SEARCH

SEARCH BY CITATION

Keywords:

  • serotonin reuptake inhibitors (SRIs);
  • benzodiazepines (BZs);
  • congenital malformations;
  • prenatal exposure

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

BACKGROUND: To determine a population-based incidence of congenital anomalies following prenatal exposure to serotonin reuptake inhibitor (SRI) antidepressants used alone and in combination with a benzodiazepines (BZ). METHODS: Population health data, maternal health, and prenatal prescription records were linked to neonatal records, representing all live births (British Columbia, Canada, N=119,547) during a 39-month period (1998–2001). The incidence and risk differences (RD) for major congenital anomalies (CA) and congenital heart disease (CHD), including ventricular and atrial septal defects (VSD, ASD), from infants of mothers treated with an SRI alone, a benzodiazepine (BZ) alone, or SRI+BZ in combinationcompared to outcomesno exposure. RESULTS: Risk for a CA or CHD did increase following combined SRI+BZ exposure compared with no exposure. However, using a weighted regression model, controlling for maternal illness characteristics, combination therapy risk remained significantly associated only with CHD. The risk for an ASD was higher following SRI monotherapy compared with no exposure, after adjustment for maternal covariates. Dose/day was not associated with increased risk. CONCLUSIONS: Infants exposed to prenatal SRIs in combination with BZs had a higher a incidence of CHD compared to no exposure, even after controlling for maternal illness characteristics. SRI monotherapy was not associated with an increased risk for major CA, but was associated with an increased incidence of ASD. Risk was not associated with first trimester medication dose/day. Birth Defects Research (Part B) 83:68–76, 2008. © 2008 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Recent scientific and public attention has focused on studies reporting an increased risk for congenital malformations, particularly cardiac anomalies following 1st trimesterto serotonin reuptake inhibitor (SRI) antidepressant medications (Alwan et al., 2007; GlaxoSmithKline, 2006; Louik et al., 2007). While SRIs are commonly used to manage depression during pregnancy (Oberlander et al., 2006), some studies have reported an increased risk for congenital (Alwan et al., 2007; Chambers et al., 1996; Wogelius et al., 2006) and cardiac anomalies (Diav-Citrin et al., 2005; GlaxoSmithKline, 2006; Malm et al., 2005) while others have failed to find such an association (Altshuler et al., 1996; Einarson et al., 2005; Ericson et al., 1999; Hendrick et al., 2003; Källén et al., 2007; Simon et al., 2002). Recently, Berard et al. (2007) reported an increased risk for congenital and cardiac anomalies also being associated with an increasedtrimester paroxetine dose. Concerns about prenatal exposure led Health Canada (2004) and the FDA (US Food and Drug Administration, Medwatch, 2005) to issue warnings about paroxetine use during pregnancy.

To date, a variety of cohort and population-based study designsbeen used to assess relationships between gestational exposure and risk for anomalies (Berard et al., 2007; Ericson et al., 1999; GlaxoSmithKline, 2006; Källén et al., 2007; Simon et al., 2002; Wogelius et al., 2006). However, outcomes have been conflicting and our understanding of neonatal risk in this setting remains uncertain due to a number of key methodological challenges (Hines et al., 2004). While the need for large study populations is frequently acknowledged, congenital anomalies typically occur at very low incidences and these studies have been challenged by uncertainty abouttiming of gestationalexposure, maternal recall bias, lack of precise definitions distinguishing major from minor anomalies, controlling for the impact of depression itself, or other maternal illness that may be associated with congenital anomalies and the lack of matched pharmacological and related maternal health data (Alwan et al., 2007; Louik et al., 2007). Moreover, studies have not investigated the effects of combined psychotropic medication exposure and have only controlled for such exposures by eliminating them from analysis (GlaxoSmithKline, 2006; Wogelius et al., 2006). In particular, studies have not examined neonatal outcomes following prenatal SRI use in combination with benzodiazepines, accounting for the impact of maternal illness severity or the length of 1st trimester use.

To address these gaps in the literature, we undertook a population-level study using health data linking all maternal health data with records for prescriptions dispensed during pregnancy and birth outcomes for their infants over a 39-month period in our province (British Columbia) to study whether the risk for major congenital malformations and congenital heart defects differs between first trimester SRI+BZ exposure and no exposure at all. Secondarily, we also compared neonatal outcomes following monotherapy SRI exposure with outcomes with no exposure. We also sought to determine whether risk for congenital anomalies is associated with first trimester SRI medication dose controlling for length of first trimester exposure and maternal illness severity, which might have also contributed to increasing neonatal risk. We expected that neonatal risk would increase with poly-psychotropic drug exposure beyond no exposure, even when controlling for maternal illness characteristics.

METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

This study was undertaken with approval from the University of British Columbia (UBC) Research Ethics Board, the Children's and Women's Health Centre of British Columbia Research Review Committee, the BC Ministry of Health Services, and the BC College of Pharmacists. Data analysis was undertaken by T.O., J.A., and W.W.

Data Set Compilation

Data used in this study came from five administrative sources housed in the BC Linked Health Database (Chamberlayne et al., 1998) (BC registry of births, hospital separation records, the PharmaCare registry of subsidized prescriptions; the Medical Services Plan physician billing records; and the registry of Medical Services Plan subscribers) linked to PharmaNet, a province-wide network recording all prescriptions dispensed by BC pharmacists outside hospitals. These data were processed by the Centre for Health Services and Policy Research (CHSPR), UBC; PharmaNet provided records with the same unique, non-identifying study ID as was provided by CHSPR to enable data linkage. The cohorts used in this study were assembled from records of 203,520 registered live births (hospital and home births) in British Columbia occurring between April 1, 1997 and March 31, 2002. To ensure accurate matching between all data sets, and accounting for data entry errors and records for multiple births, the final study cohort comprised records related to 119,547 live births, representing 92.7% of the live births in British Columbia (previously described in Oberlander et al., 2006). Of the total 356,727 prescriptions for psychotropic medications in the PharmaNet data set, we identified 75,456 for one of the following SRIs: citalopram, fluoxetine, fluvoxamine, paroxetine, sertraline, and venlafaxine. The file identified the drug by brand name and generic name, the date that the drugs were dispensed, and the number of days supplied together with a unique study number for the mother.

To account for maternal illnesses that may have also contributed to congenital anomalies, physician billing data were used to determine whether the mothers had diseases and complications related to pregnancy (ICD9 codes from 640 to 648, Complications Mainly Related to Pregnancy). In addition, any diagnosis of epilepsy or seizures was also identified from maternal records regardless of timing of exposure.

Hospital discharge records report gestational age based on last menses. We define exposure in the first trimester to have occurred if the period for which SRIs and/or benzodiazepines were dispensed (i.e., days of dosing covered by the prescription) overlapped with the period from the LMP to LMP plus 90 days (first trimester). To avoid any confounding effect from anticonvulsant exposure, and the inherent multiple maternal diagnoses and neonatal risks associated with these medications, 359 records were removed where neonates had first trimester exposure to this class of medication. We define poly-drug exposure to have occurred if both an SRI and a BZ were dispensed for the same day during the first trimester.

Study Group Identification

Information on diagnosis of maternal mood was obtained from Ministry of Health Services, Medical Services Plan billing records (ICD9 diagnostic code). To define a diagnosis of depression, all ICD9 codes that explicitly referred to depression (National Centre for Health Statistics, 2006; Teleplan Record Specifications version 4.0,2004) were selected (Oberlander et al., 2006).

Exposure groups were defined as SRI monotherapy and SRI+benzodiazepines (BZ) used in combination. Outcomes were compared between infants exposed to SRI monotherapy and SRI+BZ combination with infants with no exposure to either of these drugs in the first trimester, respectively.

Neonatal Outcomes

Based on previous studies (Berard et al., 2007; GlaxoSmithKline, 2006; Wogelius et al., 2006), neonatal outcomes were identified using ICD9 codes for major congenital anomalies (codes: 740.0 to 759.9), and the subset of cardiovascular defect (745.0–747.9)). We excluded the following congenital anomalies, which were defined as minor by Berard et al. (2007) (743.6, 744.1, 744.2–744.4, 744.8, 744.9, 747.0, 747.5, 750.0, 752.4, 752.5, 754.6, 755.0, 755.1, 757.2–757.6, 757.8, 757.9, 758.4). Specific codes for ventricular septal defects (VSD) (745.4), and atrial septal defects (ASD) (745.5) were also used.

Data Analysis

Data analysis was undertaken using four approaches: First, given that congenital anomalies are rare events within large populations, where odds ratios can be very large when the numerator is close to zero, we report risks and risk differences (RDs; 95% confidence intervals), rather than odds ratios (Davies et al., 1998; Kraemer et al., 2003). Confidence intervals for incidence were calculated using Newcombe and Altman's method (Newcombe and Altman, 2000) and confidence intervals for the risk difference were calculated using Newcombe's method ten (Newcombe, 1998).

