Changes in Smoking Habits in Pregnant Women During COVID-19 Pandemic. A Comparison of Two Cohorts

Background: Although smoking in pregnancy has well known deleterious effects, about 50% women who smoked out of the pregnancy continue to smoke while pregnant. The main is to study changes in smoking in pregnancy during covid-19 pandemic. Methods: We compare two cohorts of pregnant women: one recruited in 2018 (pre-pandemic cohort) and one recruited from March 2020 on (pandemic cohort). Results: The percentage of women who smoked decreased from 13% in the pre-pandemic cohort to 8% in the pandemic cohort (p<0.001). This step-down was similar in all age groups and both employed and unemployed/no active women. Among women in the pandemic cohort, those whose pregnancy began after March-2020 (when the pandemic was declared) reported 1% smoking prevalence, which is much lower than smoking prevalence in women whose pregnancy began before the pandemic was declared. Conclusion: Smoking was less frequent in pregnancies during the covid-19 pandemic and even less frequent in women whose pregnancy began after the pandemic was declared. The effect that lockdown, stay-at-home orders or healthier behaviour habits could have had on smoking in pregnancy would require further research.


Background
Smoking tobacco in pregnancy is well known as deleterious for both mother and baby health [1,2].
Although some studies showed a tendency towards decreasing prevalence [3], more than 50% women who smoked daily out of the pregnancy continued to smoke while pregnant [4]. The pandemic of covid-19 has led to changes in health-related behaviour, including mask use, stay-at-home orders and others, which have induced some bene cial effects regarding diseases other than covid-19; for instance, the 2021 outbreak of in uenza has been much lower than usual or even inexistent [5,6]. Although some studies have reported changes in habits during the pandemic [6], only a few small qualitative studies have analysed behavioural changes in pregnant women during the pandemic [7][8][9]. Therefore, little is known about whether pregnant women have increased, decreased or maintained their habits in this period.
In this article, we study changes in smoking habit during pregnancy comparing two cohorts: one recruited in 2018 -before the pandemic began-and another recruited from March 2021 -when the pandemic was ongoing-in a single hospital located in Northern Spain.

Setting and patients
This is a joint analysis of two cohorts recruited at the University Hospital Marques de Valdecilla (HUMV), Cantabria, Spain. Pre-pandemic cohort was recruited in 2018 and included 948 pregnant women and their children. Their main characteristics have been described elsewhere [10,11]. In brief, women were consecutively recruited between 1st January and 31st August 2018. In order to participate in the cohort, they should have been admitted for delivery in the HUMV and reside in the region were the HUMV is (Cantabria, Northern Spain).
The pandemic cohort was recruited in 2020. Its pro le has been already reported [12]. Recruitment began on 26th May and nished on 22nd October. This cohort was composed by three sub-cohorts. Sub-cohort 1 was retrospectively recruited and includes women delivering at HUMV between 23rd March (the rst date the hospital introduced routinely SARS-CoV-2 infection test via PCR for all women admitted for delivery) and 25th May, 2020. Sub-cohort 2 was prospectively recruited and includes women delivering at HUMV from 26th May on. Sub-cohort 3 was prospectively recruited and includes women attended at HUMV for their routine 12-week of pregnancy consultation; many of their pregnancies are still ongoing when carrying on this analysis. The rationale for these three sub-cohorts is to differentiate the pandemic consequences in pregnancy according to the period of risk for each woman, which was as follows: women in the rst sub-cohort were neither exposed to nor aware of SARS-CoV-2 for most of their pregnancy, and their exposition was limited to the last trimester of pregnancy, which mostly coincided with the rst pandemic wave. Women in the second sub-cohort could have been both exposed to and aware of the pandemic from their second trimester of pregnancy, which coincided with the rst pandemic wave, while their third trimester was concurrent with the lower levels of covid-19 incidence between the rst and the second waves. Finally, women in the third sub-cohort were aware of the pandemic all throughout their pregnancy and their exposition was higher in the second and posterior waves.

Gathered information
Information on maternal educational level (classi ed as primary school, secondary school, vocational training and university), occupational situation (classi ed as employed, unemployed or inactive, student) was self-reported. Age, number of previous children, length of pregnancy (later classi ed as less than 34 weeks, 34-36 weeks + 6 days, 37 weeds or more), type of delivery (eutocic, instrumental or Caesarean section), weight at birth (later classi ed as less than 2500 grams, 2500-4000 grams and more than 4000 grams) and type of feeding at hospital discharge (exclusive breastfeeding, mixed -breastfeeding and arti cial formula-, and formula) were obtained from medical records. Smoking habit in pregnancy and averaged daily number of cigarettes in pregnancy were obtained as reported by the pregnant women.

Statistical analysis
In this article, we compare the pre-pandemic and the pandemic cohorts with each other, regarding the smoking habit in pregnancy and its association with pregnancy, delivery and neonate characteristics. We also carry on the same comparison between all three sub-cohorts in the pandemic cohort.
Association between smoking and cohort according to woman, pregnancy or child characteristics was tested using Chi-square test and association between the average daily number of cigarettes and cohort according to the same characteristics was tested using the Student's t test. Results are provided as number with percentage or mean with standard deviation. Odds ratios and their 95% con dence intervals were obtained to compare smoking in pregnancy in the pandemic and the pre-pandemic cohorts, using the later as reference. Therefore, odds ratios lower than 1 indicate that smoking was less frequent in the pandemic cohort, while odds ratios over 1 indicate that smoking was more frequent in the pandemic cohort. All p values are two-tailed.

