(32) Identification and characterization of migraine around pregnancy in population-based registry data

Migraine is a primary headache affecting up to 40% of women at childbearing age and is influenced by hormonal changes, including pregnancy. The intermittent use and large variety of pharmacological treatment are challenging for drug safety studies in pregnancy. We aimed to develop algorithms that identify i) women with migraine before and during pregnancy and ii) characterize migraine type and severity, allowing estimating confounding by indication.

Material and Methods

We linked population-based data from the Medical Birth Registry, the National Patient Registry, the Norway Control and Payment of Health Reimbursement (KUHR) and the Norwegian Prescription database for the years 2009-2019.

Of 370,698 women at childbearing age (15-49 years), we included 541,775 pregnancies with one-year lookback time available, and 301,002 pregnancies with five-year lookback before the last menstrual period (LMP) date.

We used different algorithms to identify women with migraine either before or during pregnancy.  Algorithm A1 is based on ICD-10 (G43) or ICPC (N89) diagnostic codes alone, in primary and secondary care; A2 is based on filled triptan (ATC code N02C) prescriptions; and A3 combines A1 and A2. In migraineurs identified through A3, we assessed type of migraine according to the specific diagnoses and migraine severity based on type, route of administration and timing of acute and prophylactic migraine drug dispensation. For the time before pregnancy, three severity groups are possible (mild, moderate, severe) whereas during pregnancy, four are possible (mild, moderate, severe, very severe). We allocated women without any filled prescription but a diagnostic code in the “mild” category.

Results:

We identified the following numbers of pregnancies with migraine in the one-year prior to LMP according to the different algorithms: 14,798 (2.7%) according to A1; 17’576 (3.2%) for A2; and 22’467 (4.1%). These proportions increased to 7.4-9.5% when the look-back period was expanded to five years before LMP. The following number of pregnancies were identified according to the different algorithms during pregnancy: 10,841 (2.0%) according to A1; 5734 (1.1%) for A2; and 13,696 (2.5%) for A3. The proportion of pregnancies with migraine with or without aura were stable in the prepregnancy (0.1%) year and during pregnancy (0.1%). Other or unspecified migraine was the most common diagnosis type in the prepregnancy year (2.6%) and during pregnancy (1.9%).

For migraine severity, we identified 4352 (0.8%) mild, 14’919 (2.8%) moderate and 3196 (0.6%) severe cases during the one-year lookback. During pregnancy, we identified 5013 (0.9%) mild, 6477 (1.2%) moderate, 1563 (0.3%) severe and 795(0.1%) very severe cases.

Conclusion:

The algorithms were able to identify only a proportion of migraineurs compared to other studies, only if they had a diagnosis or filled a prescription during the observation period. The longer the observation period, the more were identified. A combination of diagnostic codes and prescription of migraine specific drugs is useful to increase the proportion, but identification of type and severity remains challenging. It seems that moderate migraine seems to decrease during pregnancy. However, this might be due to the fact that women quit therapy in pregnancy and may not reflect true severity.