Topiramate During Pregnancy and Risk of Birth Defects
In this large, retrospective analysis of pharmacy and medical claims data, no significant differences were seen between topiramate and comparison drug- or disease-based cohorts in the frequencies of OC or MCM occurrences. RR analysis indicated that the upper bound of RR ranged from 3.52 to 6.85 for OCs and from 1.41 to 1.90 for MCMs (Table 2 and Table 3). The relatively small number of events may limit the strength of inferences that can be drawn from these results. For MCMs, a significantly lower frequency of malformations was seen relative to the diabetes cohort (Table 3).
The frequency of OC events in the random sample comparator group (0.16%) in the current study is similar to that reported for the general population (0.17%) by the Centers for Disease Control and Prevention from 2004 to 2006. Analysis of MCM occurrence in the random sample comparator cohort revealed a higher rate for MCMs (3.77%) in this cohort than is expected in a healthy general population (1.00–2.30%). As a medical database population with medical and pharmacy claims, the random sample population in this study may be more generalizable to the database than to the general population. Studies evaluating the correspondence of claims data with malformations have reported that the positive predictive value for OCs is approximately 90%, which is consistent with the closeness of the frequency of OC events seen in this study and of that reported in the literature. By contrast, the positive predictive value for MCMs is approximately 60%, so this study yields a modest overestimation in the frequency of MCM events.
Claims databases, such as WKPS, allow for assessments across large populations but have some limitations, including the dependence on ICD-9 coding, which may introduce bias because of diagnoses that were not coded or coded incorrectly at the site of care. However, the positive aspects of claims databases are that they lack enrollment and recall bias as might occur with a registry.
Topiramate is indicated for use as monotherapy and adjunctive therapy for epilepsy and as preventive therapy in migraine headache. In this study, there was no increased risk of OC or MCM events for mothers exposed to topiramate as compared with mothers with these disease states. Although the topiramate cohort also contained women with migraine and epilepsy diagnoses (Appendix B), the numbers of these events were too limited to have had a significant effect on these outcomes.
Appendix C
Appendix D
Other studies have indicated that maternal diabetes and obesity during pregnancy may increase the risk of congenital malformations in infants. In this study, compared with the topiramate cohort, the rate of OC and MCM occurrence was higher in infants born to mothers with diabetes (Table 2 and Table 3). In addition, the RR of MCM events was significantly lower for infants exposed to topiramate compared with those born to mothers with diabetes (RR = 0.65 [95% CI: 0.47–0.89]). A further analysis of the impact of BMI (a risk factor for diabetes) on OC and MCM outcomes in the random comparator sample showed a corresponding increase in the rate of MCM events with increasing BMI (Fig. 2). The ability to run statistical analyses was limited by the small number of events.
Maternal comorbidities, such as tobacco use, can influence pregnancy outcomes and may increase the risk of OC and MCM occurrence. The current study, based on insurance claims, did not fully ascertain data on lifestyle factors that could affect pregnancy outcomes such as tobacco use or the use of folic acid, and this incomplete ascertainment (a limitation of insurance claims-based analyses) may be different across cohorts; thus, each of the drug-exposed and disease-state cohorts may have spuriously higher prevalences than the random sample comparator group.
However, the incomplete ascertainment of BMI through claims associated with different levels of obesity suggested an expected association with claims for malformations. In addition, this study was hampered by the inability to validate MCMs and medication start/stop times. Medical databases are susceptible to epidemiologic biases, such as selection bias (in this study, the selection of excluded teratogens and impact of exclusions on the random sample) and information bias, as a result of incomplete documentation of clinical details.
Typically, there is a lack of a dose response between maternal AED use and congenital malformations. In this study, a dose–response analysis showed that the frequency of MCM events was highest (6.13%) in the cohort receiving the lowest daily topiramate dose (≤50 mg) during the first trimester (Fig. 3). Because the data for this study were derived from commercially insured women, the doses of topiramate (and other AEDs) used may differ from those used among women in other settings, such as those covered by Medicaid. Polytherapy with multiple AEDs were not excluded from the analyses. Because polytherapy can lead to an increased frequency of MCM events, polytherapy use in this population may have affected the analysis and the frequency of MCM events. In addition, any potential relationship between topiramate dose and frequency of MCMs would be difficult to define based on the low numbers of available events.
Even with the recent labeling changes to topiramate, there have been limited data available to clinicians to evaluate the risks associated with topiramate use in a clinical setting. This large, retrospective claims database study can add to data from other registry and epidemiologic studies on the risk of topiramate use in women of child-bearing age, perhaps providing some context to other chronic disease states and assisting clinicians in tailoring therapy to individual patients.
