Understanding Neuropsychological Statistics in Diagnosing Brain Injury


Posted on 2nd April 2008 by Gordon Johnson in Uncategorized

, , , , , , , ,

Yesterday’s blog threw out a few numbers to illustrate some basic starting principles about neuropsychology. As an aid to our further discussion of this neuropsychology, today I will give some basic numerical principles to help in further understanding the numeric part of neuropsychological assessment.

First, neuropsych scores are typical given in one of three scoring methods: Standard score, percentile score and T scores. T scores are a little bit too complicated to try to explain to a laymen, so I will limit this discussion to standard scores and convert them to percentile scores.

Most people are somewhat familiar to standard scores, because IQ’s are given in them. Yesterday I used the example of our successful professional who had a post accident IQ of 135. An IQ of 100 is perfectly in the middle. Something below 70 is evidence of significant impairment. Each time you move down the standard score grid by 10 points, it represents a significant drop.

Here are the basic categories of Standard scores, with their percentile equivalents.

Very superior — 130 and above — 98% and above
Superior __ 120 to 129 — 92% to 97%
High Average — 110 to 119 — 76% to 91%
Average — 90 to 109 — 25% to 75%
Low Average — 80 to 89 — 8% to 24%
Borderline — 70 to 79 — 3% to 7%
Impaired — below 70 — 2% and below

T scores use the same basic concept, and again using 10 points as the break point, but with a T score, the mid point is 50. Some neuropsychologists may disagree as to the exact point that separates these categories, but this is certainly representative of the concept.

The second term to understand in terms of understanding the statistical analysis done by a neuropsychologist is the concept of “deviations”. While I am incapable of synthesizing the dozens of different explanations of this concept into one cohesive definition, in essence, when you move from one category like very superior, to superior, you have moved one deviation. When you move from very superior to high average, that would be two deviations. Movements of two deviations are deemed to be significant.

Yesterday’s example of an IQ score of 135, which was very superior, to an average processing speed score of 100, is a movement of three standard deviations. That could be quite significant, but of course is only one factor to be looked at in doing a full blown “assessment.”

Tomorrow: assessing premorbid IQ and other ability levels.

Neuropsychological Assessment to Establish Brain Injury


Posted on 1st April 2008 by Gordon Johnson in Uncategorized

, , , , , , ,

In yesterday’s blog, I talked about the essentials prerequisites to proving to a jury that a plaintiff is disabled by brain injury. I said there:

  • “Now, we have more cases than we did in 1996 where the neuroimaging is abnormal. Yet, we still must show the same things: an accident with the potential to injure the brain, acute evidence that the brain was injured, deficits that can be determined in how a person functions and a CHANGED PERSON. Neuroimaging adds to the equation, but doesn’t eliminate any of the other issues. The only thing I would seriously change from what I said in 1996 is that there are other ways in addition to neuropsychological assessment, that deficits in ways in which the brain are working, can be identified.”
The big change in the way I look at the structure of a brain injury case than I did when I started in this field nearly 20 years ago, is that I don’t see pure discrepancy analysis within a neuropsychological test battery to show relative deficits, as the cornerstone to diagnosis. That is a lot of jargon; let me explain what I mean. First some terms:

Neuropsychologist: is a not an M.D., but a Ph.D. in psychology, who has typically finished a post doctoral fellowship and training in neuropsychology, which is essentially the field of brain behavior and assessment.

Neuropsychological assessment begins with the administration of a battery of psychometric tests. Then the neurospcyhologist will do an analysis of the pattern of the test scores, the clinical interview of the patient and known potential traumatic or disease processes, to make an assessment as to what pathology may exist in the brain, and from what potential causes.

Discrepancy analysis is the technical, statistical analysis of the neuropsychological test battery to determine whether there are relative weaknesses in an intraindividual comparison, upon which conclusions about pathology can be made.

An intraindividual comparison is a method of determining whether or not a portion of a brain is performing abnormaly for that person, based on the pattern of tests scores, primarily within the specific battery of tests that are being performed at that time.

A relative weakness is a test score on a specific test within the battery where the score is sufficiently lower than other tests, that it shows that a particular part of the brain may be functioning in a pathologically changed way.

All of these technical terms and approaches are usually necessary because only in rare cases does an individual have previous neuropsychological assessments that precede their injury or disease. It is thru these technical approaches to evaluations, that a neuropsychologist can make determinations of pathology, without prior batteries to contrast current testing with.

To demonstrate how the statistical part of the assessment would work lets assume a simple example – focusing on a small part of the test battery. Let us assume we are assessing a very smart professional, who had excelled throughout his or her academic life, obtaining an advanced degree and always testing at the high end of all standardized tests.

One of the key elements to all neuropsychological assessments is the administration of the IQ test. Our hypothetical individual does as expected and receives an IQ score of 135, which is considered very superior. (More on the categories of achievement levels in tomorrow’s blog.) In contrast, when given tests which measure this individuals processing speed, the score was 100, which is still average, but is more than 35 points lower than the IQ score. If this person’s processing speed was compared to all individual’s, the score would be considered normal. But if Discrepancy Analysis is used to make an intraindividual comparison of the IQ score to the processing speed score, that person would be found to have a relative weakness. That relative weakness could begin to form the basis of an opinion about pathology, and perhaps pathology related to a specific event.

The key issue in engaging in formal discrepancy analysis would be a determination of how rare it is for someone with a 135 IQ to have a 35 point difference between that score and the processing speed.

One piece of this puzzle that most neuropsychologists would not mention, but I personally find significant, is that if this individual had consistently been in the top few percentiles on standardized testing, we can almost presume that they were capable of fast thinking. If you don’t think fast, you don’t get high scores on college or graduate school admissions tests.

But my practical approach contrasted to the technical approach of most neuropsychologists, is symptomatic of another major schism in the field: the method used to determine pre-morbid (pre-injury or disease) abilities.

More on these issues later this week.