A Systematic Review

Management of Primary Headaches During Pregnancy, Postpartum, and Breastfeeding

Ian J. Saldanha MBBS, MPH, PhD; Wangnan Cao PhD; Monika Reddy Bhuma BDS, MPH; Kristin J. Konnyu PhD; Gaelen P. Adam MLIS, MPH; Shivani Mehta BA; Andrew R. Zullo PharmD, PhD; Kenneth K. Chen MD; Julie L. Roth MD; Ethan M. Balk MD, MPH

Disclosures

Headache. 2021;61(1):11-43. 

In This Article

Results

Our search for primary studies (direct evidence; Figure 1) returned 8549 citations and our search for SRs (indirect evidence; Figure 2) returned 2788 citations. For direct evidence, we screened 400 full-text articles. The most frequent reasons for excluding full-text articles were that study participants did not have primary headache or the study did not report primary headache-specific data (68 articles) and that participants were not pregnant, attempting to become pregnant, postpartum, or breastfeeding (50 articles). For indirect evidence, we screened 376 full-text articles. The most frequent reasons for excluding these articles were that they did not meet minimum criteria for SRs (141 articles) and that there were no interventions of interest (50 articles). We ultimately included 16 primary studies and 26 SRs.

Figure 1.

PRISMA diagram depicting identification of primary studies in this systematic review

Figure 2.

PRISMA diagram depicting identification of systematic reviews (indirect evidence)

Characteristics of Included Evidence

Table 1 describes the 16 included primary studies (direct evidence) as well as their intervention details and participant baseline characteristics. Published between 1990 and 2018, the studies comprised three RCTs (reported in five articles[12–16]), eight NRCSs (reported in 16 articles[17–32]), and five single-group studies (reported in six articles[16,33–37]).

The studies included a total of 14,185 women, all of whom were pregnant. No studies included women attempting to become pregnant or who were postpartum or breastfeeding. Pregnancy trimesters and gestational ages varied across studies. When reported, average patient ages across the 16 studies ranged from 23 to 34 years. Among the only four studies that reported data for average gravidity and parity, gravidity was usually two or three and parity was usually one (Table 1).[12–15,34,36] Only one study, an RCT, reported on the racial distribution of the patients, 76% of whom were black.[12–14] Seven studies reported maternal benefit outcomes, six reported maternal adverse effects, and 11 reported fetal/child adverse effects.

Appendix D describes the 26 included SRs (indirect evidence), including number of databases searched, years searches were run, number of studies included, interventions assessed, and outcomes reported. Published between 2000 and 2020, the SRs included a total of 740 studies (not accounting for some overlap across SRs), with a median of 15 studies per SR (interquartile range 10–37). All 26 SRs assessed pharmacologic interventions: eight assessed NSAIDs,[38–45] two antiepileptics,[46–48] two β-blockers,[49–51] two calcium channel blockers,[50,52] two antiemetics (5-HT3 receptor antagonists),[53,54] two antipsychotics,[55,56] two antihistamines,[57,58] and one each assessed serotonin and norepinephrine reuptake inhibitors (SNRIs),[59,60] tricyclic antidepressants,[59,60] benzodiazepines,[61,62] corticosteroids,[63] oral magnesium,[64] triptans (5-HT1B/1D receptor agonists),[65] analgesics/antipyretics,[66] and intravenous magnesium.[67] Twelve SRs reported maternal adverse effects and 23 reported fetal/child adverse effects.

Risk of Bias

We assessed all three RCTs to be at overall high RoB because of methodological concerns, such as lack of blinding of participants, personnel, and outcome assessors, and likely selective outcome reporting (Appendix E). We assessed seven of the eight NRCSs as overall high RoB mostly because of serious risk of confounding due to inadequate adjustment for differences between groups being compared. We assessed four of the five single-group studies as overall low RoB.

Table 2 provides an overview of the direct and indirect evidence (and the gaps) identified in this review. Table 3 provides more detailed summaries of findings. Table 4 summarizes statistically significant findings of harms from SRs (indirect evidence).

Key Question 1: Prevention of Primary Headache

Pharmacologic Interventions for Prevention. Benefits of Pharmacologic Interventions for Prevention: We did not identify any evidence for the beneficial effects of pharmacologic interventions to prevent primary headache in women who are pregnant, attempting to become pregnant, postpartum, or breastfeeding.

Harms of Pharmacologic Interventions for Prevention: One primary study and 11 SRs provided evidence regarding harms of pharmacologic interventions used to prevent primary headache. No studies evaluated non-pharmacologic interventions to prevent primary studies.

Antiepileptics: The direct evidence (one single-group study: Castilla-Puentes 2014[33]) was insufficient to make conclusions about the harms of topiramate when used for prevention of migraine. Among the 81 infants exposed to topiramate during pregnancy, 10 infants (12.3%) developed congenital anomalies. Two infants (2.5%) had cleft palate. The following anomalies were found in one infant (1.2%) each: hydrocephalus, meningomyelocele, spina bifida, an unspecified cardiovascular congenital anomaly, syndactyly, polydactyly, gastrointestinal obstruction, and pyloric stenosis.