Second, because linked-population level health data do not permit randomization between exposure groups enabling us to control for maternal risk factors that could have either directly or indirectly contributed to a risk for congenital anomalies, we supplemented the comparisons of raw differences with regression models to control for maternal characteristics that could have also contributed to neonatal risk. We used weighted linear probability models (corrected for heteroskedasticity) to provide estimates of risk differences. In these models, we controlled for maternal illness characteristics, diseases, and complications of pregnancy diagnosed more than 60 days before birth, depression in the first trimester, and a dummy variable indicating that the mother filled a prescription after she knew that she was pregnant (see below), plus a variable indicating whether the patient had been prescribed methadone. Methadone use during pregnancy was used as a proxy marker for other maternal health characteristics, such as poor nutrition and other drug use, which could also contribute to an increased risk for congenital anomalies (but were unmeasured in our administrative health data). Namely, if mothers, who were also treated with methadone and had related health risk factors, were also more likely to be taking an SRI, this would bias our results.

Third, compliance among patients starting antidepressant therapy has been frequently noted (Åkerblad et al., 2003; Reis et al., 2004) and non-compliance could mask associations (i.e., bias the relationships towards zero), and therefore might alter the significance of any exposure-outcome relationships. To control for non-compliance, we constructed a variable that indicated that the mother had filled a prescription more than 50 days after the LMP and included it in the regression analyses. It was assumed that mothers, except possibly the hidden non-compliers, who took the trouble to fill a prescription, would have used the previously dispensed pills.

Fourth, to study the relationship between first trimester medication dose and congenital anomalies, we developed indicators of whether the dose was low, medium, or high in two steps. First, because of differences in dose range inherent to each SRI and BZ medication, each dispensed medication dose was converted to a z score. Second, we classified the dose for SRI monotherapy as low if the z score was less than −0.5, medium if the z score was between −0.5 and 0.5, and high if the z score was greater than 0.5. For combined BZ and SRI, we used cutoffs of −0.75 and 0.75. For some patients, dose changed over time. In these cases, we used the dose in effect on the estimated date of conception. Differences in the rates of anomalies associated with high doses (compared with medium and low doses) and for high and medium doses (compared with low doses) for both SRI monotherapy and for SRI+BZ were assessed using Fisher's exact test and logistic regression analysis. In the regression analysis, we controlled for the variables described above, and in a separate analysis we additionally controlled for days of exposure to SRI medication.

We calculated the incidence of anomalies (and 95% confidence intervals) associated with each specific SRI both when used individually and when used in combination with a BZ. We calculated the difference in incidence (risk difference) compared with no exposure, and the associated 95% confidence intervals for the difference. We also produced estimates of the risk difference after controlling for potential confounders described above using a weighted linear probability model. We tested the null hypothesis that the risk difference was the same for all SRI's using an F-test.

To ensure robustness of our regression results, we conducted sensitivity analyses by controlling for (1) exposure to clonazepam or clobazam since these drugs are sometimes used as anticonvulsants; (2) restricting the period of exposure to the period of organogenesis; (3) controlling for exposure to antipsychotics since these drugs and/or the attendant illness may present risk to the fetus, and to non-SRI antidepressants; and (4) by using logistic regression and probit regression to estimate the parameters of the models.

Confidence intervals using Newcombe's and Altman's methods were computed by the authors. All other statistical analysis was conducted using Stata SE 10 (STATA SE10, 2006).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Maternal Demographic Characteristics

Cohort and medication exposure group characteristics are tabulated in Table 1. Paroxetine was the most commonly prescribed SRI, and lorazepam and clonazepam the most frequent benzodiazepines prescribed during the first trimester. Key differences in maternal characteristics emerged between exposure groups. Mothers who had received an SRI alone had 1.8 times more family physician visits, were three times more likely to have had drugs subsidized through the welfare system, and were 16 times more likely to have been diagnosed as depressed in the year before LMP with the “no exposure group” (i.e., not depressed and not receiving an SRI during pregnancy).

Table 1. Maternal Demographic Characteristics: First Trimester Exposure
Maternal medication groupMean maternal agePrenatal care visits (all trimesters)Average number of diagnoses of depression in the year before LMPAverage number of visits to a psychiatrist in he year before LMPAverage number of visits to a physician in the year before LMP
Depression exposure alone (n=7,883)29.611.30.90.315.3
SRIs only (n=2,25)a29.611.03.20.921.3
Benzodiazepines only (n=968)b30.110.81.50.723.8
SRIs+benzodiazepines (n=392)29.910.64.92.931.6
No exposure (n=10,720)29.510.80.20.011.5
 
