FORECASTING FUNCTIONAL GAINS PLATEAU FOR NEUROREHABILITATION PATIENTS

Monday, October 25, 2010
Sheraton Hall E/F (Sheraton Centre Toronto Hotel)
Tammy Sieminowski, MD, MEng, Bridgepoint Hospital, Toronto, ON, Canada and Michael Carter, PhD, University of Toronto, Toronto, ON, Canada

Purpose: Inpatient neurorehabilitation leads to functional gains for most patients, however complete functional recovery may not occur and eventually gains plateau. Timing discharge from rehabilitation with onset of functional gains plateau is desirable.  We developed a mathematical model forecasting onset of functional gains plateau based on the pattern of individual functional gains that may provide decision support to multidisciplinary teams.

Method: Retrospective analysis of data from 240 patients admitted for neurorehabilitation. A logarithmic equation was generated for each patient using quantified standardized observations of functional status recorded biweekly at multidisciplinary rounds and plotted over time. Functional gain plateau score, defined as the functional gains made over the final 14 days of rehabilitation, was calculated for 191 patients admitted over the initial 18 month period. Scores were randomized into two groups. Group 1 plateau scores were distributed, 50th and 95th percentile values obtained, three ranges delineated (high, medium, and low), then validated with Group 2 plateau scores. Using an additional 49 patients’ individual functional gains equations, daily projected interval 14 day functional gain scores were calculated and the final day in the interval classified according to the validated plateau score ranges. Time from second recorded functional score to onset medium and low gains ranges, and to discharge day were calculated using Kaplan-Meier survival analysis.

Result: Group 1 (n = 98) functional gains plateau scores fit a lognormal distribution(K-S = 0.00993, p = 0.27). Ranges were validated using Group 2 (n = 93) observed (8/93, 37/93, 48/93) versus expected proportions. All ranges were accepted using Binomial probabilities;β(x>=8; 93, 0.05) = 0.1, β(x<=37; 93, 0.45) = 0.18, β(x>=48; 93, 0.50) = 0.42.   For Group 3 (n = 49) patients at the time of second recorded functional score: 38/49 were in the high gains range with mean time to onset medium gains range 10 days (SE 1.2), low gains range 27days (SE 2.3), to discharge 37 days (SE 3.2).  10/49 patient were in the medium gains range, with mean time to onset low gains 4 days (SE 0.8), and to discharge 26 days.(SE 3.9).

Conclusion: Onset of functional gains plateau for patients undergoing neurorehabilitation can be quantitatively forecast based on individual functional gains.  This may be useful in determining discharge dates and overall length of stay.