Final Review

1
Final Review
2
Bias

• A flaw in either the study design or data
  analysis that leads to an erroneous result
• Selection bias
• The tendency of a sample to exclude some members
  of the population being sampled and
  over-represent others
• Measurement Bias
• Systematic differences of the measurements
  recorded in different study groups

3
Confounding Variables

• A confounder is something extraneous to a study
  that happens to one group but not the other and
  that could influence the studys final results
  (AKA Extraneous Variables)
• Are undesirable variables that influence the
  relationship between variables under study
• They influence the outcome of an experiment, but
  are not the variables that are of interest

4
Variables

• Independent variable
• A variable that is manipulated by the
  investigator
• Variable that is presumed to cause a dependant
  variable
• Dependent variable
• A variable that is measured by the investigator
• Caused by the independent variable

5
Experiment vs. Quasi-experiment

• Experimental research involves the use of
  randomization
• Random selection and/or assignment
• Purpose is to test an hypothesis
• Quasi-experimental research does not use
  randomization

6
Experimental Designs

• Pretest-Posttest Group Design with random
  assignment (Classic exp.)

• Two-group Posttest-only Randomized Experiment

7
Quasi-Experimental Designs
Pretest-Posttest Non-equivalent Groups Design
Program Variations
8
Experimental Research

• Involves
• The random assignment of subjects to groups
• Then manipulating one variable to see its effect
  on another variable
• While controlling for as many other variables as
  possible

9
Validity

• A study is valid if its measures actually measure
  what they claim to and if there are no logical
  errors in drawing conclusions from the data
• There are many types of validity
• Whats important is that they all have to do with
  threats and biases which would damage a studys
  meaning

10
Four Types of Validity

• Conclusion Validity
• Is there a relationship between the variables
• Internal Validity
• Is the relationship a causal one
• Construct Validity
• Was the program implemented and the outcome
  measured as intended
• External Validity
• Can the effect be generalized

11
Definitions

• Construct
• An abstract or general idea
• Operationalize
• To put into operation make operational
• The act of putting the cause and effect
  constructs to work making them operate

12
More Definitions

• Prevalence
• The proportion of people in a population who have
  a disease at a certain point in time
• 50 have LBP orMS 133/100,000
• Incidence
• How many new cases of a disease occurred in a
  population during a specified interval of time
• A proportion

13
Outcome Measures

• The tools used to track the progress of a
  patient's health
• The more measurable or quantifiable the OM is the
  better it is
• ROM is a good example
• A patient stating they are feeling a little
  better today is a bad example
• OMs are collectively used to assess outcomes

14
Properties of Useful Outcome Measures

• Utility
• The outcome measure must be useful
• Reliability
• It must be dependable
• Validity
• It must be able to measure what it is supposed to
  measure
• Relatively easy to use
• Cost effective

15
Properties of Outcomes Cont.

• Sensitivity
• An OMs ability to identify patients with a
  specific condition
• Specificity
• An OMs ability to identify those who do not have
  the condition
• Responsiveness
• Ability to measure differences over time

16
Sensitivity Specificity

• Sensitivity is the probability that a test is
  positive (or a symptom is present) given that the
  person has the disease
• AKA the true positive rate
• Specificity is the probability that a test is
  negative (or a symptom is not present) given
  that the person does not have the disease
• AKA true negative rate

17
Sensitivity Specificity Cont.

• Highly sensitive tests are excellent screening
  tests
• Test a lot of people, discover conditions
• Poor for confirming diagnoses
• Highly specific tests are great confirmatory
  tests
• Not so great for screening

18
False Test Findings

• False positive (b)
• A person who has a positive test, but who is
  actually negative
• False negative (c)
• A person who has a negative test, but who is
  actually positive

19
Continuous Test Results

• Examples would be measurements of range of motion
  or lifting capacity
• Often continuous results are reduced to
  categorical or dichotomous
• e.g., Lifting capacity
• lt20 lbs weak
• 21-50 lbs moderately strong
• 51-80 lbs strong
• gt81 lbs very strong

20
Cutoff Score

• Conversion of continuous test scores to a
  dichotomous outcome requires the choosing of a
  cutoff score
• Scores above the cutoff are considered positive
  while scores below are negative
• Blood pressure readings above a certain level
  (cutoff) are considered to be abnormal

21
Change Score

• The difference between the patients initial
  score and the outcome score
• Purposes for measuring change
• To measure change of an individuals performance
• To determine which individuals changed a little
  and which a lot
• To identify factors that lead to a good change
• To draw inferences about treatment effects

22
Properties of Outcomes

• If an outcome measure performs poorly (i.e., has
  poor validity or reliability) use it only in
  conjunction with other accepted patient outcome
  measures
• BTW - An outcome measure is any evaluative
  procedure that measures clinical change
• Change ?

