Abstract: Objective To investigate the trend of length of stay (LOS) and identify variables that predict LOS for inpatient rehabilitation after stroke. Methods A total of 3049 inpatients who completed the stroke rehabilitation program at Beijing Bo'ai Hospital from January 2002 to December 2011 were reviewed and analyzed with nonparametric rank sum test and binary Logistic regression analysis. Results The median of LOS was 82 days, and tended to decrease obviously in 2010 and 2011. On Logistic regression analysis, there were 13 variables associated with LOS, in which operation, muscle spasm, gain of activities of daily living (ADL), onset admission interval, nosocomial infection,complications, speech disorder, the type of stroke, the number of hospitalization, methods of payment were the risk factors, and age,functional ambulation classificateion and Fugl-Meyer Assessment were protect factors. Conclusion Reducing LOS must base on rehabilitation efficacy, and the improvement of hospital management process, the prevention of complications and nosocomial infection control, and early rehabilitation intervention can reduce the LOS.

Key words: stroke, rehabilitation, length of stay, related factors