3.4.1 Human Data
184.108.40.206 In utero exposures
Case reports and case series reported adverse effects in infants exposed to ﬂuoxetine in utero but the Expert Panel noted that such studies are not adequate for evaluating developmental effects of ﬂuoxetine. The Expert Panel focused their evaluation of prenatal ﬂuoxetine toxicity on studies with denominators, i.e., the sampling frame that was used for selecting study subjects could be identiﬁed.
The most complete study for evaluating toxicity in infants following prenatal exposure to ﬂuoxetine was conducted by Chambers et al. (89). The study evaluated pregnancy outcomes in 228 women taking ﬂuoxetine and a comparison group of 254 women exposed to either acetaminophen, dental x-ray, or < 1 oz. alcohol/week prior to learning of pregnancy. The ﬂuoxetine group was divided into early or late exposures (before or after 25 weeks gestation, respectively). No difference in major mal-formations was found between liveborn infants exposed to ﬂuoxetine in early gestation and controls.
However, the proportion of infants with multiple minor anomalies was increased. Late pregnancy exposures were associated with increased incidence of prematurity, reduced birth weight and length at full term, and poorer neonatal condition characterized by admission to special care nursery and adaptation problems (e.g., jitteriness, tachypnea, hypoglycemia, hypothermia, poor tone, respira-tory distress, weak or absent cry, or desaturation on feeding). Relative risks for late exposure were calculated by comparing late and early exposures and adjusting for numerous confounding factors.
Relative risks were 4.8 (95% CI: 1.1 – 20.8) for prematurity, 2.6 (95% CI: 1.1 – 6.9) for admission to a special care nursery, and 8.7 (95% CI: 2.9 – 26.6) for poor neonatal adaptation. Though conclusions about major malformations are limited by small sample size, the Expert Panel found this study to
have well-deﬁned procedures and outcome measures and a thorough assessment of outcome com-pared to other studies. However, the Panel urged caution in interpreting the long-term implications of multiple minor anomalies. Also, the possible confounding effects of maternal depression need to be considered.
Two other large studies involving medical record reviews conducted by Simon et al. (104) and Ericson et al. (102) were felt to be well-designed and contribute to the utility of available data, although the Expert Panel noted that the medical record reviews in these studies do not provide outcome measures as strong as those in Chambers et al. (89).
Simon et al. (104) performed a record linkage study among members of a population based prepaid health plan to identify pregnancies for whom 1 or more prescriptions were ﬁlled for antidepressants within a 270 day interval before delivery. Signiﬁcant reductions in gestational age at delivery (-0.9;
95% CI: -0.5,-1.3) were observed for women with SRI exposure (129/185 women using ﬂuoxetine) during pregnancy in comparison to matched unexposed women after adjusting for maternal tobacco use, other substance use, race, and number of prior births. Risk of preterm birth dichotomized as
< 36 weeks gestation conferred a signiﬁcantly increased risk for SRI exposure (OR:: = 4.38; 95% CI:
1.57, 12.22). Mean birth weights were signiﬁcantly lower among SRI exposed infants in comparison to unexposed (-172 grams; 95% CI: -46, -299, controlling for the above potential confounders), but no effects were noted on birth weight when corrected for gestational age. None of these differences was signiﬁcant when infants exposed to TCAs were compared with matched unexposed infants. The authors concluded that the observed effects are speciﬁc to SRI exposure rather than underlying maternal depression. The Expert Panel noted that the percent of preterm birth (<37 weeks) was comparable in the SRI- and TCA-exposed groups (10.3 and 10.0%, respectively). The absence of information on the full distribution of gestational age in each of the groups precludes a more meaningful interpretation of these data.
In a review of the Swedish Medical Birth Registry, Ericson et al. (102) found that use of antidepressants (ﬂuoxetine [n = 15], other SRIs [n = 531], and non-SRI antidepressants [n = 438]) was associated with premature delivery (adjusted OR: 1.43, 95%; CI 1.14, 1.80), but no associations were noted by class of antidepressants; therefore, the authors suggested that the disease rather than the treatment was associated with prematurity. The Expert Panel believes that an alternative explanation for these ﬁndings is that a similar mode of action among these medications may be affecting the rate of prematurity.
The additional studies of prenatal ﬂuoxetine toxicity were found to be limited by small numbers of subjects and/or study design. Though these studies alone were of limited utility, the Panel found some of the studies useful when evaluated together as a group, especially with better quality studies. No increase in major congenital malformations following in utero exposure was identiﬁed in any of the other studies reviewed by the Panel (19, 90, 96, 101, 104), although the Panel found the methods of most of these studies to be inadequate (i.e., limited sample size and statistical power for hypothesis testing) to detect a potentially important risk for congenital anomalies.
In a review of medical records, Cohen et al. (103) found that neonatal complications requiring admission to a special care nursery occurred in 1 of 11 (9%) pregnancies with ﬁrst or second trimester exposure to ﬂuoxetine vs. 16 of 53 (30%) pregnancies with third trimester exposure. Laine et al. (18) evaluated
tonergic symptoms (e.g., myoclonus, restlessness, tremor, shivering, hyperreﬂexia, incoordination, and rigidity) in infants from ten ﬂuoxetine- and ten citalopram-exposed pregnancies in which exposures were implied to have occurred during the period just prior to delivery. Compared to a control group (n = 20), there was no difference in pregnancy duration, but serotonergic symptoms were greater in the SRI group during the ﬁrst 4 days of life; there was no difference at 2 weeks of age. The study by Oberlander et al. (110) found that 2-day-old infants exposed to SRIs (ﬂuoxetine: n = 7, paroxetine: n = 11, sertraline:
n = 4) in utero responded to pain with less facial activity, a slower heart rate, and greater maintenance of cardiac modulation compared to control infants. It is not known if effects were due to prenatal brain alterations or continued presence of ﬂuoxetine in infant blood. Findings from another study on a broader range of prenatal SRI exposures were consistent with these neonatal ﬁndings (112).
