Assessment issues Flashcards
What is evidence-based neuropsychology practice?
Providing care integrating clinical data, research literature, individual patient characteristics, financial realities, and cultural/social norms.
What is flexible-battery approach?
Hypothesis-driven approach, systematic screening of all domains, more detailed assessment when deficits are identified
What is process approach?
Focusing less on test scores and more on behavioral processes while obtaining that score
What is normative comparison standard?
Species specific standard (e.g., language development) - lack is a pathognomonic sign (abnormal)
Population average - important to know whether distribution is normal or skewed
Does not provide sufficient information to identify a deficit within an individual, premorbid level of function is important
What is individual comparison standard?
Comparison of scores relative to premorbid level of ability
What is an example of ordinal data?
Likert scale data
What is an example of interval level data?
T scores
What is the difference between sensitivity and specificity?
High sensitivity rules out dx with a negative test result (sn-nout)
High specificity rules in a dx (sp-pin)
What is positive/negative predictive value?
Likelihood that the person has the disease when the test is positive
Or negative when the condition is absent
How is prevalence rate related to SN, SP, PPV, and NPV?
Prevalence rates affect PPV and PNV but not SN and SP
Decreased prevalence rate - decreased PPV
What is post-test probability?
PPV for a positive test
1-NPV for a negative test
What is the difference between pre-test probability and pre-test odds?
Pre-test probability = prevalence rate
Pre-test odds = prevalence/1 - prevalence
What is LR+?
Likelihood ratio positive - estimate of how much a positive test result will change the odds of having the disease
Based on ratio of SN and SP
The more LR deviates from 1, the stronger the likelihood that the disease is present (LR >1)
What is LR-?
LIkelihood ratio negative - change in odds of not having a disease when a test result is negative
The more LR deviates from 1, the stronger the likelihood that the disease is absent (LR < 0.01)
What is the difference between odds ratio and likelihood ratio?
OR - measure of effect size, how harmful/beneficial an exposure is to an individual
LR - extent to which a particular test value increases the likelihood of disease being present or absent
What is parametric statistical modeling?
Impairment defined relative to group
Based on extrapolation of the central limit theorem
Bayesian statistical modeling
Impairment relative to an individual comparison standard
Using corrective variables to improve accuracy of prediction
What are some cautions in interpreting scores as impaired?
Family-wise error rate - increasing likelihood of a positive test finding when multiple tests are administered
Lower IQ = more low scores
Higher IQ = more variability
What is confirmatory bias?
Excluding contradictory test findings or history when it does not fit with a suspected disease or condition
LR of a positive test compares
True positives to false positives
LR of negative test compares
False negatives to true negatives
LR of 1 indicates
That test result is just as likely in those with and without condition and the result does not add much relevant information
What is the risk of focus on deficit measurement?
Underappreciating diversity and overattributing lo scores to brain injury/disease
What are 5 principles for interpreting scores? (from Iverson in LBB)
1) low scores are relatively common across all test batteries
2) low scores depend on where you set your cutoff score
3) low scores vary by number of tests administered
4) low scores vary by demographic characteristics of examinee
5) low scores vary by level of intelligence
What is the limitation of using normal curve in NP?
NC relates to a single score, while NP rely on multiple tests, which yield multiple scores.
E.g., a FP rate of at least one lost score will be considerably higher than that of a single test score
What is the relationship between low scores and intelligence?
Patients with below average intelligence will have more low scores than those with above Ave I.
Among healthy older adults, 22-38% will have one or more scores below 5th %ile across a battery of memory tests; the number of low scores varies by intelligence
What is the relationship between the length of battery and low scores?
The more tests given, the more likely a person to have >1 scores 1SD below the mean
Increasing number of tests will result in increased number of false positive for people with below ave I and increased rate of false negatives for people with above ave I
What are some factors contributing to low scores?
Measurement error
Normative sample characteristics
Longstanding weaknesses
Situational: fluctuations in motivation, fatigue, inattentiveness, minor illness
What is the relationship between SN, SP, and cutoff scores?
Higher cutoff - improve SN, reduce SP (more likely to correctly identify those with problems but also include those without)
What are sex differences in NP to consider?
Women better on tasks of verbal learning and memory, verbal fluency, processing speed, and motor dexterity.
Men better on motor speed, some visual-spatial and visual-constructional tasks, arithmetic reasonign and computations.
What is the key factor affecting test interpretation?
Sample characteristics, including size, non-normal distributions, skew, etc.
What are factors that affect comparison of performance across tests?
Sample differences, measurement error, score magnitude, extreme scores, ceiling and floor effects, extrapolation/interpolation of scores
