Skip to Content
All Abstracts
All Abstracts

Model for Predicting Outcome Appears Good

Categories: Recovery - Long Term

The Question

Is the researcher’s model a good prediction tool for outcome?

Past Studies

Past Studies prove that predicting outcome from traumatic brain injury is hard to do. Outcome refers to the ability of an injured person to return to activities that were performed before the injury. There are many ways of measuring outcome (for instance, return to work, return to driving, or return to living alone). There are also many factors that influence the outcome , such as the severity of the injury, physical problems after the injury, family support, and abilities that existed at the time of injury. Recently, the trend in research is to measure outcome in several ways and to examine several factors that may predict outcome all in the same study. In addition to these factors predicting outcome, the factors are also related to one another, which increases the complexity of the analysis of the data. Only by examining these complex relationships can we understand a complex situation like recovery from traumatic brain injury. Using statistics, models of relationships can be developed and then checked for accuracy.

In 2001, Dr. Thomas A. Novack and colleagues at the University of Alabama created such a model framework. Their goal was to develop a predictive equation for determining outcomes. They examined the influence of individual characteristics and the injury on outcome. The researchers found that injury severity was less important to 12-monthoutcome than the pre-injury characteristics of the individual and the individual’s functional recovery at six months after the injury date. Statistical analysis of the data supported that their model appeared to be a good one to predict outcome. However, due to limitations in their study the cross-validity of the model was not established. In research terms, cross-validity means that a test can be used effectively for a particular purpose time and time again. A model that is cross-validated produces results that are generalizable. For the Novack study to be considered cross-validated, the same study needed to be carried out on the second group of individuals with similar characteristics.

This Study’s Goal

This study’s goal was to replicate and establish cross-validation for the model in the Novack study (mentioned above) using a group of individuals with similar characteristics. There were 107 individuals in the original Novack study and 294 individuals in this study. The majority of the participants were males with moderate to severe traumatic brain injury. The researchers examined the contribution of the individual’s pre-injury characteristics, injury severity, and functional and cognitive (thinking skill) ability levels to the outcome at one-year after injury.

The researchers compared and statistically analyzed the results from the group of individuals that they assessed and the group from the Novack study. The results of both groups was “fitted” into a structural equation to evaluate the “goodness of fit.” In other words, did the new data “fit” the statistical model that Novack and his co-workers had established based on the first set of data? The “fit” was good, supporting the model’s validity. It was confirmed that injury severity appeared to affect cognitive and functional status, but did not directly predict the outcome at one year. Cognitive and functional status appeared to directly influence a one-year outcome. Pre-injury individual characteristics and injury severity did not directly influence the outcome.

Who May Be Affected By These Findings

Researchers, healthcare professionals.


Injury severity did not directly influence the outcome. However, injury severity did indirectly influence the outcome through its effects on cognitive and functional status. The researchers conclude that along with treatment to reduce injury severity, rehabilitation treatments should aim at improving individuals’ cognitive and functional abilities to improve one-year outcomes.

Bottom Line

The researcher’s model appears to be a good prediction tool for outcome based on statistical analysis.

Please take a moment to comment on the value of this abstract:
Click here to take a brief survey

Find This Study

Bush, B. A., Novack, T. A., Malec, J. F., Stringer, A. Y., Millis, S. R., & Madan, A. (2003). Validation of a model for evaluating outcome after traumatic brain injury. Archives of Physical Medicine and Rehabilitation, 84, 1803-1807.

Join Our Monthly Giving Program and Become a BIAA Amplifier. Text describing how your gift sustains our services.

Become a BIAA Amplifier by making a monthly donation in support of brain injury services and research.

Your monthly donation means, together, we will help more people.