Introduction: There has been limited publication of temporal trends in rates of stroke during pregnancy. The HCUP National Inpatient Sample (NIS) is a set of longitudinal databases, originally encompassing all discharges from 20% of the available community non-rehabilitation hospitals each year. In the year 2000, 994 hospitals yielded 7,450,992 unweighted discharge records. Beginning in 2012, the most recent year included in this analysis, the sampling scheme changed to a 20% random sample of discharges from each participating hospital. This resulted in data from 4,378 hospitals, yielding 7,296,968 unweighted discharge records.
Methods: Using pregnancy related discharges in patients greater than 14 years(y) of age from the NIS between 2000 and 2012, we set out to determine if rates of pregnancy related hospital stays which included a diagnosis of stroke, have changed over time. Stroke was defined by ICD-9 codes of 430.xx for subarachnoid hemorrhage (SAH), 431.xx and 432.9x for intracranial hemorrhage (ICH); and 433.x1, 434.x1, and 436.xx for ischemic stroke (IS). Pregnancy was defined by ICD-9 codes of 630.xx to 679.xx (undelivered), and DRG (v18 or v24) between 370 and 375 or ICD9 V27.x (delivered). Appropriate NIS discharge weights were used to estimate national rates and determine variance. Stroke rates are reported per 100,000 pregnancy related discharges. Trends over time were tested.
Results: Rates for stroke increased from 15 (4 ICH, 4 SAH, 7 IS) in 2000, to 26 (6 SAH, 8 ICH, 12 IS) in 2012 (p<.001). Stroke in women who delivered during the hospitalization increased from 4.8 to 6.6 (ns), while in those hospitalized during gestation and discharged undelivered, the stroke rate increased from 103 to 223 (p<.001). The increase was predominately in IS. Rates increased from 12 to 18 for women aged < 25 y; from 16 to 27 for women between 25 and 34 y; and from 28 to 47 for women 35 y and older, over the 12 year period.
Conclusion: The diagnosis of stroke in pregnancy related hospital stays increased by 73% between 2000 and 2012. The reason for this could be changes in coding or diagnoses, or a true increase in stroke incidence, but further study is needed.