In contrast to the Chambers et al. study, Goldstein et al. (100) reported a lower incidence of neonatal irritability or jitteriness (4.5%) compared to other investigators and stated there were no apparent patterns of abnormal infant conditions in cases of ﬂuoxetine exposure during the third trimester. The Goldstein study was not considered to be reliable due to methodologic limitations such as unstated but possibly large loss to follow-up and reliance for outcome data on reporters of questionable reliability.
Other neurotoxicity studies, limited by study design and found to be of little utility, suggested that prenatal ﬂuoxetine exposure had no effect on neurodevelopment in children up 12 months (19) and 71 months of age (107, 109).
Although no differences were found by Simon et al. (104) in seizure disorder, motor delay, speech delay, or other motor abnormalities when comparing children with pregnancy exposure to ﬂuoxetine and an unexposed reference group, the data collected to demonstrate this lack of developmental deﬁcits were considered inadequate by the Expert Panel. No difference was found in birth weight or growth during the ﬁrst 12 months for infants born to 11 women treated with 20 – 40 mg ﬂuoxetine during pregnancy as compared to 10 control women (19), but the small size of this study provided limited power to detect differences.
220.127.116.11 Breast milk exposures
Symptoms in infants breastfed by mothers taking ﬂuoxetine were reported in case studies as outlined in Section 3.1.2. While some studies reported symptoms similar to those reported with prenatal exposure (e.g., hyperactivity, crying, irritability, reduced sleep, and poor feeding), no symptoms were reported in other breastfed infants exposed to ﬂuoxetine. There were no controlled studies designed to evaluate symptoms in infants exposed to ﬂuoxetine through breast milk.
In a well designed retrospective-cohort study conducted by Chambers et al. (113), it was found that infants nursed by mothers taking ﬂuoxetine (n = 26) had a 392 g deﬁcit in body weight gain (95% CI:
5 – 780 g), with weight gain at ~1.2 SD below control group (n = 38) values from 2 weeks to 6 months of age. However, all of the postnatally exposed infants had been prenatally exposed to ﬂuoxetine in the third trimester, as contrasted to only 10.5% of the controls. Thus, the growth deﬁcits found in this study may have been due to prenatal exposure or the effects attributed to postnatal exposure may have been partly due to residual levels of ﬂuoxetine/norﬂuoxetine from third trimester exposure. A major strength of this study is that unlike most other ﬂuoxetine studies which lack an unmedicated depressed comparison group, this study provides the only evidence of infant deﬁcits speciﬁcally related to ﬂuoxetine and not the underlying depressive disorder.
18.104.22.168 Childhood exposures
Side effects in children taking ﬂuoxetine are similar to those of adults and include manic reaction, hyperkinesia, rash, personality disorder, agitation, constipation, diarrhea, headache, nervousness, somnolence, insomnia, suicide attempt, depression, endometrial hyperplasia, hostility, euphoria, and migraine (4, 116). Additional side effects reported for children include thirst, hyperkinesia, epistaxis, urinary frequency, and menorrhagia (4). Some reviews expressed concern that children may be par-ticularly sensitive to excessive arousal and irritability (117-119). However, other studies did not ﬁnd activation to be the most common side effect, as more commonly observed effects included gastroin-testinal effects, drowsiness, and headache (120, 122, 123).
In a medical review, the FDA (116) expressed concern about prolonged QTc interval and growth decrements in children taking ﬂuoxetine. Although prolonged QTc interval was not replicated in later studies, the signiﬁcance was found to be robust using Fridericia correction, which is unlikely to result in statistical signiﬁcance for random variability. In addition the R-isomer of ﬂuoxetine prolonged QTc intervals in adults. Therefore, the FDA Medical Ofﬁcer believed prolonged QTc interval to be a true drug effect in children.
FDA concerns about growth decrements in children were based upon a 19-week study that reported height and weight increases of 1.0 cm and 1.2 kg in children treated with ﬂuoxetine vs. 2.0 cm and 2.3 kg in control children (P = 0.008) (116). The original study examining growth in children was not available to CERHR. Impaired growth was reported in abstracts but there are no known published studies examining this endpoint.
In October, 2003, the FDA issued a Public Health Advisory regarding their review of suicidality in children taking ﬂuoxetine or seven other antidepressant drugs (136). It was concluded that preliminary data suggest an excess of reports of suicidal ideation and suicide attempts and that additional data, analysis, and public discussion of available data on this issue are needed. In their latest Public Health Advisory, the FDA stated that the contribution of antidepressants to suicidal thinking and behavior is not yet clear, and cautioned clinicians, patients, families, and caregivers to closely monitor children or adults receiving ﬂuoxetine or other antidepressants for worsening of depression or suicidal thoughts, especially during initiation of therapy and following dose adjustments (137). Manufacturers were asked to update their labels with stronger cautions and warnings about the need for monitoring of symptoms.
The Expert Panel ﬁnds the literature on childhood exposures to be markedly deﬁcient due to small sample sizes, inadequate follow-up ranging from 6 to 13 weeks, high attrition, and multiple diagnoses.
Therefore, it is not possible to reach a conclusion regarding possible differences between ﬂuoxetine and control treatments in the context of underlying methodologic limitations.