Frequency of particular drug use in a specific therapeutic class in the first trimester:
SRIs:
 DrugFrequency of use (%)    
Paroxetine37.0    
Sertraline24.3    
Fluoxetine24.2    
Venlafaxine7.1    
Fluvoxamine4.6    
Citalopram2.8    
 
Benzodiazepines:
 DrugFrequency of use (%)    
Lorazepam44.0    
Clonazepam21.4    
Oxazepam15.0    
Alprazolam6.8    
Temazepam5.1    
Diazepam5.0    
Clobazam1.6    
Bromazepam0.3    
Triazolam0.3    
chlordiazepoxide hcl0.2    
flurazepam hcl0.2    
Nitrazepam0.1    
Clorazepate dipotassium0.0    

Major Congenital Anomalies

When an SRI was used in combination with BZ, the crude risk difference for CA relative to no exposure was 2.43 percentage points (0.49 to 5.31%). However, after adjustment for potential confounders, the risk difference was no longer statistically significant (Table 2a). Compared with no exposure, SRI monotherapy did not increase the risk for major CAs.

Table 2a. Major congenital Anomalies (incidence, 95% CI and RD compared with risk associated with no exposure.)
Exposure groupMajor congenital anomalies (N)Incidence (95% confidence interval)Unadjusted Risk Difference (95% confidence interval)Regression adjusted Risk Difference (95% confidence interval)
No exposure (n=107,320)3,3693.14 (3.04–3.25)  
SRI's only (n=2,625)752.86 (2.29–3.57)−0.28 (−0.86–0.43)−0.61 (−1.44–0.21)
Benzodiazepines only (n=968)313.20 (2.27–4.51)0.06 (−0.88–1.37)−0.41 (−1.51–0.69)
SRI's + benzodiazepines (n=359)205.57 (3.63–8.45)2.43 (0.495.31)1.65 (−0.49–3.79)

Congenital Heart Disease (CHD)

An SRI used in combination with a BZ increased the risk of CHD by 1.19 percentage points compared with no exposure. The risk difference remained statistically significant after adjustment for potential confounders, adjusted RD=1.18% (0.18 to 2.18%) (Table 2b). SRI monotherapy did not increase the risk for CHD overall compared with no exposure. The risk for VSD did not increase with poly-drug exposure (Table 2c), while the risk for an atrial septal defect (ASD) (Table 2d) increased with SRI monotherapy compared with no exposure RD=0.16% (0.03 to 0.43%). This difference remained statistically significant after adjustment for potential confounders, adjusted RD=0.21% (0.05 to 0.36%).

Table 2b. Cardiovascular Congenital Defects
Exposure groupCardiovascular defect (N)Incidence (95% confidence interval)Unadjusted Risk Difference (95% confidence interval)Regression adjusted Risk Difference (95% confidence interval)
No exposure (n=107,320)5120.48 (0.44–0.52)  
SRI's only (n=2,625)170.65 (0.40–1.03)0.17 (−0.08–0.56)0.21 (−0.14–0.56)
Benzodiazepines only (n=968)50.52 (0.22–1.20)0.04 (−0.26–0.73)−0.13 (−0.55 to 0.29)
SRI's+benzodiazepines (n=359)61.67 (0.77–3.60)1.19(0.293.12)1.18(0.182.18)
Table 2c. Ventricular Septal Defects (VSD)
Exposure groupVSD (N)Incidence (95% confidence interval)Unadjusted Risk Difference (95% confidence interval)Regression adjusted Risk Difference (95% confidence interval)
No exposure (n=107,320)2190.20 (0.19–0.25)  
SRI's only (n=2,625)60.23 (0.13–0.55)0.02 (−0.09–0.33)0.10 (−0.12–0.33)
Benzodiazepines only (n=968)30.31 (0.11–0.91)0.11 (−0.11–0.69)0.05 (−0.26–0.36)
SRI's + benzodiazepines (n=359)20.56 (0.15–2.01)0.35 (−0.07–1.09)0.35 (−0.26–0.96)
Table 2d. Atrial Septal Defects (ASD)
Exposure groupASD (N)Incidence (95% confidence interval)Unadjusted Risk Difference (95% confidence interval)Regression adjusted Risk Difference (95% confidence interval)
  1. Notes to Tables 2a–d:

  2. a

    • Major congenital anomalies refers to ICD9 codes from 740 to 759 excluding minor anomalies (743.6, 744.1, 744.2–744.4, 744.8, 744.9, 747.0, 747.5, 750.0, 752.4, 752.5, 754.6, 755.0, 755.1, 757.2–757.6, 757.8, 757.9, 758.4).

