Key words COVID-19 - SARS-CoV-2 - preterm birth - lockdown measures - preterm delivery
Schlüsselwörter COVID-19 - SARS-CoV-2 - Frühgeburt - Lockdown-Maßnahmen
Introduction
In the context of the SARS-CoV-2 pandemic, countries around the world have implemented
different measures to reduce the incidence of SARS-CoV-2 infection. Initially, obstetricians
were
concerned about the possible consequences of SARS-CoV-2 infection during pregnancy
[1 ], [2 ]. Furthermore, it was assumed that the
quality of antenatal care would decline as a result of mandatory contact reductions.
These worries were initially supported by studies showing an increase in preterm births
and stillbirths
[3 ], [4 ], [5 ].
In contrast, studies from Ireland and Denmark demonstrated a decrease in preterm birth
rates as well as in the numbers of very low birth weight (VLBW) and extremely low
birth weight (ELBW)
neonates during defined lockdown periods [6 ], [7 ]. These studies, both with small sample numbers, were complemented by a study
from the Netherlands [8 ] based on data from approximately 1.6 million newborns. This study also demonstrated
a reduction in preterm births during the lockdown
period. Since the publication of these data, it has been discussed whether this
is a coincidence or if a causal relationship could exist. A meta-analysis demonstrated
that in high-income
countries a moderate reduction in preterm birth rates was achieved by infection
control measures.
However, the studies published so far have some limitations. It has not yet been investigated
to what extent there might be a relationship between a reduction in population mobility
and the
risk of preterm birth. Population mobility might offer a better representation
of the coherence of the population with regard to nonpharmacologic interventions rather
than assuming homogeneous
effects during complete lockdown periods [9 ]. Hence, the main hypothesis of our study was that the risk of preterm birth diminishes
with decreasing population
mobility. Furthermore, we hypothesized that some of the observed effects in 2020
might be confounded by an overall long-term trend towards a reduction in preterm births.
Additionally, there is
a lack of studies that have investigated these effects in twin pregnancies, which
typically are associated with a higher overall risk of preterm birth.
Therefore, the aim of the present study was to investigate whether an association
between lockdown measures or a lockdown-induced reduction in population mobility and
preterm birth rates can
be demonstrated based in a large full cohort (all births from 2010 to 2020) from
Bavaria while also adjusting for long-term trends.
Material and Methods
Setting
Due to increasing numbers of infections, a number of measures were prescribed by law
in 2020 to reduce the incidence of SARS-CoV-2. Starting from March 16th, the Bavarian
government urged
the population to reduce their social contacts to a minimum and to practice hygiene
measures. After a decrease in infection rates, gradual opening up with a lessening
of restrictions on
social activities was permitted from April 20th while the general hygiene measures
remained in force. The first lockdown ended on May 6th, 2020 with the end of the curfew
for the Bavarian
population.
After a new rise in the number of infections, contact restrictions were introduced
again from November 2nd. These measures were tightened as infection rates remained
high and were
maintained until the end of 2020 ([Fig. 1 ], [10 ]). The most relevant Bavarian measures are shown in [Fig. 2 ].
Fig. 1 Daily reported counts of SARS-CoV-2 infections in Bavaria from January 2020 to January
2021 (data from: [10 ]).
Fig. 2 Infection control measures from March – May and November – December 2020 in Bavaria.
Red: tightened measures; green: lightened measures.
Study population and examined lockdown periods
In Germany, it is mandatory to report specific fixed parameters for all births to
the Institute for Quality Assurance and Transparency in Healthcare (IQTIG). In Bavaria,
the Bavarian
Institute for Quality Assurance (BAQ) accepts these obstetric quality data and
forwards them to IQTIG. For this study, Bavariaʼs centrally collected obstetric data
were made available by the
BAQ in anonymized form for a secondary analysis of singleton and twin births.
