Trend of antidepressants before, during, and after pregnancy across two decades—A population‐based study

Abstract Introduction Factors that influence antidepressant (AD) prescription and use during pregnancy are multiple including, in particular, the balance between the potential risk of untreated depression and the potential risk of AD treatment. Surveillance of temporal trends of AD use might identify areas requiring further research. We studied the use of ADs before, during, and after pregnancy using national data across two decades in Denmark. Methods We included 1,232,233 pregnancies leading to live birth in Denmark between 1 January 1997 and 31 December 2016. Information on redemption of AD prescriptions was obtained from the Danish National Prescription Register. Results We identified 29,504 (2.4%) pregnancies having at least one AD prescription (96,232 AD prescriptions) during pregnancy. The majority redeemed more than one prescription (69.7%) often for a single kind of AD (83.5%), and in 94% of the AD‐exposed pregnancies, the estimated duration of treatment was 1 month or longer. Prescription of ADs during pregnancy increased steadily from 0.4% in 1997 to 4.6% in 2011, but decreased thereafter to 3.1% in 2016. The proportion of pregnancies with ADs in 2011 was 6.05‐fold higher than that in 1997. The temporal trends in AD prescription in the years before and after pregnancy were similar to the trend during pregnancy. The decreasing use of ADs during pregnancy after 2011 was mainly driven by a decrease in the use of selective serotonin reuptake inhibitors (SSRIs), especially citalopram, the main type of SSRIs used in Denmark. Conclusion Prescription of ADs during pregnancy in Denmark increased steadily from 1997 to 2011 but decreased sharply thereafter. More research is needed to show whether the same trend exists in other populations, like women of reproductive age, men of reproductive age, and old people, and other countries. We also need to find explanation for the decreasing trend in recent years and potential risk for untreated depression.


S IG NIFIC ANT OUTCOME S
The proportion of pregnant women who redeemed antidepressants (ADs) during pregnancy increased six-fold from 1997 to 2011, but decreased sharply thereafter-similar trends were found before and after pregnancy.
The decrease in the use of ADs after 2011 was mainly driven by a decrease in the prescribing of serotonin reuptake inhibitors (SSRIs) and in particular citalopram, the main type of SSRI used in Denmark.

LI M ITATI O N S
Redeeming prescribed ADs during pregnancy is not always equivalent to the use of ADs, and exposure during pregnancy may thus be lower than estimated from pharmacy data.
Factors that influence AD use during pregnancy are multiple and include social, cultural, and economic factors, but in particular, the balance between the potential risk of AD treatment and untreated depression to fetus and pregnant women (Gentile, 2017;Gomez-Lumbreras et al., 2019;Grigoriadis et al., 2013;Petersen, Gilbert, Evans, Man, & Nazareth, 2011;Prady, Hanlon, Fraser, & Mikocka-Walus, 2018). Concerns about the safety of AD exposure to unborn babies are a major determinant of cessation of AD medication use during pregnancy (Petersen et al., 2011), although pregnancy can represent a time of increased vulnerability for the onset or return of depression (Bennett et al., 2004). The postpartum period is also a vulnerable time period with concerns of safety of medicine via breastfeeding and increasing demand for prescriptions of ADs (Munk-Olsen, Gasse, & Laursen, 2012). Most studies have indicated that new types of ADs-like selective serotonin reuptake inhibitors (SSRIs) are generally safe and not associated with birth defects or neurodevelopmental impairment (Petersen, Evans, Gilbert, Marston, & Nazareth, 2016;Prady et al., 2018), which might partly explain the increasing use of AD (and SSRIs in particular) during pregnancy in most countries. Side effects of AD use to the pregnant women are also important factors, which might affect prescriptions of ADs from physicians and adherence of AD use. Surveillance of temporal trends in AD use is basic and essential to research on health of pregnant women and unborn babies and might inform potential safety signals and identify areas requiring further research (Charlton et al., 2015).
In this study, we present the use of ADs in Denmark before, during, and after pregnancy between 1997 and 2016.

| Study population
We identified pregnancies leading to live birth between 1 January 1997 and 31 December 2016 in Denmark (N = 1,243,729) from the Danish Medical Birth Registry (Bliddal, Broe, Pottegard, Olsen, & Langhoff-Roos, 2018;Knudsen & Olsen, 1998). We excluded pregnancies where the pregnant woman's identification number was missing (n = 9), pregnancies that had a missing value of gestational age (n = 11,363) or gestational age <20 weeks (n = 124), leaving 1,232,233 pregnancies in the study population. In Denmark, all residents are assigned a unique identification number recorded in the Danish Civil Registration System (Schmidt, Pedersen, & Sorensen, 2014), and this identification number enables linkage between the many nation-wide registries, such as the Danish Medical Birth Registry and the Danish National Prescription Registry (Kildemoes, Sorensen, & Hallas, 2011) we used in this study. The study was a population-based study and approved by the Danish Data Protection Agency.