  3. b

    • Cardiovascular congenital defects refers to ICD9 codes from 745 to 747 excluding 747.0 patent ductus arteriosus and 747.5 Absence or hypoplasia of umbilical artery.

  4. c

    • VSD refers to ICD9 code 745.4.

  5. d

    • ASD refers to ICD9 code 745.5.

  6. e

    • Table excludes pregnancies with exposure to an anticonvulsant during the first trimester (357 observations) and pregnancies with exposure to both SRI & BZ in the first trimester, but not at the same time (35 observations).

  7. f

    • Regression analysis controls for characteristics listed in Table 1, diseases and complications of pregnancy diagnosed more than 60 days before birth, depression in the first trimester and a dummy variable indicating that the mother filled a prescription after she knew that she was pregnant. Sample includes 7,883 depressed, not medicated mothers, N=119,155.

  8. g

    • We report results from the weighted least squares regression analysis. We repeated the analysis using logistic regression analysis. The conclusions regarding statistical significance were unaffected. The OR for cardiovascular congenital defects for SRI+BZ was 3.077 95% CI (1.22–7.78).

No exposure (n=107,320)750.07 (0.06–0.09)  
SRI's only (n=2,625)60.23 (0.10–0.50)0.16(0.030.43).21 (0.05–0.36)
Benzodiazepines only (n=968)10.10 (0.02–0.58)0.03 (−0.05–0.51)0.02
(−0.16–0.20)    
SRI's + benzodiazepines (n=359)00.00 (0.00–1.06)−0.07 (−0. 09–0.99)−0.01 (−0.31–0. 30)

Specific SRI Exposure and Risk for Anomalies

No differences between SRI medications were observed for the risk of major anomalies (Table 3a,b). An increased risk for cardiovascular defects was observed following exposure to citalopram (n=3; 2.97%; 95% CI (1.02–8.37); RD 2.49% (0.54–7.89) when compared to no exposure. While this risk difference remained statistically significant after controlling for potential confounders, the F-test did not reject the null hypothesis that the risk difference did not differ across SRI mono-therapies. Specific CHDs identified in the 3 citalopram-exposed neonates are listed in Table 3b.

Table 3a. SRI Monotherapy Major Congenital Anomalies (incidence, 95% CI and RD compared with risk associated ith no exposure.)
Exposure groupMajor congenital anomalies (N)Incidence (95% confidence interval)Unadjusted Risk Difference (95% confidence interval)Regression adjusted Risk Difference (95% confidence interval)
No exposure (n=107,320)3,3693.14 (3.04ndash;3.25)  
Citalopram (n=101)43.96 (1.55–9.74)0.82 (−1.59–6.61)0.40 (−3.13–3.93)
Fluoxetine (n=638)213.29 (2.16–4.98)0.15 (−0.98–1.84)−0.26 (−1.68–1.17)
Fluvoxamine (n=119)21.68 (0.46–5.92)−1.46 (−2.68–2.78)−1.52 (−4.02–0.98)
Paroxetine (n=993)292.92 (2.04–4.16)−0.22 (−1.10–2.49)−0.56 (−1.70–0.59)
Sertraline (n=608)193.13 (2.01–4.83)−0.01 (−1.13–1.69)−0.41 (−1.84–1.02)
Venlafaxine (n=250)62.40 (1.10–5.14)−0.74 (−2.04–2.00)−1.18 (−3.20–0.84)
Table 3b. SRI Monotherapy Cardiovascular Congenital Defects
Exposure groupMajor congenital anomalies (N)Incidence (95% confidence interval)Unadjusted Risk Difference (95% confidence interval)Regression adjusted Risk Difference (95% confidence interval)
  1. a

    *CHD for infants with citalopram exposure: Infant 1: 745.4 Ventricular septal defect. Infant 2: 747.0 Patent ductus arteriosus; 745.5 Ostium secundum type atrial septal defect; 746.8 Other specified anomalies of heart; 746.6 Congenital mitral insufficiency. Infant 3: 745.5 Ostium secundum type atrial septal defect; 745.4 Ventricular septal defect.

