Pregnancy Outcome Following Prenatal Exposure to Triptans
We followed the recommendations of the Meta-analysis Of Observational Studies in Epidemiology Group. After developing the inclusion criteria, 2 authors searched the literature and, independently in selected studies, extracted data and evaluated the methodology of the studies. With the assistance of a trained research librarian, we searched electronic databases for potentially pertinent articles that were published in any language from 1991 (when the first triptan medication was introduced) until December 2013. These databases included Medical Literature Analysis and Retrieval System Online (via Object, View and Interaction Design [OVID]), Excerpta Medica Database, SCOPUS, Toxicology Information Online Special (via Toxicology Data Network), DART: Developmental and Reproductive Toxicology, ReproTox, Teratogen Information System, OVID International Pharmaceutical Abstracts, Cumulative Index to Nursing and Allied Health Literature, Shepard's Citations, Google Scholar, Cochrane Library, World Cat, Digital Dissertations, Global Health, Institute for Scientific Information Proceedings, and Biosciences Information Service Previews.
We started by using the chemical identification database to identify synonyms of different triptans. In addition to searching the literature by drug names, both scientific and generic (sumatriptan [Imitrex, Imigran, Cinie, Illument, Migriptan], rizatriptan [Maxalt], naratriptan [Amerge, Naramig], zolmitriptan [Zomig], eletriptan [Relpax], almotriptan [Axert, Almogran], frovatriptan [Frova, Migard, Frovamig], and avitriptan [BMS-180,048], etc), we searched by drug categories based on pharmacological action (antimigraine agents, antianalgesics, serotonin agonists, serotonin 5-HT1B and 5-HT1D, headache treatment, headache relief). These terms were then combined with various Medical Subject Headings categories (including pregnancy; pregnancy complications; abnormalities; embryonic and fetal development; maternal exposure; teratogens; and congenital, hereditary, and neonatal diseases and abnormalities) and keywords (including birth defects; abnormalities; congenital, hereditary, and neonatal diseases and abnormalities; malformations, embryonic, and fetal developments; embryonic structures; teratogens; prenatal pregnancies; deformities; and embryopathies). The Web of Science search included: Science Citation Index Expanded, Social Sciences Citation Index, Conference Proceedings Citation Index – Science, and Conference Proceedings Citation Index – Social Science & Humanities.
To find additional sources of published and unpublished data, we examined the bibliographies of the retrieved articles.
Figure 1 illustrates our strategy for the study selection process.
(Enlarge Image)
Figure 1.
Diagram for search strategy and study selection for the meta-analyses.
Studies had to meet the following inclusion criteria: (1) they had to be randomized control trials, case–control or observational cohort studies; (2) they had to report the incidence of MCMs, prematurity and/or spontaneous abortions in women who had migraines and were exposed to triptan medications during at least the first trimester of pregnancy (exposed group); (3) the study must have had at least one of the following comparison groups: women who had migraine and were not exposed to triptans during pregnancy (migraine – no triptans, disease control group) and/or healthy women who did not have migraines and were not exposed to triptans during pregnancy (healthy control). Prematurity was estimated by analyzing the duration of pregnancy to delivery time.
Three investigators reviewed the studies to determine whether they met inclusion criteria, and a standardized data collection form was used to extract information concerning each study. This information included the article title, name of the first author, year of publication, study design, study location, characteristics and source of study population, sample size, outcome measures and methods for data collection, exposure measurement, ascertaining outcomes, handling loss to follow-up, and controlling for confounding factors. It also included the number of cases for prematurity, spontaneous abortions, and fetal MCMs associated with exposure to sumatriptan or other medications from the triptan group. If the number of births was not stated explicitly in published reports, we obtained this information from communication with the authors. We used the Newcastle–Ottawa Scale to assess the quality of the methodology used in the case–control and cohort studies. To rate the quality of the methods of selecting groups (cases and controls or exposed and nonexposed cohorts), we used a scale from 0 to 9, with 0 indicating the lowest quality. To evaluate the case and controlled subjects, we used a scale from 0 to 4. To rate the comparability of groups, we used a scale from 0 to 2 and awarded a study 1 point if it controlled for maternal age and 2 points if it controlled for maternal age and any additional factor. To rate the methods for ascertaining exposure or ascertaining outcome, we used a scale from 0 to 3, with 0 again indicating the lowest quality. We considered cohorts to have adequate follow-up if fewer than 10% of participants were lost to follow-up and if the loss occurred in a way that was unlikely to introduce bias. Studies had to achieve a quality rating of 6 or greater to be included in this meta-analysis.