Study cohorts
Deliveries during the lockdown periods in 2020 constituted the study group, while
births from 2010 to 2019 represent the historical cohort. The study specifically focused
on births during
the defined Bavarian lockdown periods (Lockdown 1 from March 16th, 2020 to May
6th, 2020; Lockdown 2 from November 2nd, 2020 to December 31st, 2020). According to
gestational age at delivery
(weeks + days), the analyzed population was divided into term deliveries and
preterm births as follows:
Preterm births were additionally divided into subgroups as follows:
In addition, the frequency of the two different subgroups according to birth weight
was included in the analysis:
very low birthweight (VLBW, < 1500 g)
extremely low birthweight (ELBW, < 1000 g)
Singleton and twin pregnancies were analyzed separately. Higher-order multiple pregnancies
were excluded from the analysis.
Mobility patterns in Bavaria
In addition to the two lockdown periods, we investigated the general impact of population
mobility in 2020 on preterm births compared to 2019.
For this purpose, anonymized phone data of two large mobile phone providers (Deutsche
Telekom, Telefónica) were analyzed. The exact procedure is described in the COVID-19
Mobility Project
[11 ]. Briefly summarized, mobility is defined as a change in the mobile cell to which
the mobile phone is logged on. This technique allows structural
changes in population mobility to be identified.
Statistical analysis
Descriptive statistics of preterm births are presented for both singleton and twin
births as numbers and percentages. To explore the effects of lockdown measures on
preterm birth rates in a
first univariate analysis, Fisherʼs exact test was used to compare raw prevalence
during lockdown periods compared to equivalent periods from 2010 to 2019.
In the multivariable analysis, the risk of preterm birth was modeled for the overall
sample from 2010 to 2020 using generalized additive regression models [12 ], adjusting for seasonal and long-term trends. More precisely, penalized splines
were incorporated in logistic regression models to account for non-linear long-term
trends
(year), seasonal trends (cyclic spline for calendar week) and weekly patterns
(cyclic spline for weekday) while estimating the potential effects of the lockdown
periods (binary explanatory
variables) or mobility change (continuous explanatory variable) in the year 2020.
To explore these effects further, similar additive regression models were also fitted
for gestational age
and birthweight as continuous outcomes. All statistical analyses were performed
using the statistical computing environment R (Version 4.1.2) with corresponding add-on
packages for the
additive regression models [13 ].
Ethical approval
Since this was a secondary analysis of centrally collected anonymized data, no ethics
vote was required for this study.
Results
Overall collective and examined study periods
In total, the overall study period included 1 263 959 births in Bavaria, Germany from
2010 to 2020. Of these, 1 217 442 (96.32%) were singleton births, 45 211 (3.58%) were
twin births and
1302 (0.10%) were higher-order pregnancies. In 2020, a total of 125 089 births
were reported, of which 120 669 (96.47%) were singleton pregnancies, 4296 (3.43%)
were twin pregnancies and the
remaining 124 (0.10%) pregnancies were higher-order pregnancies.
Preterm birth rates between 2010 and 2020
Annual birth counts and their corresponding preterm birth rates between 2010 and 2020
are shown in [Table 1 ]. Over the years, the tendency to preterm births
has decreased in Bavaria.
Table 1 Bavarian (preterm) birth rates from 2010 to 2020.
2020
2019
2018
2017
2016
2015
2014
2013
2012
2011
2010
n: case numbers, wks: weeks of gestation
Singleton births, n
120 669
120 057
118 969
119 425
115 664
112 240
107 653
103 567
101 890
98 292
99 036
Preterm births, < 37 + 0 wks
6 787
7 203
7 101
7 316
7 246
7 185
7 002
6 796
6 718
6 572
6 609
Preterm birth rate, %
5.62
6.00
5.97
6.13
6.27
6.40
6.50
6.56
6.59
6.69
6.67
Twin births, n
4 296
4 426
4 318
4 451
4 371
4 260
4 252
3 966
3 638
3 571
3 662
Preterm births, < 37 + 0 wks
2 361
2 521
2 396
2 513
2 507
2 523
2 376
2 274
2 176
2 169
2 245
Preterm birth rate, %
54.96
56.96
55.49
56.46
57.36
59.23
55.88
57.34
59.81
60.74
61.31
Lockdown periods and preterm birth
In total, there were 361 737 preterm births during the studied periods (35 333 in
the actual lockdown periods and 326 404 in the corresponding periods between 2010
and 2019), and they were
included in our univariate analysis. Numbers of deliveries for each studied period
are shown in [Table 2 ].
