Introduction
COVID-19 containment and control policies were implemented worldwide in response to the SARS-Cov-2 pandemic. The preventative and protective measures placed on communities in order to reduce viral transmission created an opportunistic experiment [
1] which may add to our future understanding of what causes preterm birth (PTB) [
2]. A Danish nationwide study first described a dramatic reduction in extremely PTBs (xPTB) during the strict lockdown period [
3]. Following this finding, similar reductions in PTBs were reported from the Netherlands [
1], Japan [
4], Italy [
5], Tennessee [
6], New York [
7], Israel [
8] and the USA [
9], and a report from Ireland described a reduced proportion of very low birth weight babies [
10]. However, smaller studies from California [
11], Philadelphia [
12], Israel [
13], Spain [
14] and London [
15], as well as a recent nationwide Swedish study [
16], could not confirm these findings.
A recent meta-analysis indicated that during the COVID-19 pandemic high-income countries generally saw a reduction in PTBs and stillbirths, while low- to middle-income countries (LMIC) saw increases [
17]. Notably, there are very few reports from LMIC. The conflicting findings can in part be explained by health system inefficiencies and/or an inability to deal adequately with the pandemic [
17]. In Nepal, for instance, the number of women giving birth in institutions dropped precipitously [
18]. These regional differences are further supported by a publication reporting data from 17 countries [
19], where the xPTB rate was reduced between 11 and 22% in the European region and increased by 48% in China and India. Furthermore, some studies have reported elevated stillbirth rates [
5,
13,
15] and a study from Nepal reported an increase in both PTBs and stillbirths during their lockdown periods [
18].
Stillbirths and early neonatal mortality rates were not assessed in the three nationwide studies from The Netherlands, the USA and Denmark [
1,
3,
9]. Since PTB rates could be reduced in response to increasing perinatal mortality rates [
20], there is a need to assess live birth rates alongside perinatal mortality rates (stillbirth and early neonatal death) during the different periods studied. Accordingly, this study aimed to use data pertaining to all Danish pregnancies and births, captured in the extensive nationwide electronic registers to study the prevalence of PTB, stillbirth and early neonatal mortality from singleton pregnancies during the first strict lockdown period in Denmark and during the continuous period of COVID-19 restrictions. Furthermore, we describe the stringency of Danish COVID-19 policies and the timeline of their implementation in order to contextualize our findings.
Discussion
Prematurity is a complex and challenging pathophysiological condition [
27] which is the leading cause of death in children under 5 years old [
28,
29]. The complex and multifaceted aetiology of PTB is linked to a wide range of psychosocial, medical and environmental factors, but remains only partly understood [
30]. Emerging data suggest that the COVID-19 pandemic and subsequent mitigation measures have provided a natural experiment, through which we can compare PTB and perinatal mortality rates against the human psychosocial and environmental characteristics of pre-COVID-19 and COVID-19 restriction periods.
Using population-based Danish register data, we have documented that the strict lockdown period is associated with a ~ 70% reduction in singleton xPTBs, while no difference was noted for the other PTB groups (Table
2). This result is compatible with the previously reported ~ 90% reduction in xPTBs assessed from the Danish Neonatal Screening Biobank (DNSB) [
3], taking into account that three out of four extremely preterm infants died during the first day of life (very early neonatal death)—precluding their appearance in the DNSB (Table
1). The proportion of very PTBs was slightly, ~ 20%, reduced, but there is no evidence for a protective effect of the lockdown in this group. It is important to note that the reduction in xPTBs during the lockdown period does not appear to be mirrored by an increase in stillbirths (Tables
1 and
2). As several studies had reported an increase in stillbirths [
5,
13,
15] during the pandemic, a decrease in xPTBs could have been explained by an increase in stillbirth [
20]. This does not seem to have been the case in Denmark.
A reduction in PTB or xPTBs has been observed in three nationwide studies from high-income countries (The Netherlands, the USA and Denmark [
1,
3,
9]). The stringency of lockdown in all three countries was greater than 70% (Figs.
1,
3, and
5 in the Supplement), with clear effects on location behaviour, e.g., workplace attendance reduced around 50% (Figs.
1-
6 in the Supplement). A nationwide study conducted in Sweden found no effect on the rate of PTB [
16]; however, the Swedish COVID-19 response was characterized by comparatively lenient lockdown measures, with a stringency of approximately 65% (Fig.
7 in the Supplement). Mobility data from Sweden reflects these measures with an average reduction of workplace attendance of approximately 30% (Figs.
7,
8 in the Supplement). These data suggest that the reduction in xPTBs correlates with the nature and extent of the lockdown, and, in consequence, the behavioural and psychosocial changes associated with an effective lockdown.
The data presented in the current study covers a broader period of the Danish COVID-19 restrictions and demonstrates the temporal variation in xPTB rates after the easing of the lockdown (April 15–September 30) as illustrated by a marked increase in xPTB rates in May and June (Fig.
2C) followed by reduced levels in July, August and September. The net result is that for the whole period with COVID-19 restrictions, the odds ratio of xPTB was reduced, albeit not significantly (Table
2). The months May and June coincide with a reopening of society and a relative return to normal activities (Figs.
