Monitoring recovery after injury: procedures for deriving the optimal test window

Researchers studying the impact of treatments designed to facilitate recovery after neural injury face competing demands. On the one hand, because treatment effects often emerge slowly over days, and because researchers seek evidence of stable long-term effects, it is common practice to observe experimental subjects for many weeks after treatment. On the other hand, the cost of performing studies and the need to evaluate a multitude of alternative treatment procedures requires optimal efficiency, pushing researchers towards shorter test procedures. With these issues in mind, researchers have appeared to derive a test window based on previously published methodologies and inspection of their recovery curves, with testing terminated after the recovery curve reaches asymptote (approaches a slope of 0). An alternative procedure is introduced here that evaluates the stability of the data set over time. Using correlational techniques, researchers can determine whether (1) testing should be continued for additional days; or (2) equivalent statistical power can be achieved in fewer days. This provides a rational decision rule to help researchers balance competing demands. Applying these techniques to a procedure that evaluates the impact of acute treatments on recovery from spinal cord injury, it is shown that equal statistical power can be achieved in half the time, greatly increasing the efficiency with which alternative treatments can be evaluated.