Table 2 Preterm birth rates during both lockdown periods in 2020 compared to 2010 – 2019.
2010 – 2019
2020
Raw OR
p value
Adj. OR
Adj. p value
n (%)
n (%)
n: case numbers, %: percentage, wks: weeks of gestation, VLBW: very low birth weight
(< 1500 g), ELBW: extremely low birth weight (< 1000 g), ‡ p < 0.05.
Raw odds ratio (OR) refers to deliveries during the lockdown periods in 2020 compared
to the corresponding periods in 2010 – 2019, the adjusted OR takes all births from
2010 to
2020 into account and was adjusted for non-linear long-term trends, seasonality
and weekday effects.
Lockdown period I
March 16th to May 5th
Total number of singleton births during Lockdown
145 018
16 015
Number of singleton preterm births, n (%)
9 295 (6.41)
914 (5.71)
0.88
< 0.001‡
0.99
0.73
< 32 + 0 wks
1 312 (0.9)
115 (0.72)
0.79
0.02‡
0.82
0.04‡
< 28 + 0 wks
477 (0.33)
50 (0.31)
0.94
0.77
0.96
0.81
VLWB
1 190 (0.82)
110 (0.69)
0.84
0.08
0.84
0.09
ELBW
548 (0.38)
47 (0.29)
0.78
0.10
0.79
0.13
Total number of twin births during Lockdown, n
5 649
551
Twin preterm births, n (%)
3 195 (56.26)
292 (52.99)
0.88
0.15
0.91
0.31
< 32 + 0 wks
593 (10.44)
51 (9.26)
0.87
0.42
0.95
0.77
< 28 + 0 wks
186 (3.28)
18 (3.27)
1.00
1.00
1.24
0.42
VLWB
545 (9.60)
51 (9.26)
0.96
0.88
1.06
0.71
ELBW
207 (3.65)
17 (3.09)
0.84
0.63
1.09
0.74
Lockdown period II
November 2nd to December 31st
Total number of singleton births during Lockdown
168 989
18 159
Number of singleton preterm births, n (%)
11 156 (6.60)
1 037 (5.71)
0.86
< 0.001‡
0.96
0.24
< 32 + 0 wks
1 550 (0.92)
172 (4.83)
1.03
0.68
1.10
0.24
< 28 + 0 wks
587 (0.35)
70 (0.39)
1.11
0.39
1.20
0.17
VLWB
1 455 (0.86)
173 (0.95)
1.11
0.21
1.18
0.05
ELBW
688 (0.41)
90 (0.50)
1.21
0.09
1.30
0.02‡
Total number of twin births during lockdown, n
6 748
608
Twin preterm births, n (%)
3 975 (58.91)
353 (58.06)
0.97
0.70
1.11
0.24
< 32 + 0 wks
605 (8.97)
79 (12.99)
1.51
0.002‡
1.70
< 0.001‡
< 28 + 0 wks
164 (2.43)
25 (4.11)
1.72
0.02‡
1.69
0.02‡
VLWB
630 (9.34)
73 (12.01)
1.32
0.04‡
1.54
0.003‡
ELBW
202 (2.99)
26 (4.28)
1.44
0.09
1.52
0.06
In the more advanced regression analysis, we incorporated all births from 2010 to
2020. The additive generalized regression models demonstrated a seasonal increase
in preterm births in the
first winter months during the studied period as well as a considerable long-term
trend towards a lower risk of preterm birth. In contrast, there is a lower incidence
of preterm births in
the summer months. Analysis of the individual weekdays showed only small impact
on the risk of preterm birth ([Fig. 3 ]).
Fig. 3 Preterm birth rates and possible impacts. a Trend of the Bavarian preterm birth rate between 2010 and 2020. b Impact of seasons on the risk of preterm birth.
c Impact of weekdays on the risk of preterm birth.