23
OA Tools

• Classified as subjective or objective
• Subjective
• Relies on information derived from patients
• Some think they are unreliable, but research has
  shown the opposite
• Objective
• Tests carried out on patients
• ROM, strength, proprioception, etc.

24
Subjective OA Tools

• General Health
• Pain Perception
• Condition-specific
• Psychometric
• Disability Prediction
• Patient Satisfaction

25
Pain Perception

• Pain scales
• Visual analog scales
• Patient marks their pain level on a line that
  represents a scale ranging from no pain to
  terrible pain
• Numerical pain scales
• A range of numbers (usually from 0-10) with 0
  being no pain and 10 being worst imaginable
• Patient circles a number that is closest to their
  pain

26
General Health OA Tools

• These tools can be used for all patients,
  regardless of their ailment, since they are not
  condition specific
• Because they deal with quality of life issues can
  be used for patients with HAs, back pain,
  digestive problems, etc.
• Will usually parallel the patient's level of
  disability

27
Physiological Outcomes

• Physical performance tests
• Strength and endurance
• Psychosocial
• How the patient reacts to pain stimulus
• ROM and flexibility
• Including inclinometer evaluation, SLR, etc.
• Proprioception
• One-leg standing test
• Functional activity measures
• e.g., Timed up and go test

28
Spine-specific Measures

• The Oswestry Disability Index (ODI) and the
  RolandMorris disability questionnaire (R-M) are
  the most commonly used outcome measures for
  spinal disorders
• 10 sections with total possible score of 50
• If the first statement is marked the section
  score 0, if the last statement is marked 5
• Multiply by 2 to get percentage (only if all
  sections are completed)

29
Scoring Cont.

• Example
• If one section is missed or not applicable the
  score is calculated

30
Oswestry - Score Interpretation

• 0-20 Minimal Disability
• 20-40 Moderate Disability
• 40-60 Severe Disability
• 60-80 Crippled
• 80-100 Bed Bound or Exaggerating

31
Roland-Morris

• Simpler, faster and more acceptable to patients
  than ODI
• More sensitive measure of activity intolerances
  in acute and subacute patients
• No descriptions of the varying degrees of
  disability as in ODI

32
Neck Disability Index (NDI)

• Is a modification of the Oswestry Low Back Pain
  Disability Index
• Also has 10 sections that are scored from 0 5
• Use the same scoring procedure as in ODI

33
Health Questionnaires

• Are not condition-specific so can be applied to
  virtually any type of physical complaint
• SF-36 Health Survey is the most popular
• 8 scales dealing with physical and mental health
  issues

34
NOIR Applies to Outcomes Measures

• Nominal
• Palpating for the presence or absence of
  subluxation
• Ordinal
• Grading muscle strength on a 0-5 scale
• Interval
• Temperature
• Ratio
• Range of motion or dynamometer strength

35
TEST RELIABILITY

• The degree of consistency of a test when it is
  repeated
• Consistency is a function of random error
• All tests have some random error (variability)
• Reliable tests have less

36
Reliability

• Test-retest reliability
• Measures stability over time in repeated
  applications of a test
• Internal consistency
• Tests the degree of correlation between
  individual questions on a questionnaire and the
  total score or individual questions and other
  questions
• Determines the degree that questions measure the
  same construct and are consistent with each other

37
Test-retest reliability

• Questionnaire 1
• 1 hh hh 2 hh hh
• 3 hh hh 4 hh hh
• 5 hh hh 6 hh hh
• 7 hh hh 8 hh hh
• 9 hh hh 10 hh hh

Questionnaire 2 1 hh hh 2 hh hh 3 hh hh
4 hh hh 5 hh hh 6 hh hh 7 hh hh 8 hh
hh 9 hh hh 10 hh hh
?
38
Internal consistency
Questionnaire 1 1 hh hh 2 hh hh 3 hh hh
4 hh hh 5 hh hh 6 hh hh 7 hh hh 8 hh
hh 9 hh hh 10 hh hh
Does - Q1 Q8 Q1 Q9 Q2 Q7
39
Reliability Cont.