We conducted several meta-analyses for pregnancy outcomes by comparing the results of the triptan-exposed group with each of 2 control groups, and also comparing the results of the group exposed to migraine no-triptans to healthy controls.
From each study, we extracted data unadjusted for potential confounders, arranged it in a 2 × 2 table, and calculated the odds ratio (OR) and 95% confidence intervals (CIs) for dichotomous outcomes (the rate of spontaneous abortion, prematurity, and/or MCMs) following exposure to triptans. A random-effects meta-analysis model was used to combine the rate data for malformation and other pregnancy outcomes by using the Comprehensive Meta Analysis Version 2.0. (Biostat, Englewood, NJ, USA). We included more than 2 studies and used random-effects models to reduce the high statistical heterogeneity. Heterogeneity of effects was assessed utilizing the Q statistic. In addition, we estimated heterogeneity using I-square statistic, which estimates the proportion of total variance that is attributable to between-study variance. An associated I value between 25–50% signifies low heterogeneity, between 50–75% signifies moderate heterogeneity, and >75% signifies high heterogeneity. We were unable to assess for publication bias utilizing the funnel plot due to the small number of available studies contributing to each outcome (less than 10). The test has low power, which makes it difficult to detect asymmetry with good accuracy.
Methods
Data Sources
We followed the recommendations of the Meta-analysis Of Observational Studies in Epidemiology Group. After developing the inclusion criteria, 2 authors searched the literature and, independently in selected studies, extracted data and evaluated the methodology of the studies. With the assistance of a trained research librarian, we searched electronic databases for potentially pertinent articles that were published in any language from 1991 (when the first triptan medication was introduced) until December 2013. These databases included Medical Literature Analysis and Retrieval System Online (via Object, View and Interaction Design [OVID]), Excerpta Medica Database, SCOPUS, Toxicology Information Online Special (via Toxicology Data Network), DART: Developmental and Reproductive Toxicology, ReproTox, Teratogen Information System, OVID International Pharmaceutical Abstracts, Cumulative Index to Nursing and Allied Health Literature, Shepard's Citations, Google Scholar, Cochrane Library, World Cat, Digital Dissertations, Global Health, Institute for Scientific Information Proceedings, and Biosciences Information Service Previews.
We started by using the chemical identification database to identify synonyms of different triptans. In addition to searching the literature by drug names, both scientific and generic (sumatriptan [Imitrex, Imigran, Cinie, Illument, Migriptan], rizatriptan [Maxalt], naratriptan [Amerge, Naramig], zolmitriptan [Zomig], eletriptan [Relpax], almotriptan [Axert, Almogran], frovatriptan [Frova, Migard, Frovamig], and avitriptan [BMS-180,048], etc), we searched by drug categories based on pharmacological action (antimigraine agents, antianalgesics, serotonin agonists, serotonin 5-HT1B and 5-HT1D, headache treatment, headache relief). These terms were then combined with various Medical Subject Headings categories (including pregnancy; pregnancy complications; abnormalities; embryonic and fetal development; maternal exposure; teratogens; and congenital, hereditary, and neonatal diseases and abnormalities) and keywords (including birth defects; abnormalities; congenital, hereditary, and neonatal diseases and abnormalities; malformations, embryonic, and fetal developments; embryonic structures; teratogens; prenatal pregnancies; deformities; and embryopathies). The Web of Science search included: Science Citation Index Expanded, Social Sciences Citation Index, Conference Proceedings Citation Index – Science, and Conference Proceedings Citation Index – Social Science & Humanities.