1C,
2 in the supplement). For example, the lockdown is characterized by a stable ~ 50% reduction in workplace attendance, during May and June workplace attendance climbs steadily to ~ 20% reduction where, with the exception of the vacation period in July, it remains relatively stable. Thus, the periods characterized by high stringency and considerable mobility reduction (lockdown) correlate with a significant reduction in xPTB, whereas the lockdown easing phase (May and June) coincides with a significant increase in xPTB rate.
The stillbirth rate for the whole COVID-19 restriction period (February 27–September 30) was reduced by ~ 30% (Table
1). The monthly number of total stillbirths (Fig.
2B) and stillbirths among extremely preterm pregnancies (Fig.
2D) do indicate a uniform reduction in the number of stillbirths in each month during the COVID-19 restriction period. Consequently, the stillbirth rate does not account for the reduction in xPTBs during the Danish lockdown and is itself reduced as a possibly unintended beneficial consequence of the COVID-19 restrictions generally.
The reduction in combined extremely preterm stillbirth and live birth rates is statistically significant in February and March, while the increase in May and June is not (Fig.
3). This seems to reflect that the improvement in perinatal health started prior to the lockdown, during which period the Danish government communicated frequently with the public regarding the risk of COVID-19 in Denmark and societal preparedness (Fig.
1) [
31].
Perinatal mortality was moderately reduced during the lockdown period and was, as a result of a considerably reduced stillbirth rate, significantly reduced over the extended COVID-19 restriction period (Table
1). Early, and very early, neonatal mortality was not changed during these periods (Table
1). Consequently, despite the strain COVID-19 is expected to place on health care resources [
32], the observed reduction of xPTB is unlikely to be driven by a change in obstetric policies, formal or otherwise.
One may speculate on the causes of the reduction in xPTBs and stillbirths—one suggestion is a reduction in exposure to harmful pathogens [
33] or changes in genital tract microbiota due to hygiene precautions; massive reductions in the transmission of meningitis [
34] and pertussis [
35] and an absent influenza season 2020–2021 [
36] are consistent with this argument. Furthermore, the number of laboratory-confirmed chlamydia cases dropped ~ 25% during the lockdown compared to the 2015–2019 average, followed by an increase in June and July 2020, and a return to the 2015–2019 level in August [
37]. However, the number of people tested for chlamydia during this time was similarly reduced. As a surrogate marker of ascending urogenital infections and microbiota changes, this moderate reduction during the lockdown suggests that a reduction in ascending infections is unlikely to be responsible for the reduction of xPTBs. Exposure to environmental pollutants, particularly air pollution, has been associated with PTB [
38]. However, the Danish lockdown was not associated with major changes in the level of air pollution [
39].
Reduced anxiety and stress may also play a role by influencing the expression of the chaperone protein FKBP51, which increases in response to stress and alters the oestrogen/progesterone ratio thus overcoming the progesterone effect which inhibits parturition [
40,
41]. Two, albeit nearly 30 years old, Danish studies showed an association between anxiety in gestational week 30, but not in week 16, and PTB [
42], and an association between very early PTB (< 34 weeks) and severe stressful events in mid-pregnancy [
43], respectively. Furthermore, leisure time activity during pregnancy has been associated with a reduced risk of PTB [
44], while hard work and hard physical activity during pregnancy could increase the risk of PTB [
45,
46]. An Irish questionnaire-based COVID-19 study among 71 pregnant women [
47] revealed an increased worry for elderly relatives and children rather than for their pregnancy. Attitudes to the COVID-19 lockdown are being systematically studied as part of the HOPE project [
48] in Denmark, and the feeling of being efficacious, i.e., being able to act on information given, seems to be a very significant predictor of protective behaviour [
49] in Western democracies. Surprisingly, the lockdown-associated suspension of daily duties has been shown to be of value for people with vulnerable psyches [
50]. Thus, in Denmark, the reduced workplace and social stress may outweigh the negative effects of worries associated with the pandemic. While the end of the lockdown period was associated with improved psychological well-being [
51], the practicalities of easing lockdown restrictions and returning to work, returning children to school, etc., could increase stress and anxiety. This is particularly true among women with depression (prevalence of 14.2% among Danish women) or anxiety (prevalence of 6.8% among Danish women) [
52], both of which have been associated with PTB [
53]. We cannot exclude that the psychosocial and behavioural changes during the lockdown, i.e. more people working from home, could have affected the risk of xPTB in specific groups of pregnant women.
The observational nature of this study precludes causal inference, but the sizeable effect on the xPTB rate indicates that it will be worthwhile to identify the elements of the lockdown that have conferred this unintended protective effect regarding xPTB. One might speculate that it will be possible to identify a specific psychosocial phenotype in mid-pregnancy that may benefit from specialized care. However, the complex causality of PTB [
30] suggests that defining controllable protective factors will require a cross-disciplinary effort.
The use of national registers allowed us to examine all Danish pregnancies, thus avoiding selection bias and allowing us to provide a complete assessment of birth and mortality rates.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.