• Test reliability
• The consistency of a test between repeated tests
  by the same evaluator (Intraexaminer reliability)
  and between different evaluators (Interexaminer
  reliability)
• The extent to which repeated measurement, by
  different people and instruments, at different
  times and places, get similar results

40
Test Validity

• Refers to the degree which a test or other
  measuring device is truly measuring what it is
  intended to measure
• Face validity
• Does it seem to have merit as a useful test?
• Does it have reasonable face value?
• Content validity
• A tests ability to include or represent all of
  the content of a particular construct

41
Test Validity Cont.

• Concurrent validity
• A measurements ability to vary directly with a
  measure of the same construct
• Often compares a test with a gold standard
• Convergent validity
• The ability of a measurement to correlate well
  with another measure that is related but
  different
• Gold Standard
• The most specific and sensitive test to diagnose
  a disease

42
Test Validity Cont.

• Discriminant validity
• The ability of a test to discriminate findings
  into categories like normal vs. abnormal
• Measures of constructs that theoretically should
  not be related to each other are, in fact,
  observed to not be related to each other
• Predictive validity
• A tests capacity to be a valid screening tool
  for some future health problem

43
Concurrent Validity Trial

• New test is compared to a gold standard
• An already accepted measuring procedure
• e.g. Diagnosis of spondylolisthesis clinically
  vs. radiographically
• A highly reliable, widely accepted gold
  standard yields a better study
• Consequently these trials are limited by the
  meaningfulness of the accepted standard

44
Social Validity

• Determining whether changes in subjects scores
  are consistent with normative scores
• Comparing the results of patients tests to
  normal values

45
Measurement Error

• Every measurement consists of two components
• True score Error component
• Systematic error
• Differences between the observed score and the
  true score due to the testing situation
• Random error
• Factors related to the subject being examined
  which might randomly affect measurement

46
Systematic vs. Random Error
Systematic error (Bias)
F r e q u e n c y
Random error
Mean Truth
47
Sources of Error

• Biological variation
• Accuracy or validity of the instrument used
• Inter-examiner reliability of the instrument
• Intra-examiner reliability of the person reading
  the instrument
• Others

48
Correlation Coefficients

• Intraclass Correlation Coefficient
• ICC is mainly used to measure interexaminer
  reliability
• Can be used for two or more raters
• gt0.75 indicates excellent reliability
• 0.40 to 0.75 fair to good
• lt0.40 poor

49
Contingency Table
Rater A Present Absent
Rater B Present Absent
Concordance
Disconcordance
50
Kappa

• Measures concordance of dichotomous data
• Poor agreement lt 0.40
• Fair agreement 0.400.59
• Good agreement 0.600.79
• Excellent agreement 0.80

51
Randomized, Control-group Clinical Trial (RCGCT)

• The most common type of clinical outcome
  experiment
• The strongest evidence for causality
• Is a prospective study that compares the effect
  of an intervention against a control (placebo)
• Equivalent groups are required
• Produced by means of random assignment

52
Random Assignment

• Random assignment will be more successful
• With larger samples
• Fewer number of groups
• We would be more confident of equivalence with
  100 subjects divided into 2 groups than 30
  subjects divided into 3 groups

53
Randomization

• The most accepted and most effective method of
  attaining equivalence
• Its the single most important design strategy
  for reducing the influence of bias
• Strengthens the claim that group differences in
  the study were attributable to the intervention
• Random selection is preferred because it
  decreases the likelihood of selection bias

54
Blinding

• Double-blind study
• A method used to prevent bias
• Neither the patient nor the doctor know whether a
  sham, standard treatment, or new treatment is
  involved
• As mentioned previously, blinding is nearly
  impossible in the chiropractic setting
• Single-blind study
• Patient is not aware (instrument adjusting)

55
Subdivisions of Clinical Trials

• Treatment (therapeutic) trials
• Test new treatments, new techniques, new drugs,
  or new surgeries
• Prevention trials
• Look for better ways to prevent disease in people
  who have never had the disease or to prevent a
  disease from returning
• May include treatments, vitamins, vaccines, or
  lifestyle changes

56
Subdivisions of CTs

• Diagnostic trials
• Conducted to find better diagnostic tests or
  procedures
• Screening trials
• Test the best way to detect certain diseases or
  health conditions
• Quality of life trials (supportive care trials)
• Find new ways to improve the quality of life for
  people with a chronic illnesses

57
Population

• In research, population refers to the units from
  which a sample is drawn
• A universe population is the total
  theoretically possible units of study
• A study population only includes the sampling
  units from which the sample is to be selected
• Also, a population is the set of people to which
  findings are to be generalized to

58
Random Vs. Non-random Sampling

• Random sampling
• Data collection where every person in the
  population has an equal chance of being selected
• Non-random sampling
• Every person does not have an equal chance, may
  result in systematic overrepresentation
• Stratification
• The grouping of individuals in a population into
  homogenous groups regarding a certain
  characteristic

59
Statistical Analysis

• 2 groups t-test
• Find the value for t and then see if its large
  enough to be statistically significant
• 3 groups ANOVA
• Find the value for F and then see if its large
  enough to be statistically significant