There is indirect evidence (two SRs: Veroniki 2017[46,47] and Weston 2016[48]) that antiepileptics are associated with maternal or fetal/child adverse effects. Topiramate was associated with fetal death or spontaneous abortion (combined), fetal growth restriction, cleft lip/palate, and other major congenital anomalies (Moderate SoE). Carbamazepine was associated with both major and minor congenital anomalies (Moderate SoE). Gabapentin was associated with congenital cardiovascular anomalies, hypospadias, and psychomotor developmental delay (Low SoE). Lamotrigine was associated with autism/dyspraxia, but not other adverse effects (Moderate SoE). Valproate was associated with fetal death or spontaneous abortion, major congenital malformations, cleft lip/palate, developmental delays, and autism/dyspraxia (Moderate SoE).

Venlafaxine. There is indirect evidence (one SR: McDonagh 2014[59,60]) that venlafaxine, an SNRI, was associated with preterm birth (Moderate SoE).

Tricyclic Antidepressants: There is indirect evidence (one SR: McDonagh 2014[59,60]) that use of any tricyclic antidepressant was associated with major congenital anomalies, cardiovascular anomalies, small for gestational age, neonatal convulsions, and neonatal respiratory distress, but not low birth weight (Moderate SoE).

Benzodiazepines: There is indirect evidence (one SR: Enato 2011[61,62]) that use of any benzodiazepine was associated with oral cleft and other major congenital anomalies (Low SoE).

β-Blockers: There is indirect evidence (two SRs: Yakoob 2013[49] and Abalos 2018[50]) that use of any β-blocker was associated with cardiovascular anomalies, cleft lip/palate, and neural tube defects, but not preterm birth (Moderate SoE).

Calcium Channel Blockers: There is indirect evidence (one SR: Abalos 2018[50]) that use of any calcium channel blocker was not associated with maternal (Low SoE) or fetal/child adverse effects (Low to Moderate SoE). There is also indirect evidence (one SR: Bellos 2020a[52]) that nifedipine was not associated with fetal/child adverse effects (Low to Moderate SoE).

Prednisolone: There is indirect evidence (one SR: Park-Wyllie 2000) that prednisolone was associated with oral clefts, but not other major congenital anomalies (Low SoE).

Antihistamines: There is indirect evidence (two SRs: Etwel 2017[57] and Li 2019[58]) that use of any antihistamine was not associated with spontaneous abortion, stillbirth, preterm birth, low birth weight, or major congenital anomalies (Moderate SoE).

Oral Magnesium: There is indirect evidence (one SR: Makrides 2014[64]) that oral magnesium was associated with neonatal death, but not low birth weight (Low SoE). Its use was not associated with maternal adverse effects (Low SoE).

Non-pharmacologic Interventions for Prevention. No primary study or SR addressed benefits or harms of non-pharmacologic interventions for prevention of primary headache.

Key Question 2: Treatment of Primary Headache

Among the 15 primary studies assessing treatments, 11 enrolled patients with migraine,[17–32,34,36,37] one enrolled patients with tension headache,[15] and three enrolled patients with either migraine or tension headache.[12–14,16] Nine studies assessed pharmacologic treatments: one RCT of antiemetics, antihistamines, and opioid-containing analgesics[12–14] and eight NRCSs of triptans, ergot products, and NSAIDs.[17–32] Six studies assessed non-pharmacologic treatments: two RCTs[15,16] and two single-group studies[16,35,36] of complementary, behavioral, and physical therapies, one single-group study of nerve blocks,[34] and one single-group study of noninvasive neuromodulation devices.[37]

All 18 SRs assessing treatment of primary headache focused on pharmacologic interventions; no SRs assessed non-pharmacologic interventions (Table 2, Table 3 and Table 4).

Pharmacologic Interventions for Treatment. Benefits of Pharmacologic Interventions for Treatment: Antiemetics, Antihistamines, and Opioid Analgesics: We found one RCT (Childress 2018;[12–14] direct evidence) that randomized 70 pregnant women with either migraine or tension headache to a single dose of an intravenous combination of metoclopramide (a dopamine receptor antagonist antiemetic) 10 mg and diphenhydramine 25 mg or to a single dose of oral codeine 30 mg. Compared with patients in the codeine arm, patients in the combination treatment arm experienced greater reductions in severity of acute headache attacks over 24 hours, measured using a visual analog scale from 0 to 10 at 30 minutes after the dose (net mean difference [NMD] −3.0, 95% CI −4.2 to −1.8), at 1 hour (NMD −2.1, 95% CI −3.3 to −0.9), and at 12 hours (NMD −1.6, 95% CI −2.9 to −0.3), but not at 6 or 24 hours. Patients in the combination treatment arm were also more likely to experience relief from headache with one dose (odds ratio [OR] 1.37, 95% CI 1.07 to 1.75) and to experience complete resolution of headache at 24 hours (OR 5.42, 95% CI 1.86 to 15.8). Combination treatment also provided relief from headache 42.2 minutes sooner (95% CI 20.7 to 63.7) than codeine treatment.

