heuser at
al. reported that documentation of >5 contractions in 10
minutes averaged over 30 minutes was associated with an increased rate
of 5-minute Apgar scores <7 (TS: 61/5363 [1.1%], No TS:
336/44972 [0.7%], OR: 1.53 [95% CI: 1.16-2.01]).
(22) Stewart et al. reported that maximum
contraction rate was not associated with an increased risk of a
five-minute Apgar ≤3 (≤4: 0/152 (0%), 5: 0/134 (0%), 6: 1/179 (1%),
≥7: 0/119 (0%), p for trend=0.86). (23)
Ahmed at al. found that TS (> 5 contractions in 10 minutes
averaged over a 30-minute window) was not associated with an Apgar score
<7 at 5 minutes (No TS: 14/7118 [0.2%], TS: 2/890
[0.2%], p=0.897). (26) Bofill et al.
reported that TS was not associated with median 5-minute Apgar scores
(≥3 TS episodes: 9 [IQR: 8-9], No TS: 9 [IQR: 8-9], p=0.502) or
with the risk of 5-minute Apgar scores ≤3 (≥3 TS episodes: 0/131
[0.0%], No TS: 3/631 [0.5%]).
(32)
Composite Outcomes
Two studies featured composite outcomes which were composed of eligible
outcomes but could not be separated based on available data. Smith et
al. compared cases with an UmA BE ≤10mmol/L or a 5-minute Apgar score ≤6
to controls without those characteristics. Using automated analysis of
external tocography recordings from up to 4 hours before delivery, they
found that TS regardless of duration was not more common in cases than
controls (No TS: 66/5095 [1.3%], Any TS: 11/1139 [1.0%],
p=0.45). (28) Mussi et al. found that TS
or prolonged contractions were not associated with either a UmA pH
<7.1 or a UmA BE >-10 (No TS: 11/339 [3.2%],
TS: 5/91 [5.5%], p=0.35)
Figure 2 shows unadjusted odds ratios for each dichotomous outcome in
the presence of TS. Mutlu Meydanli et al. was excluded from the graph as
group numbers were not
reported.
Fetal Heart Rate Assessments
Three of the twelve included studies compared FHR patterns in labours
with TS to those without TS. All rejected the null hypothesis. Stewart
et al. reported that the maximum 10-minute contraction rate was related
to the presence of variable, late, or prolonged decelerations (Any
decelerations: ≤4: 61/152 [40%], 5: 88/179 [49%], 6: 77/134
[57%], ≥7: 69/119 [58%] p for trend <0.0.001, OR for
>5: 1.67 [95% CI: 1.20-2.32]). Ahmed et al. found that
the rate of non-reassuring FHR (NRFHR) traces in labours with TS was
higher than those labours without TS (TS: 38/890 [4.3%], no TS:
179/7118 [2.5%], OR: 1.73 [95% CI: 1.21-2.47]). Bofill et al.
found higher rates of NRFHR patterns with TS compared to without (≥3 TS
episodes: 28/131 [21%], No episodes: 91/631 [14%], OR: 1.61
[95% CI: 1.01-2.59]).
Uterine Activity and Delivery Methods
Three of the included studies reported on TS in relation to caesarean
delivery for any indication. None found a statistically significant
association. Heuser at al. reported that TS for any half hour period in
the last two hours of labour was not associated with caesarean delivery
(TS: 670/5363 [12.5%], No TS: 5240/44972 [11.7%], OR: 1.08
[95% CI: 0.99-1.18]). Stewart at al. reported that TS in the first
four hours after induction was not associated with caesarean delivery
(TS: 43/253 [17.0%], No TS: 66/331 [19.9%], OR: 0.82 [95%
CI: 0.54-1.26]). Bofill et al. found that TS at any stage after
induction was not associated with caesarean delivery (TS: 37/131
[28.2%], No TS: 215/631 [34.1%], OR: 0.76 [95% CI:
0.50-1.15]).
Two studies reported data on caesarean delivery for FHR trace findings.