| Information on redeeming of ADs
Information on redeemed AD prescriptions was obtained from the Danish National Prescription Registry (Kildemoes et al., 2011), which contains data on all prescription medications dispensed from Danish community pharmacies since 1995. The registry has applied codes for medications using the anatomical therapeutic chemical (ATC) system. ADs were identified by the ATC code N06A. ADs were further categorized into the following groups: selective serotonin reuptake inhibitors (SSRIs), serotonin-norepinephrine reuptake inhibitors (SNRIs), tricyclic ADs (TCAs), and other ADs (ATC codes are found in Table S1). The register contains information on number of package of a medicine in a prescription and the number of defined daily dose (DDD) in one package. DDD is defined as the assumed average maintenance dose per day for a drug used for its main indication in adults (WHO, 2018).

| Information on pregnancy
Information on gestational age and the age of pregnant women at the time of birth was obtained from the Danish Medical Birth Registry (Knudsen & Olsen, 1998). Information on gestational age was reported in days by the midwife attending the delivery using a mandatory coding sheet. Ultrasound measurements have been widely used to determine gestational age in nearly all pregnancies since 1995 (Jorgensen, 1999 1966 and 1993 and 10th revision was used from 1994 onwards (Mors, Perto, & Mortensen, 2011;Munk-Jorgensen & Mortensen, 1997). The codes we used to identify depression and other psychiatric disorders are shown in Table S2. To identify ADs that were prescribed before conception but may have been consumed by women in the early period of pregnancy, we defined the study period of pregnancy as the period from 1 month (30 days) before the estimated time of conception until birth. We also defined the two periods around pregnancy as 1 year (365 days) before and after pregnancy (1 year before pregnancy was defined as the period between 395 and 30 days before the conception). We used the common concept on periods of 1 year before and after pregnancy since we aimed to present temporal trend before and after pregnancy as well rather than comparing the exact prevalence of AD in the three periods.

| Statistical analyses
For pregnancies with at least one AD prescription, we analyzed the number of AD prescription per pregnancy, the number of kind of AD per pregnancy, proportion of pregnancies with a type of AD (SSRIs, SNRIs, TCAs) or each specific AD, and the estimated duration that AD prescriptions during pregnancy might cover. If two or more kinds of ADs were prescribed on the same day, the prescription for each kind of AD was counted as a single prescription. We presented AD prescriptions with two kinds of ADs redeemed during pregnancy, which could be treatment with two kinds of AD or change from one kind of AD to another. We calculated the estimated duration that an AD prescription might cover by multiplying the number of redeemed packs with the number of defined daily doses (DDD) in one package. The total duration that AD prescriptions during pregnancy might cover was the summary of duration of all prescriptions during pregnancy. We presented duration for individual prescriptions and duration that all AD prescriptions might cover for pregnancies with one kind of AD during pregnancy. We presented proportion of pregnancies where women delivered at three age groups (<25, 25-34, 35+ years old) and proportion of pregnancies with a history of psychiatric disorders before the time of delivery (depression, other psychiatric disorders, or no psychiatric disorder) since the two factors could affect prescription and use of ADs.
We analyzed the temporal trend of overall AD prescriptions, and trends according to types of ADs, and specific AD during pregnancy.
We analyzed the temporal trend of AD prescriptions in the 1 year before and 1 year after pregnancy. We further analyzed the trends in overall AD prescriptions according to pregnant women's age at the Abbreviations: SNRIs, Serotonin-norepinephrine reuptake inhibitors; SSRIs, Selective serotonin reuptake inhibitors refer to fluoxetine, citalopram, paroxetine, sertraline, escitalopram in the table. a The two types of AD could be a multitherapy of AD treatment or a change of AD from one type to another. Four pregnancies with a prescription of fluvoxamine during pregnancy were not included in the table because the numbers were too small according to data protection guidelines from Denmark Statistics.
and women from different pregnancies are a cluster factor. The unit in the analyses was pregnancy, and one woman could have several pregnancies.
The analyses were conducted in STATA 15.1 (StataCorp LLC).

