Trend of Stroke Hospitalization, United States, 1988–1997

BACKGROUND AND PURPOSE: Age-adjusted stroke mortality in the United States has declined in recent decades. However, the course of stroke incidence is less certain. To address this issue, we determined trends of stroke hospitalization and in-hospital case fatality during 1988-1997. METHODS: Stroke hospitalization was estimated from National Hospital Discharge Survey as numerator and Current Population Survey as denominator. Hospitalization rates were determined and stratified by patient characteristics. Average length of hospital stay was also determined. In-hospital mortality was specified by sex, age, and other patient characteristics. The change in these rates over 10 years and average annual percent changes were calculated. RESULTS: During 1988-1997, age-adjusted stroke hospitalization rate increased 18.6% (from 560 to 664/100 000; P=0.043), while total hospitalization increased from 592 811 to 821 760. This increase was limited to persons aged >/=65 years. Patients in the South had the highest stroke hospitalization rates, and those in the West had the lowest. Overall, 58% of strokes were classified as ischemic, 13% as hemorrhagic, and 29% as other. Over these 10 years, stroke patients having coincident diabetes, hypertension, and congestive heart failure increased 17.4% (P=0.17), 34% (P=0.05), and 31% (P=0.091), respectively. The average length of hospital stay fell from 11.1 to 6.2 days (44.1%; P=0.012). As a result, despite an increase in hospitalizations for stroke, the total person-days in hospital actually decreased by 22% (P=0.06). CONCLUSIONS: The declining age-adjusted stroke mortality in the United States has not been accompanied by a fall in hospitalization over recent years. Thus far, however, decrease in length of stay has more than offset increased admission. At the same time, the sharp drop in hospital case fatality rates suggests that continuing decline in stroke mortality may be due, in large part, to improved survival after acute stroke.