Discussion
In this large, retrospective analysis of pharmacy and medical claims data, no significant differences were seen between topiramate and comparison drug- or disease-based cohorts in the frequencies of OC or MCM occurrences. RR analysis indicated that the upper bound of RR ranged from 3.52 to 6.85 for OCs and from 1.41 to 1.90 for MCMs (Table 2 and Table 3). The relatively small number of events may limit the strength of inferences that can be drawn from these results. For MCMs, a significantly lower frequency of malformations was seen relative to the diabetes cohort (Table 3).
The frequency of OC events in the random sample comparator group (0.16%) in the current study is similar to that reported for the general population (0.17%) by the Centers for Disease Control and Prevention from 2004 to 2006. Analysis of MCM occurrence in the random sample comparator cohort revealed a higher rate for MCMs (3.77%) in this cohort than is expected in a healthy general population (1.00–2.30%). As a medical database population with medical and pharmacy claims, the random sample population in this study may be more generalizable to the database than to the general population. Studies evaluating the correspondence of claims data with malformations have reported that the positive predictive value for OCs is approximately 90%, which is consistent with the closeness of the frequency of OC events seen in this study and of that reported in the literature. By contrast, the positive predictive value for MCMs is approximately 60%, so this study yields a modest overestimation in the frequency of MCM events.
Claims databases, such as WKPS, allow for assessments across large populations but have some limitations, including the dependence on ICD-9 coding, which may introduce bias because of diagnoses that were not coded or coded incorrectly at the site of care. However, the positive aspects of claims databases are that they lack enrollment and recall bias as might occur with a registry.
Topiramate is indicated for use as monotherapy and adjunctive therapy for epilepsy and as preventive therapy in migraine headache. In this study, there was no increased risk of OC or MCM events for mothers exposed to topiramate as compared with mothers with these disease states. Although the topiramate cohort also contained women with migraine and epilepsy diagnoses (Appendix B), the numbers of these events were too limited to have had a significant effect on these outcomes.
Appendix C
Appendix D
Other studies have indicated that maternal diabetes and obesity during pregnancy may increase the risk of congenital malformations in infants. In this study, compared with the topiramate cohort, the rate of OC and MCM occurrence was higher in infants born to mothers with diabetes (Table 2 and Table 3). In addition, the RR of MCM events was significantly lower for infants exposed to topiramate compared with those born to mothers with diabetes (RR = 0.65 [95% CI: 0.47–0.89]). A further analysis of the impact of BMI (a risk factor for diabetes) on OC and MCM outcomes in the random comparator sample showed a corresponding increase in the rate of MCM events with increasing BMI (Fig. 2). The ability to run statistical analyses was limited by the small number of events.
Maternal comorbidities, such as tobacco use, can influence pregnancy outcomes and may increase the risk of OC and MCM occurrence. The current study, based on insurance claims, did not fully ascertain data on lifestyle factors that could affect pregnancy outcomes such as tobacco use or the use of folic acid, and this incomplete ascertainment (a limitation of insurance claims-based analyses) may be different across cohorts; thus, each of the drug-exposed and disease-state cohorts may have spuriously higher prevalences than the random sample comparator group.
However, the incomplete ascertainment of BMI through claims associated with different levels of obesity suggested an expected association with claims for malformations. In addition, this study was hampered by the inability to validate MCMs and medication start/stop times. Medical databases are susceptible to epidemiologic biases, such as selection bias (in this study, the selection of excluded teratogens and impact of exclusions on the random sample) and information bias, as a result of incomplete documentation of clinical details.
Typically, there is a lack of a dose response between maternal AED use and congenital malformations. In this study, a dose–response analysis showed that the frequency of MCM events was highest (6.13%) in the cohort receiving the lowest daily topiramate dose (≤50 mg) during the first trimester (Fig. 3). Because the data for this study were derived from commercially insured women, the doses of topiramate (and other AEDs) used may differ from those used among women in other settings, such as those covered by Medicaid. Polytherapy with multiple AEDs were not excluded from the analyses. Because polytherapy can lead to an increased frequency of MCM events, polytherapy use in this population may have affected the analysis and the frequency of MCM events. In addition, any potential relationship between topiramate dose and frequency of MCMs would be difficult to define based on the low numbers of available events.
Even with the recent labeling changes to topiramate, there have been limited data available to clinicians to evaluate the risks associated with topiramate use in a clinical setting. This large, retrospective claims database study can add to data from other registry and epidemiologic studies on the risk of topiramate use in women of child-bearing age, perhaps providing some context to other chronic disease states and assisting clinicians in tailoring therapy to individual patients.
Source...