| RE SULTS
In the whole cohort, we identified 29,504 (2.4%) pregnancies where the pregnant women redeemed at least one AD prescription from 30 days before pregnancy to the date of the birth.
Among the 29,504 pregnancies where the pregnant women redeemed prescriptions for ADs during pregnancy, 30.3% redeemed one AD prescription and 69.7% redeemed two or more AD prescriptions including 50.6% with three or more AD prescriptions (Table 1).
We identified 96,232 AD prescriptions in total during all pregnancies. Of all the pregnancies with AD prescriptions during pregnancy, 24,643 (83.5%) pregnant women had redeemed only one kind of AD, while 4,315 (14.6%) redeemed two kinds of ADs, and 546 (1.9%) redeemed three or more kinds of ADs. In total, 81.9% of pregnancies were exposed to SSRIs, 13.1% to SNRIs, and 6.4% to TCAs.
Citalopram, sertraline, and fluoxetine were the main types of SSRIs prescribed during pregnancy (Table 1).
Most AD prescriptions (88.5%) cover treatment for 1 month or longer. In 94% of the AD-exposed pregnancies, the estimated duration of treatment was 1 month or longer (Table 2). Even among pregnancies with only one AD prescription, 84.7% would cover 1 month or longer (Table 2).
The proportion of pregnant women who gave a birth at 35 years old and above was 12.7% in 1997 and 20.7% in 2016 (Table S3) (Table S4).

| D ISCUSS I ON
Our study showed that 2.4% of all included pregnancies in Denmark during 1997 and 2016 had at least one AD prescription during pregnancy. In pregnancies with ADs, the majority redeemed more than one prescription (69.7%), most often for a single kind of AD (83.5%). Furthermore, in 94% of the AD-exposed pregnancies, the estimated duration of treatment was 1 month or longer. SSRIs dominated in Denmark, and citalopram was the main type of SSRIs used in Denmark. The more impressive findings were that the proportion of pregnancies with AD prescriptions during pregnancy increased steadily from 1997 to 2011, but decreased sharply thereafter to the end of the study period in 2016. The decreasing trend in AD prescriptions in recent years was also found before and after pregnancy. The decrease in redeeming of AD prescriptions was driven mainly by a decrease in prescription of citalopram.
This was a population-based study covering the entire population of pregnancies in Denmark over 20 years. Both the estimation of the duration of pregnancy (gestational age) and information on ADs were captured from nation-wide registries, and these data are therefore not subject to recall bias. The quality of prescription data is also generally held to be high because of the Danish pharmacy reimbursement structure according to which all citizens receive reimbursement from the Danish regions. Reimbursement is automatically deducted from the price charged at the pharmacy. However, redeeming ADs is not always equivalent to using ADs. A study on the data quality of prescription registration in Denmark indicates that completeness of psychoanaleptics (N06) is 95.1% (Johannesdottir et al., 2012). A study in Denmark has shown that about 85% of people who were prescribed ADs took them regularly, which may also apply to AD use before women were aware of their pregnancy (Lewer, O'Reilly, Mojtabai, & Evans-Lacko, 2015). A large proportion of pregnancies (50.6%) in our study had three or more AD prescriptions, and they were more likely to continue use AD during pregnancy.
The types of ADs used in pregnancy vary across countries  (Zoega et al., 2015). In pregnancies where the women filled two or more kinds of ADs during pregnancy, we did not separate the different situations like combination therapy with two or more kinds of ADs at the same time or change of AD from one kind to another (Benard-Laribiere et al., 2018). This could be a group of interest for further studies.
Studies have shown that many factors including pharmaceutical expenditure, clinical factors, indications for use and patient, and doctor characteristics, can affect the prescription and use of Ads (Bauer et al., 2008;Gomez-Lumbreras et al., 2019). In August 2011, the US Food and Drug Administration issued a safety warning concerning the safety of high doses of citalopram, as the administration of high doses was associated with cardiac corrected QT interval (QTc) prolongation according to findings from an unpublished randomized controlled trial (Howland, 2011). Citalopram was the most often prescribed AD to Danish pregnant women in 2011. This safety concern could, therefore, be one Our study showed that the proportion of women (same as pregnancies) who redeemed AD prescriptions during pregnancy was lower than the proportion of women who redeemed AD prescriptions in the year before pregnancy and the year after pregnancy. We should be aware that the pregnancy period, which is usually around 9 or 10 months in this study since we included 1 month before pregnancy, is shorter than the other two periods, which are 1 year before and after pregnancy. It may slightly affect the proportion of women with AD prescription in these periods although it was unlikely to change the trend over time. Although the study population in the year before pregnancy comprised women who were not pregnant, they may not be representative of the general population of women of fertile age because pregnant women or women who are planning a pregnancy may overall be healthier than women of the same age who are not pregnant.
Further studies are needed to determine whether the trend in AD prescriptions in the population of women of fertile age, men of fertile age, or old people follows the same pattern as that seen around pregnancy among pregnant women.
This study showed that the prevalence of AD use in 1 year after birth was higher than AD use during pregnancy, but lower than AD use Year In UK and Wales, however, prescription of SSRI after birth was higher than prescription of SSRI before pregnancy (Charlton et al., 2015).