To find additional sources of published and unpublished data, we examined the bibliographies of the retrieved articles.
Study Selection
Figure 1 illustrates our strategy for the study selection process.
(Enlarge Image)
Figure 1.
Diagram for search strategy and study selection for the meta-analyses.
Studies had to meet the following inclusion criteria: (1) they had to be randomized control trials, case–control or observational cohort studies; (2) they had to report the incidence of MCMs, prematurity and/or spontaneous abortions in women who had migraines and were exposed to triptan medications during at least the first trimester of pregnancy (exposed group); (3) the study must have had at least one of the following comparison groups: women who had migraine and were not exposed to triptans during pregnancy (migraine – no triptans, disease control group) and/or healthy women who did not have migraines and were not exposed to triptans during pregnancy (healthy control). Prematurity was estimated by analyzing the duration of pregnancy to delivery time.
Three investigators reviewed the studies to determine whether they met inclusion criteria, and a standardized data collection form was used to extract information concerning each study. This information included the article title, name of the first author, year of publication, study design, study location, characteristics and source of study population, sample size, outcome measures and methods for data collection, exposure measurement, ascertaining outcomes, handling loss to follow-up, and controlling for confounding factors. It also included the number of cases for prematurity, spontaneous abortions, and fetal MCMs associated with exposure to sumatriptan or other medications from the triptan group. If the number of births was not stated explicitly in published reports, we obtained this information from communication with the authors. We used the Newcastle–Ottawa Scale to assess the quality of the methodology used in the case–control and cohort studies. To rate the quality of the methods of selecting groups (cases and controls or exposed and nonexposed cohorts), we used a scale from 0 to 9, with 0 indicating the lowest quality. To evaluate the case and controlled subjects, we used a scale from 0 to 4. To rate the comparability of groups, we used a scale from 0 to 2 and awarded a study 1 point if it controlled for maternal age and 2 points if it controlled for maternal age and any additional factor. To rate the methods for ascertaining exposure or ascertaining outcome, we used a scale from 0 to 3, with 0 again indicating the lowest quality. We considered cohorts to have adequate follow-up if fewer than 10% of participants were lost to follow-up and if the loss occurred in a way that was unlikely to introduce bias. Studies had to achieve a quality rating of 6 or greater to be included in this meta-analysis.
We conducted several meta-analyses for pregnancy outcomes by comparing the results of the triptan-exposed group with each of 2 control groups, and also comparing the results of the group exposed to migraine no-triptans to healthy controls.
Meta-analytic Method
From each study, we extracted data unadjusted for potential confounders, arranged it in a 2 × 2 table, and calculated the odds ratio (OR) and 95% confidence intervals (CIs) for dichotomous outcomes (the rate of spontaneous abortion, prematurity, and/or MCMs) following exposure to triptans. A random-effects meta-analysis model was used to combine the rate data for malformation and other pregnancy outcomes by using the Comprehensive Meta Analysis Version 2.0. (Biostat, Englewood, NJ, USA). We included more than 2 studies and used random-effects models to reduce the high statistical heterogeneity. Heterogeneity of effects was assessed utilizing the Q statistic. In addition, we estimated heterogeneity using I-square statistic, which estimates the proportion of total variance that is attributable to between-study variance. An associated I value between 25–50% signifies low heterogeneity, between 50–75% signifies moderate heterogeneity, and >75% signifies high heterogeneity. We were unable to assess for publication bias utilizing the funnel plot due to the small number of available studies contributing to each outcome (less than 10). The test has low power, which makes it difficult to detect asymmetry with good accuracy.