Ahmed et al. found TS in spontaneous unaugmented labours to be
associated with an increased risk of caesarean delivery for NRFHR traces
(TS: 58/890 [6.5%], No TS: 318/7118 [4.5%], OR: 1.49 [95%
CI: 1.12-1.99]). Mutlu Meydanli et al. reported a statistically
insignificant increase in the risk of caesarean delivery for NRFHR
traces (RR: 2.4 [95% CI: 0.6-9.4]).
Heuser et al. found that TS was associated with an increased likelihood
of operative vaginal delivery (TS: 682/5363 [12.7%], No TS:
3491/44972 [7.8%], OR: 1.73 [95% CI: 1.59-1.89]).
Results for Exposures Other Than Contraction Rate
With regard to the last hour of the first stage, Bakker at al. found
that UmA pH <7.11 was associated with average relaxation times
(cases: 51s±23, controls: 63s±35, p<0.001), average Montevideo
units (cases: 261±86, controls: 236±97, p=0.02), and average active
planimeter units (cases: 9014±2461, controls: 8379±2740, p=0.04), but
not with average contraction duration (cases: 87s±9, controls: 87s±10,
p=0.68), average contraction amplitude (cases: 54mmHg±16, controls:
51mmHg±19, p=0.165), or average contraction surface (cases:
1875mmHg*s±555, controls: 1798±600, p=0.26). Similar results were
reported for the second stage.
Discussion
Main Findings
The evidence for a relationship between UA and adverse infant outcomes
is inconsistent. The reported effect sizes vary from small and not
statistically significant to highly clinically significant.
Strengths and Limitations
Individual Studies
Only one study reported direct measures of neurological function in
neonates and no study reported long-term neurological outcomes. These
interactions are inherently difficult to assess because adverse outcomes
of these types are rare and often require long-term follow-up.
As pre-specified in our protocol, NICU admission was excluded as an
outcome. Admission to the NICU was reported either as an individual or
as part of a composite outcome in three of the included studies.
(22, 26,
31) The group of babies admitted to NICU
is heterogenous and likely significantly composed of babies conditions
not associated with neurological function.
One study explicitly excluded neonates with encephalopathy, potentially
introducing significant selection bias. Selection criteria for studies
of intrapartum monitoring should allow for the inclusion of labours with
negative outcomes including HIE and fetal death.
None of the included articles reported sample size calculations. Since
the relevant adverse outcomes are rare and assuming modest estimated
effect sizes, several of the included studies were underpowered for the
variables of interest to this review.
When trying to establish a causative relationship, controlling for
confounding variables through multivariate analysis should be based on a
model of the mechanisms of the purported relationships between the
exposure, the outcome, and the potential confounder.
(34) Oxytocin and nulliparity are both
associated with increased UA. (14,
35) Therefore, in a study of UA and
neonatal outcomes, it is not usually appropriate to control for these
factors since, according to the hypothesis, they may be causally related
to the outcome via their effect on uterine activity. Some included
studies e.g. Heuser et al. reported multivariate analyses which included
these variables.
Chorioamnionitis, oligohydramnios, pre-eclampsia and the duration of
labour have all been associated with both increased UA and adverse
neonatal outcomes. (22,
26, 36,
37) For the most part, the connection
between these factors and neonatal outcomes is not plausibly via their
effect on tachysystole. Therefore, it may be appropriate to control for
these factors when analysing the influence of UA on neonatal outcomes.
They were not considered in the included studies.
Review Level
The permissive eligibility criteria for this review allowed a
comprehensive survey of the literature. This approach precluded
metanalysis. However, given the disparate nature of the evidence base,
valid metanalysis would not have been possible even with a more focused
approach.
The assessment of bias in the individual studies was limited. Current
tools for assessing bias in non-interventional studies are not robust.
(38)
Publication bias is common in systematic reviews and may particularly
affect observational studies. (39) Since
publication bias is difficult to assess accurately and owing to the
heterogeneity of the included studies, no formal assessment was
attempted. Study registration and replication have been proposed as
solutions to publication bias and selective reporting in observational
studies. (40,
41) None of the included trials had
protocols registered at ClinicalTrials.gov. No replication studies were
found.