Harms of Pharmacologic Interventions for Treatment: Antiemetics, Antihistamines, and Opioid Analgesics: The Childress 2018 RCT (direct evidence) reported that no serious maternal adverse effects occurred within 24 hours in either the combination metoclopramide 10 mg and diphenhydramine 25 mg arm or the oral codeine 30 mg arm.[12–14]

Triptans, Ergot Products, Naproxen, and Pizotifen: Eight primary studies (direct evidence), all observational NRCSs (described in 16 articles[17–32]) reported the harms of these interventions (benefits were not reported). These studies, which included three prospective cohort studies[17–21] and five retrospective cohort studies,[22–32] enrolled a total of 13,907 patients with migraine. The eight studies addressed nine different comparisons. There was insufficient evidence to make conclusions regarding seven of the comparisons described in four studies (Ephross 2014,[17–19] O'Quinn 1999,[20] Kallen 2011,[28,29] and Olesen 2000[30]): sumatriptan versus naratriptan during pregnancy, sumatriptan versus combination sumatriptan and naproxen during pregnancy, naratriptan versus combination sumatriptan and naproxen during pregnancy, any triptan versus any ergot product during pregnancy, any triptan versus pizotifen during pregnancy, any ergot product versus pizotifen during pregnancy, and sumatriptan during pregnancy versus sumatriptan before pregnancy only.

We were able to make conclusions regarding only two comparisons. First, triptan use during pregnancy, when compared with triptan use only before pregnancy, may have a low risk of adverse effects, except for increased child emotionality (adjusted relative risk [RR] 2.18, 95% CI 1.03–4.53) and hyperactivity at 3 years of age (adjusted RR 1.70, 95% CI 1.02–2.80) (two NRCSs of 8460 patients: Nezvalova-Henriksen 2013[22] and Nezvalova-Henriksen 2010;[23–27] Low SoE). Second, triptan use during pregnancy, when compared with no triptan use either during or before pregnancy, may have a low risk of adverse effects, except for increased child emotionality (adjusted RR 2.51, 95% CI 1.27–4.90) and hyperactivity at 3 years of age (adjusted RR 1.57, 95% CI 1.04–2.36) (three NRCSs of 6999 patients: Shuhaiber 1997,[21] Nezvalova-Henriksen 2010,[23–27] and Spielmann 2018;[31,32] Low SoE).

NSAIDs: There is indirect evidence (one SR: Hammers 2015[39]) that indomethacin was associated with neonatal periventricular leukomalacia, intraventricular hemorrhage, and necrotizing enterocolitis (Low SoE) (Table 4). There is indirect evidence (six SRs[40–45]) that low-dose aspirin was not associated with maternal (Moderate SoE) or fetal/child adverse effects (Low SoE).

Ondansetron: There is indirect evidence (two SRs: Kaplan 2019[53] and Picot 2020[54]) that ondansetron (a 5-HT3 receptor antagonist antiemetic) was associated with cardiovascular anomalies, orofacial clefts, diaphragmatic hernia, and respiratory system anomalies (Moderate SoE).

Antipsychotics: There is indirect evidence (two SRs: Coughlin 2015[55] and Terrana 2015[56]) that use of any antipsychotic was associated with preterm birth, low birth weight, and congenital anomalies (Low to Moderate SoE).

Analgesics/Antipyretics: There is indirect evidence (one SR: Masarwa 2018[66]) that acetaminophen was associated with attention deficit hyperactivity disorder, hyperactivity symptoms, autism spectrum disorder, and conduct disorder.

Intravenous Magnesium: There is indirect evidence (one SR: Bain 2014[67]) that intravenous magnesium was associated with maternal respiratory depression/other respiratory problems, hypotension, and tachycardia.

Corticosteroids and Antihistamines: See section Harms of Pharmacologic Interventions for Prevention.

Non-pharmacologic Interventions for Treatment. Six primary studies (direct evidence) comprising 108 patients, total, but no SRs (indirect evidence) addressed non-pharmacologic treatments for primary headache (Silva 2012,[15] Marcus [two studies] 1995,[16,35] Govindappagari 2014,[34] Hickling 1990,[36] and Bhola 2015[37]). The studies provided insufficient evidence to make conclusions about either the benefits or harms of acupuncture use versus non-use; combination thermal biofeedback, relaxation therapy, and physical therapy; combination thermal biofeedback and relaxation therapy; peripheral nerve blocks; and transcranial magnetic stimulation.

Quantitative Synthesis (for Both Prevention and Treatment)

Most of the evidence identified in this review was indirect, that is, in women with various health conditions. Moreover, as detailed in Table 3, where we identified direct evidence (i.e., in women with primary headache), we were limited by insufficient SoE and/or large inconsistencies in types of primary headache addressed or intervention comparisons evaluated. These factors led to sparseness that precluded meta-analysis.

processing....