Univariate analysis revealed a significant decrease in singleton preterm births with
delivery occurring < 37 + 0 weeks of gestation (5.71% vs. 6.41%; OR 0.88; p < 0.001)
during the
first lockdown period ([Fig. 4 ]). Similarly, we demonstrated a significant reduction in preterm births occurring
< 32 + 0 weeks gestation (0.72% vs. 0.9%;
OR 0.79; p = 0.02), while preterm births < 28 + 0 weeks gestation did not change
significantly (0.31% vs. 0.22%; OR 0.04; p = 0.77). [Table 1 ] illustrates
the proportion of preterm births per 1000 live births in 2020 compared to 2010 – 2019.
After adjusting for seasonality and long-term trends, no significant effect was found
for preterm
births (adj. OR 0.99; p = 0.73), while preterm birth rates < 32 weeks gestation
continued to be significantly lower during the first lockdown period (adj. OR 0.82;
p = 0.04).
Fig. 4 Preterm birth rates in singleton pregnancies per 1000 births from 2010 to 2019 vs.
2020.
During the second lockdown period, there was also a significant reduction in all studied
preterm births (5.71% vs. 6.60%; OR 0.86; p < 0.001), although this effect did not
remain
significant after adjustment (adj. OR 0.96; p = 0.24).
There was no significant impact on the rates of neonates born < 32 weeks of gestation
(0.95% vs. 0.92%; OR 1.03; p = 0.68; adj. OR 1.10; p = 0.24) and < 28 weeks of gestation
(0.39%
vs. 0.35%; OR 1.11; p = 0.39; adj. OR 1.20; p = 0.17), respectively.
For twin pregnancies, no significant difference in preterm birth rates was observed
during the first lockdown period (52.99% vs. 56.26%; OR 0.88; p = 0.14; adj. OR 0.91;
p = 0.31). There
was also no difference in the subgroups showing delivery < 32 weeks of gestation
(9.26% vs. 10.44%; OR 0.87; p = 0.42; adj. OR 0.95; p = 0.77) and < 28 weeks gestation
(3.27% vs.
3.28%; OR 1.00; p = 1.00; adj. OR 1.24; p = 0.42).
During the second lockdown, there was no difference in the preterm birth rate for
all twin pregnancies (58.06% vs. 58.91%; OR 0.97; p = 0.70; adj. OR 1.11; p = 0.24).
However, there were
more children born < 32 weeks of gestation (12.99% vs. 8.97%; OR 1.51; p = 0.002;
adj. OR 1.70; p < 0.001) and < 28 weeks gestation (4.11% vs. 2.43%; OR 1.72; p = 0.02;
adj. OR
1.69; p = 0.02) in 2020 ([Table 2 ]).
Lockdown periods and birth weight
Analyses of VLBW and ELBW fetuses do not provide evidence for significant differences
during the first lockdown period in 2020 compared to 2010 – 2019, neither for singleton
(VLBW: 0.69%
vs. 0.82%; OR: 0.84; p = 0.08; adj. OR: 0.84; p = 0.09; ELBW: 0.29% vs. 0.38%;
OR: 0.78; p = 0.10; adj. OR 0.79; p = 0.13) nor for twin pregnancies (VLBW: 9.26%
vs. 9.60%; OR: 0.96;
p = 0.88; adj. OR: 1.06; p = 0.71; ELBW: 3.09% vs. 3.65; OR: 0.84; p = 0.63;
adj. OR: 1.09; p = 0.74).
Regarding the second lockdown, the rate of VLBW neonates was higher for twin pregnancies
(12.01% vs. 9.34%; OR 1.32; p = 0.04; adj. OR 1.54; p = 0.003) but not for singleton
pregnancies
(0.95% vs. 0.86%; OR 1.11; p = 0.21; adj. OR: 1.18; p = 0.05). With regard to
ELBW, rates were significantly higher in singleton pregnancies (0.50% vs. 0.41%; OR
1.21; p = 0.09; adj. OR
1.30; p = 0.02), but not in twin pregnancies (4.28% vs. 2.99%; OR: 1.44; p = 0.09;
adj. OR: 1.52; p = 0.06) ([Table 2 ]).