No exposure (n=107,320)5120.48 (0.44–0.52)  
Citalopram (n=101)3*2.97 (1.02–8.37)2.49 (0.547.89)2.28 (0.194.36)
Fluoxetine (n=638)50.78 (0.34–1.82)0.31 (−0.14–1.35)0.08 (−0.54–0.70)
Fluvoxamine (n=119)00.00 (0.00–3.13)−0.48 (−0.52–2.65)−0.55 (−1.45–0.36)
Paroxetine (n=993)70.70 (0.34–1.45)0.23 (−0.14–0.97)0.12 (−0.38–0.62)
Sertraline (n=608)30.49 (0.17–1.44)0.02 (−0.31–0.96)−0.09 (−0.65–0.47)
Venlafaxine (n=250)10.40 (0.07–2.23)−0.08 (−0.41–1.75)0.01 (−0.77–0.79)

An increased risk for both CA and CHD was observed following exposure to fluoxetine+BZ compared with no exposure (Table 3c, Table 3d). The risk difference for CA remained statistically significant after controlling for potential confounders, but the risk difference for CHD did not. The F-test did not reject the null hypothesis that risk differences did not differ across SRI's when they were used in combination with BZ.

Table 3c. Individual SRI+BZ Major Congenital Anomalies (incidence, 95% CI and RD compared with risk associated with no exposure.)
Exposure groupMajor congenital anomalies (N)Incidence (95% confidence interval)Unadjusted Risk Difference (95% confidence interval)Regression adjusted Risk Difference (95% confidence interval)
No exposure (n=107,320)3,3693.14 (3.04–3.25))  
Citalopram (n=25)14.00 (0.71–19.54)0.86 (−2.43–16.41)−1.78 (−8.24–4.67)
Fluoxetine (n=81)78.64 (4.25–16.78)5.50 (1.1113.64)5.18 (0.3010.07)
Fluvoxamine (n=25)28.00 (2.22–24.97)4.86 (−0.92–21.83)4.84 (−3.67–13.35)
Paroxetine (n=141)74.96 (2.43–9.89)1.83 (−.72–7.75)0.98 (−2.30–4.26)
Sertraline (n=84)44.76 (1.87–11.61)1.62 (−1.27%–8.48)1.11 (−3.03–5.25)
Venlafaxine (n=38)12.63 (0.47–13.50)−0.51 (−2.68–10.36)−0.46 (−5.39–4.47)
Table 3d. Individual SRI+BZ Cardiovascular Congenital Defects
Exposure groupMajor congenital anomalies (N)Incidence (95% confidence interval)Unadjusted Risk Difference (95% confidence interval)Regression adjusted Risk Difference (95% confidence interval)
  1. a

    This table is based on the 3,423 mothers who, in the first 90 days of pregnancy, had been exposed to an antidepressant but had not been exposed to an anticonvulsant, antipsychotic or benzodiazepine.

No exposure (n=107,320)5120.48 (0.44–0.52)  
Citalopram (n=25)00.00 (0.00–13.32)−0.48 (−0.52–12.84)−0.89 (−3.00–1.22)
Fluoxetine (n=81)22.47 (0.68–8.56)1.99 (0.208.09)1.94 (−0.30–4.18)
Fluvoxamine (n=25)00.00 (0.00–13.32)−0.48 (−0.52–12.84)−0.63 (−2.62–1.36)
Paroxetine (n=141)21.42 (0.39–5.02)0.94 (−0.09–4.55)0.70 (−0.79–2.19)
Sertraline (n=84)11.19 (0.21–6.44)0.71 (−0.27–5.96)0.82 (−0.95–2.58)
Venlafaxine (n=38)12.63 (0.47–13.50)2.15 (−0.01–13.02)1.90(−1.41–5.22)

Impact of First Trimester Dose

There was no association between increased first trimester dose and risk for major congenital anomalies or congenital heart disease in the simple comparisons or in the regression models, for SRI+BZ or SRI monotherapy even controlling for length of time of medication use and maternal illness characteristics.

Sensitivity Analysis

The conclusions were unaffected by any of the sensitivity analyses with one notable exception. The increased risk associated with citalopram monotherapy did not remain statistically significant when we restricted exposure to the period of organogenesis.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

When SRIs were used in combination with a BZ, risk for CA and CHD increased over risks associated with no exposure. However, when controlling for maternal illness characteristics, using a weighted regression model, only an increased risk for CHD remained. With the exception of an increased risk for CHDs following citalopram exposure, monotherapy SRI or BZ exposure was not associated with an overall increased risk for CA or CHD. Risk for major congenital anomalies was not associated with the dose of medication used per day during the first trimester, even controlling for length of first trimester medication use. SRI+BZ exposure increased the risk for CA when the exposure included fluoxetine. However, the numbers of fluoxetine+BZ with a CA and citalopram-exposed neonates with a CHD were very small (n=7 and n=3, respectively). Moreover, given that the majority of VSDs and ASDs are frequently considered clinically insignificant (Hoffman et al., 2002), the clinical relevance of these anomalies (Table 3b) could not be determined, nor could we attribute a causal relationship between these exposures and these anomalies. A surveillance or diagnostic bias associated with prenatal mental maternal mental illness, SRI and BZ treatment leading to an increased observation of these anomalies could not be ruled out. SRIs have been associated with increased admission to special care nurseries and an increased risk for all investigations (i.e., clinical exams, cardiac echocardiograms etc) inherent to care in that setting. Importantly, the risk for major congenital anomalies associated with exposure to depressed maternal mood alone was not significantly different from no exposure.

