Pediatric rehabilitation: trends in length of stay.

Length of stay (LOS) is one of the most important pieces of data used to measure clinical rehabilitation outcomes, but there is a paucity of LOS data regarding pediatric acute rehabilitation. The purpose of this study was to predict LOS based on admission diagnosis to an acute pediatric inpatient rehabilitation unit. The hypothesis was that LOS will vary according to admission diagnosis. One thousand one hundred forty-five patients were admitted to our acute inpatient rehabilitation unit from January 1, 2000 to December 31, 2007. Common admission impairment groups were orthopedic conditions (29.3%), brain injury (17.9%), brain tumor (7.9%), pain syndrome (7.5%), complex medical conditions (7.4%), stroke (7.2%), meningoencephalitis (7.2%), and spinal cord injury (4.4%). The distribution of LOS data was significantly skewed to the right (for example, mean 40.6 days and median 26 days, respectively, in brain injury group). The median LOS of patients admitted with orthopedic conditions was the shortest (13 days), compared to patients admitted with spinal cord lesions, which was the longest (32 days). Logarithm-transformed mean LOS was different among the admission impairment groups (F=28.7, p < 0.01). However Tukey's Honestly Significance Difference test further showed that: 1) LOS data was not always statistically different across admission impairment groups, and 2) LOS of patients with orthopedic conditions and spinal cord lesions was the shortest and longest, respectively, compared to other admission impairment groups. No proportional decrease in median LOS was observed across the impairment groups from 2000 to 2007. LOS for pediatric acute inpatient rehabilitation varied according to admission diagnosis. When considering future comprehensive pediatric rehabilitation outcome studies, focusing on a uniform impairment group is suggested. In addition, attention to other factors such as functional status changes, severity of illness, payment types, and psychosocioeconomic status should be considered.