Mobility change and preterm birth
During the first lockdown period, a maximum mobility reduction of 63.9% was observed
in the Bavarian population on April 13th, 2020 due to contact restrictions. Compared
to 2019, the first
lockdown period showed a mean mobility reduction of 34.8%.
In the second lockdown, the mean mobility was only 10% lower than in the previous
year. [Fig. 5 ] shows the changes in mobility of the Bavarian population
based on the data provided by the Covid-19 Mobility Project [14 ].
Fig. 5 Mobility changes in 2020 compared to 2019 in Bavaria.
When all births are considered, the decreased population mobility in 2020 had a significant
impact on the odds of a preterm birth (adj. OR [25% reduction] = 0.95, p = 0.002)
while adjusting
for long-term trends and potential seasonality.
However, subgroup analysis only showed a significant effect for twin births (adj.
OR [25% reduction] = 0.89, p = 0.02) and the effect did not reach significance for
singleton pregnancies
(adj. OR [25% reduction] = 0.97, p = 0.14).
Further modeling suggests that decreased population mobility might also have a potential
impact on gestational length in a combined cohort of singleton and twin births (0.19
days per 25%
reduction, p = 0.002). However, in a subgroup analysis, this effect did not reach
significance in singleton (p = 0.05) and twin pregnancies (p = 0.10).
Additional analysis showed that decreased population mobility has a small positive
effect on birth weight. Thus, for a 25% decrease in population mobility, an increase
in birth weight of
5.3 g was estimated across all births (p = 0.04). When analyzing the effect of
population mobility on preterm VLBW and ELBW infants, the odds of neonates having
a birth weight < 1500 g
decreased with reduced population mobility (VLBW, adj. OR [25% reduction] = 0.91,
p = 0.03), but the effect failed to reach significance for neonates with a birth weight
< 1000 g (ELBW,
adj. OR [25% reduction] = 0.90, p = 0.14).
Discussion
In the context of the infection control measures during the SARS-CoV-2 pandemic, studies
showed a significant reduction in preterm births [8 ] and VLBW and ELBW
neonates [7 ]. The discussion on whether and how these measures could be responsible for the reduced
risk of preterm birth is still ongoing. While analyses from
China [15 ], Sweden [16 ] and Spain [17 ] failed to demonstrate an effect of country-specific
lockdown measures on preterm birth risk, other studies indicate an association
with these measures [18 ], [19 ], [20 ]. However, a meta-analysis of 15 studies revealed a moderate effect of lockdown measures
on preterm births in high-income countries, whereas this effect was not
detectable in three low-income countries [21 ].
In our initial univariate analysis, we also found significantly lower preterm birth
rates in singleton pregnancies in both lockdown periods. In contrast to previously
published studies, we
performed these analyses in a high number of cases. Thus, initially, we were able
to confirm the results of the studies from Ireland and the Netherlands.
After adjusting for long-term trends and seasonality, however, multivariable semiparametric
regression analysis failed to provide evidence for a significant effect of both lockdown
periods.
Hence, it should be questioned whether the preterm birth rate really is linked
to the establishment of infection control measures during these periods. So far, only
one study has investigated
potential long-term trends. That study summarized centrally reported data from
Norway, Sweden and Denmark [22 ]. The results of that study and our work are
congruent.
An analysis based on fixed temporal lockdown periods must be assessed critically.
At the beginning of the SARS-CoV-2 pandemic, it was demonstrated that the population
in Germany was already
aware of SARS-CoV-2 through media reports prior to the implementation of specific
legislative measures on infection prophylaxis (e.g., the use of face masks and social
distancing) and this may
have led to an early reduction in the time-dependent reproductive rate R [23 ]. Furthermore, it is difficult to assess how strictly these restrictions and
recommended behavioral measures were applied by the population over the course
of the pandemic.
To further objectify population coherence with respect to legislative contact restrictions,
we performed semiparametric regression modeling based on population mobility but were
not able to
demonstrate that decreased mobility had an impact on the risk of preterm birth
in singleton pregnancies. To our knowledge, no other study has yet examined the effect
of population mobility on
a potential reduction in preterm birth rates. The data show that the coherence
of the Bavarian population was significantly stronger during the first lockdown than
during the second one.