Comparing our findings with previously reported outcomes remains confusing and contradictory due to multiple study designs, outcomes, drug exposures, and the impact of maternal illness itself. Some population-based studies have reported an increased risk for omphalocele and craniosynostosis following paroxetine exposure (Alwan et al., 2007), while others found no increase in risk for congenital malformations (Ericson et al., 1999; Simon et al., 2002). An increased risk for ventricular and atrial septal defects was reported by Källén et al. (2007) using population level data following paroxetine exposure. Furthermore, using a case control study design (Louik et al., 2007) and unpublished propriety industry surveillance data (GlaxoSmithKline, 2006) indicate that first trimester paroxetine exposure increases the risk for congenital malformations and cardiac defects (OR 1.82; 95% CI 1.17–2.82). However, when controlling for other exposures, an increased risk for cardiovascular malformations was not observed (OR 1.54, 95% CI 0.81–2.92) (GlaxoSmithKline, 2006).

In contrast to previous reports (Berard et al., 2007; GlaxoSmithKline, 2006; Wogelius et al., 2006), we specifically sought to advance our understanding of the teratogenic risk associated with prenatal SRI exposure by distinguishing risk differences between SRI+BZ poly-drug exposure and no exposure, accounting for rare outcomes in large data sets and control for the impact of maternal illness characteristics that could have also influenced neonatal risk. Our study found that the risk for cardiac anomalies increasedwhen SRIs were used in combination with a benzodiazepine medication, even when controlling for maternal age, illness, and income, beyond the risk associated with mono-drug exposure. Such findings might reflect a poly-drug interaction (competitive inhibition) and a possible pharmacological mechanism associated with the increased risk. SRI monotherapy exposure was only associated with an increasedfor atrial septal defects.poly-drug-associated risk was also higher than the risk associated with exposure to depressed maternal mood alone. Finally, using regression models, including length of prenataltrimester exposure, we were not able to identify an increased risk associated with an increased SRI dose/day.

From the outset, we were aware that we needed to overcome a number of key methodological limitations that have faced previous work in this field. These included the risk for mis-reporting confidence intervals inherent to studies of very rare events in large populations, the variety and overlap between neonatal diagnostic outcomes, imprecision on length and timing of 1st trimester exposure, and the impact of maternal illness itself. To deal with these challenges, we applied the following strategies: First, to obtain measures of timing and duration of 1st trimester SRIs and benzodiazepines exposure, we used the date of the LMP and the actual date on which the drugs were dispensed. Second, using ICD9 diagnostic codes, we were able to define mutually exclusive neonatal outcomes for major anomalies. Third, because these anomalies occur very rarely, even in large populations, we report risk differences and confidence intervals using Newcombe's and Altman's methods rather than reporting odds ratios and p values produced by standard statistical approaches. Fourth, given that maternal illness itself may carry a risk for anomalies but that key aspects of maternal health can not be directly assessed using population level health data, we used characteristics of maternal illness severity (i.e., times visiting a psychiatrist in the year before LMP, number of times receiving a diagnosis of depression in the year before LMP, etc.) in a regression model to assess whether maternal illness severity was also associated with increased risk for anomalies. Finally, given that risk may be associated with increased 1st trimester dose, but that inherent to SRI and BZ dose is length of exposure, we also examined the impact of dose accounting for duration of exposure and maternal illness severity.

The increased risk associated with poly-drug exposure was not altogether unexpected. Benzodiazepines have been used to manage anxiety during pregnancy and have been associated with cleft palate, musculoskeletal abnormalities, and cardiovascular abnormalities (ASD) (Iqbal et al., 2002), though results have not been consistent (Tikkanen and Heinonen, 1992; Zierler Rothman, 1985) and typically benzodiazepines are not considered teratogenic. As well, multi-drug interactions or increased maternal plasma concentrations resulting from combinations of clonazepam and phenobarbital have been associated with increased risk for congenital heart disease and hip dislocation (Czeizel et al., 1992). The increased risk for anomalies observed with SRI+ BZ exposure suggests similar mechanisms might account for increased risk with poly-drug exposures.