Results
In this study, 2116 pregnant women were included; 948 belonging to the pre-pandemic cohort and 1168 to the pandemic one. Among the pandemic cohort, 270 women were retrospectively recruited and had delivered a bay before 26th May, 2020 (sub-cohort 1); 350 were prospectively recruited at delivery from 26th May, 2020 on (sub-cohort 2), and 548 were prospectively recruited at week 12 of pregnancy (subcohort 3). Only 53 women in sub-cohort 3 had already delivered a baby by the time of this analysis, so this sub-cohort was excluded regarding variables related to delivery. Supplementary Table 1 provides the description of the participants. Smoking in pregnancy was less frequent in the pandemic cohort (8%) than in the pre-pandemic one (13%) (p < 0.001). Both cohorts also differed in educational attainment, occupational situation, type of delivery, length of pregnancy and weight at birth.
The association between smoking in pregnancy and cohort, according to maternal characteristics is depicted in Table     To further analyse changes in smoking during the pandemic, Table 3 displays the relationship between smoking in pregnancy and sub-cohort, according to maternal characteristics. Smoking was less frequent in pregnant women belonging to the sub-cohort 3 (1%) than in sub-cohorts 1(12%) and 2 (15%), irrespective of their age, educational level or occupation situation. Women in sub-cohort 1 and sub-cohort 2 declared similar rates of smoking. In Table 4, smoking in sub-cohorts 1 and 2 in the pandemic is compared according to type of delivery and neo-born characteristics. Smoking was more frequent in sub-cohort 2 if delivery was eutocic (17% vs. 10%, p = 0.05). No other differences were found between both sub-cohorts.

Discussion
According to our results, smoking was less frequent in pregnant women during the covid-19 pandemic than in pregnant women in 2018, before the pandemic began. However, this reduction mainly occurred in the sub-cohort 3, which is formed by women who became pregnant when the pandemic had already started, rather than in sub-cohorts 1 and 2 whose pregnancies commenced several months before being aware of the pandemic. The cutdown in smoking during sub-cohort 3 seems to be independent of maternal age, educational level and occupational situation.
In the only study we found comparing pandemic and pre-pandemic cohorts, Overbeck et al (2021) [13] reported smoking prevalence in Danish pregnant women cutdown from 6.8% in 2015/16 to 4.0% in 2020 during the lockdown. Ceulemans et al (2021) [14], in an international study carried out in Ireland, Norway, Switzerland, The Netherlands and United Kingdom, informed of a smoking prevalence in pregnant women as low as 2.6% during the pandemic, although they did not inform on a pre-pandemic comparison group. A Spanish cohort reported that prevalence of smoking was 12.8% in non-SARS-CoV-2 infected women and 10.2% in infected women, although these gures include both smokers in pregnancy and former smokers [15].
Other changes in reproductive health behaviour have been described. For instance, Stout et al [16] have informed of a 14% reduction in pregnancy initiation after the beginning of the pandemic, contrasting with a trend to increase in the same areas throughout the three preceding years. A survey conducted by the Guttmacher Institute reported a trend to delay pregnancy, cancellations of contraceptive or other sexual and reproductive health care [17]. Our group have depicted that maternal leaving of work has been earlier in pregnancies initiated during the pandemic [18]. These changes, together with the decrease in smoking in pregnancy we are reporting, delineate a tendency to self-protective behaviour in pregnant women and women in reproductive age during the pandemic.
The mechanisms for these changes could be multiple. Firstly, the women themselves could have adopted healthier behaviours, which would have been induced by the sense of individual risk as transmitted in media or by governmental spokespersons. Secondly, societal shutdowns and stay-at-home order could have induced those changes, voluntarily or not; by-the-way, early new reported higher risk of covid-19 infection in smokers, which led to restriction or prohibition of smoking in public places (e.g., in terrace bars). Thirdly, health care services have modi ed their functioning, including -among others-delays of appointments, increase in telemedicine. We have no information to estimate the relative importance of these three factors on changes in reproductive behaviour.
This study has some limitations. Firstly, data on smoking were self-reported, which makes some room to misclassi cation if some women -being aware of the social rejection on tobacco, especially in pregnancyhave hidden their use. However, we think that this misclassi cation -if exists-would have been nondifferential between the cohorts we are comparing; so that the decreasing trend would still hold. Secondly, our study has been carried out in a single centre; therefore, further studies are required in other locations to test its generalizability. Among the strengths of this study, we would remark the prospective recruitment of most women and the homogeneous way of recording information.

Conclusions
In conclusion, we are reporting that smoking prevalence in pregnancy has decreased during the COVID-19 pandemic. This reduction coincides with other changes in reproductive behaviour. Further research is needed to acknowledge whether this cutdown would remain after the pandemic be controlled.