Assuming a causal relationship, this could theoretically explain the significant
increase in ELBW neonates during the second lockdown.
Multiple pregnancies are generally at higher risk of preterm birth and this cohort
of pregnancies is of considerable interest with respect to the prevention of preterm
birth.
However, due to the increased a-priori risk of preterm birth, multiple pregnancies
have been excluded from comparable studies, with the exception of a study by Klumper
et al. [24 ]. Hence, our study is one of the first to also investigate the risk of preterm birth
for twin pregnancies in the context of lockdown measures. We did not show a
significant difference in twin preterm births during both lockdown periods while
the risk of preterm birth at < 32 and < 28 weeks of gestation was significantly higher
in the second
lockdown period. This rather unexpected result is in line with a study from the
Netherlands which also showed an increase in the risk of preterm birth < 28 weeks
of gestation [24 ]. The interpretation of these findings is challenging and we can only guess whether
the deterioration in population adherence and the increased incidence of
SARS-CoV-2 during the second lockdown had significant impacts on the increased
risk of preterm birth before 32 and 28 weeks of gestation in twin pregnancies.
Since we were able to demonstrate that reduced population mobility could have a favorable
effect on the preterm birth rate in twin pregnancies, one could conclude that mothers
of twins should
stay at home to increase the length of their pregnancy. In this context, however,
it is important to note that other risk factors for preterm birth were also influenced
by nonpharmacological
interventions during the SARS-CoV-2 pandemic; for instance, contact restrictions,
social distancing, increased awareness of hygiene measures, and the use of mouth-nose
protection are also
thought to be responsible for significantly reduced rates of respiratory infections
such as influenza [25 ], [26 ], [27 ]. The Robert Koch Institute reported a significant decrease in respiratory infections
in the German population following the introduction of contact
restrictions and hygiene recommendations during the studied periods [28 ]. Furthermore, lower levels of air pollution could also be related to the reduced
rates
of preterm births [29 ]. Significantly lower levels of pollutants were measured in Germany during the analyzed
period. Compared to the corresponding period in
the previous year, a reduction of NO2 by 37% was detected in Munich between March 15th, 2020, and April 30th, 2020, for
example [30 ].
In addition, a possible negative effect of the measures during the SARS-CoV-2 pandemic
has to be discussed. For instance, pregnant women reported anxiety and depressive
symptoms more often
[31 ], which have been identified as possible risk factors for preterm birth [32 ].
A limitation of the study is that the considered second lockdown period ends at the
end of 2020, while the second wave of infections and corresponding mitigation measures
continued in
2021.
Another limitation of our analysis is that the considered adjustment for the effects
of seasonal, weekday and long-term trends could theoretically also somewhat reduce
the true effects of
lockdown or mobility changes. As the pandemic influenced the behavior of the population
for nearly the whole of 2020, parts of the spline effect at the border of the study
period (2020, [Fig. 3 ]) might not just be part of an overall long-term trend but could also reflect the
impact of the lockdown periods or reduced mobility. It is difficult to
disentangle such partly overlapping effects.
In summary, it is of course very hard to design a study that can prove a causal relationship
between lockdown periods, reduced population mobility and a lower preterm birth rate.
Our
observational study can therefore only provide additional evidence for a multifactorial
phenomenon leading to a lower preterm birth rate during these periods.
Nevertheless, the strength of our study is that we investigated a possible impact
of population mobility on the risk of preterm birth in singleton and twin pregnancies.
Another strength is
the large number of cases: in total, our study includes 1.2 million births, which
allowed us to investigate various outcome parameters for preterm birth in singleton
and twin pregnancies in
detail.
Conclusion
Reduced preterm birth rates during the two lockdown periods in 2020 were also observed
for singleton and twin pregnancies in Bavaria, Germany. These effects were no longer
detectable after
adjusting for seasonal and long-term trends, indicating that they might not be
large enough to have a real clinical impact. Reduced population mobility was associated
with lower preterm birth
rates for twin pregnancies even after adjusting for seasonal and long-term trends.
Further studies are needed and should be conducted to estimate these effects, including
in smaller subgroups
such as twin pregnancies.