Previously, we observed significantly increased paroxetine levels in neonates of mothers treated with SRI medication and the benzodiazepine clonazapine (Oberlander et al., 2004), findings that may have reflected competitive inhibition of the hepatic metabolic isoezyme, CYP3A4. These findings also raise the question of whether such increased SRI drug levels may increase risk for congenital anomalies associated with poly-drug exposure where pharmacological factors may lead to increased drug levels and associated neonatal risk. During gestation, serotonin and related receptors play key roles as trophic factors in the development of the monoamine-dependent cardiac structures and functions (Nebigil and Maroteaux, 2001). It is conceivable that changing prenatal intrasynaptic levels of serotonin by altering drug levels, could change this trophic role, leading to an increased risk for congenital anomalies and cardiac malformations (Nebigil et al., 2001). Moreover, multi-drug interactions may lead to increased drug levels secondary to a competitive inhibition for common metabolic pathways shared by both SRIs and BZs. Moreover, increased parent drug or metabolite levels, or the generation of reactive intermediates, are all possible mechanisms. Increased maternal 3rd trimester and delivery paroxetine plasma levels have also been reported following the use of this SRI with the benzodiazepine clonazepam (Oberlander et al., 2004); this may reflect competitive inhibition of the hepatic CYP450 3A4 isoenzyme that metabolizes both medications. Other examples of pertinent drug-drug interactions include increased plasma concentrations of alprazolam and diazepam by fluvoxamine, and increased serum concentrations of lorazepam by valproic acid (Hemeryck and Belpaire, 2002; Perucca, 2002).

A number of limitations associated with the use of administrative health data need to be mentioned. While our data set was able to identify the congenital anomalies diagnosed, we were not able to verify the exact time at which the diagnosis was made or whether it was verified by echocardiogram evaluations or other imaging tests. The incidence of congenital anomalies and CHD in our population is consistent with other reports, though there are substantial variations depending on the severity of the anomaly and population studied. The occurrence of CHDs between studies ranges from 4/1,000–50/1,000 with the incidence of moderate to severe forms at about 6/1,000 (Hoffman and Kaplan, 2002). Isolated VSDs vary from 2 to 5% (Roguin et al., 1995) and ASDs range from 0.03–0.1% (mean 0.09%) (Hoffman and Kaplan, 2002). The clinical significance or severity of the anomaly could not be verified in our current study and many VSDs and ASDs are minor and resolve without any clinical sequeala (Hoffman and Kaplan, 2002). While we attempted to control for maternal health (i.e., diabetes, etc.) and severity of illness, which could have potentially contributed to congenital anomalies, the use of tobacco and alcohol during pregnancy or maternal weight gain, illicit drug use, or maternal parity could not be directly studied with these administrative data sets.

Using population-based linked health cohort data, we report an increased risk for cardiac malformations when first trimester SRIs medication use was combined with benzodiazepines. The risk associated with SRI monotherapy did not increase beyond exposure to depression or no drug/no depression. Risk for anomalies in SRI monotherapy was not associated with maternal doses/day, maternal age, nor measures of illness severity. These results highlight the importance of considering the competitive interactions or toxicity associated with poly-psychotropic drug exposure. While poly-drug exposure increased the risk for all anomalies, the risk associated with combined exposures remained very low and the clinical significance of the anomalies remains to be determined. Moreover, the added risk associated with SRI medication exposure should be evaluated in the context of depressed maternal mood during pregnancy. The identification and treatment of depression during pregnancy is an urgent health care concern for mothers, their families, and health care professionals. Antenatal depression is by itself a risk factor for maternal and fetal/neonatal health that might require pharmacological management and these findings only highlight one aspect of the risks associated with medical management of perinatal depression. None of these findings should diminish the urgency to identify and manage maternal mental illness during pregnancy. This decision should be made by an informed patient with her physician as risk needs to be assessed on a case-by-case basis (Wisner et al., 1999).

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The researchers gratefully acknowledge funding from the BC Ministry of Children and Family Development through the Human Early Learning Partnership. The authors also acknowledge the financial support of The Michael Smith Foundation for Health Research. T.F.O. is supported by a HELP Senior Career Award and is the R. Howard Webster professor in Child Development (UBC, Faculty of Graduate Studies). We are also grateful to Colleen Fitzgerald for her administrative help and to Ursula Brain and Tracey Weir for their editorial comments. The views presented in the article are solely those of the authors and do not represent the policy of HELP or the Province of British Columbia. Dr. Oberlander has had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. None of the authors has a conflict of interest with these data or our findings. As the corresponding author, I have had full access to all the data in the study and had final responsibility for the decision to submit for publication (T.F.O.).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES