Exam notes Flashcards

Question

Systematic review

Answer 1

- Synthesis of all eligible controlled trials about a specific question - Common method used to pool all data from a RCT-> meta-analysis

Answer 2

⇒ Normal 1. - Within the usual range of variation in a given population or population groups, or frequently occurring in a given population or groups, 95.44%, within 2 standard deviations of mean 2. In good health, indicative or predictive of good health or conducive to good health (for a diagnostic test -> normal is in a range in which the probability of a specific disease is low) 3. Gaussian distribution ------------------------------------------→ - Normal limits: compare normal values obtained when measurements are made in 2 groups, one that is healthy and has been found to remain healthy, the other ill, or has subsequently found to become ill - Result may be 2 overlapping distributions (overlap may have normal or abnormal patients) read the reading

Answer 3

⇒ Reliability - Degree to which the results obtained by a measurement procedure can be replicated under identical conditions - Lack of may arise from divergences between observers or instruments or instability of the attribute being measured 1. Inter-rater reliability - Clinicians agreeing - By chance alone will be 50% of the time - Examiners discipline, experience level, consensus on producer used, training before study and use of symptomatic subjects does not improve reliability 2. Intra-rater reliability - Level of agreement by same observer 3. Kappa - rates agreement - from 0-1 - level of agreement beyond chance divided by the level of agreement possible beyond chance look at Kappa value and degree of agreement.

Answer 4

- Method is sound - Usually tested against a gold standard Internal validity - internal study design is such that we can have confidence that it sets out to do what it claims to do - usually achieved by critical appraisal External validity - Capacity of study to be generalizable to target population outside the sample studied - Based on sample of subjects in the study Measurement validity - Degree to which a measurement measures what it purports to measure 1. Construct validity - Extent to which measurement corresponds to theoretical concepts concerning phenomenon under study - Eg. Spinal degeneration worsens with each adult decade as does hair loss and weight gain 2. Content validity - Extent to which measurement incorporates domain of phenomenon under study - Eg. Measurement of functional health status should embrace ADL’s, occupation, family and social functioning 3. Criterion validity - Extent to which measurement correlates with an external criterion of phenomenon under study - Concurrent: same point in time, ankle jerk missing and person has L5.S1 disc herniation - Predictive: measurements validity is expressed in terms of ability to predict criterion eg. Finding exposure to a large dose of radiation and being able to predict that the person will develop cancer in the future Face validity - Indicates after reading proposal one believes that it seems valid - Eg. Read about a new instrument designed to measure neck disability and after pursuing the question you decide it appears valid (on the face of it)

Answer 5

⇒ Diagnostic tests - any type of information that could be helpful in making a diagnosis - To identifying anatomical structure involved or process responsible, make informed decision about treatment, give patients accurate info about prognosis - Refine probability of presence/absence of a disease or disorder (the results of a diagnostic test should change the likelihood of a particular diagnosis - What we think before test + test information = what we think after ⇒ Sensitivity - Proportion of truly diseased persons in screened population who are identified as diseased by diagnostic test ie. Probability that any given true case will be identified by test “true positive rate” - if low -> proportion of patients with disease who have a positive test is low ⇒ Specificity - Proportion of truly non-diseased persons who are so identified by screening/diagnostic test Ie. Measure of probability of correctly identifying a non-diseased person with a screening test “true negative rate” - If high -> proportion of patients without the disease who have a negative test is high view tables

Answer 6

⇒ Pre-test probability: Probability patient has disease before undergoing test 1. Clinical data - derived from clinical assessment (previous test results from patient, personal clinical experience) - reliant on careful, unbiased estimate 2. published data - giving prevalence of disease in a defined clinical setting/population 3. clinical prediction rules - developed from research into factors that may predict a disorder that responds to a specific treatment - ie. Have these symptoms -> more likely to have this disease

Answer 7

⇒ Post-test probability: probability patient has disease given test results Likelihood ratios (LR’s) - Alternative way of describing performance of a diagnostic test - Can be used to calculate probability of disease after a + or – test - Likelihood that a given test result would be expected in a patient with the target disorder compared with the likelihood that the same result would be expected in a patient without the target disorder - Defines how much a + or - test result modifies the probability of disease -be able to calculate it -liklihood ratio

Answer 8

1. Is the study valid (internal validity)? - Measurement: was there an independent blind comparison with a gold standard reference test? - Representative: was test applied in appropriate spectrum of patients? - Ascertainment: was reference standard applied or ascertained regardless of test result? - Were test methods described in sufficient detail to permit replication? 2. Are the valid results of this diagnostic study important to me? - What were the results? - Are likelihood ratios provided or is data necessary for their calculation provided? 3. Can I apply this valid, important result to my patient (relevance)? - Will reproducibility of test result and its interpretation be satisfactory in my setting? - Are results applicable to my patient? - Will results change my management? - Will patient be better off as a result of the test?

Answer 9

Traffic light system is used to examine the patient. Red light- illogical test that hasn't been tested - illogical test that was tested and found not to work - logical test that has been tested and found not to work Yellow Light: - logical test that hasn't been tested - logical or illogic test that has been tested and received different results or with moderate success Green light: - Logical test that has been tested and found to work - illogical test that has been tested and found to work - several moderately useful test combined and found to work when considered together Why do you examine your patient? To detect any contra-indication or non-indication to chiropractic treatment.

Answer 10

Selection bias- allocation of subjects to groups Performance bias- implementation of interventions e.g. real vs sham Attrition bias- follow up of participants Detection bias- Evaluation of outcomes

Answer 11

Precision: measure of likelihood of random errors, reflected in confidence intervals around estimate of effect. CI's can be used to examine statistical significance and provide us with more info than P values

Answer 12

- exact effect size (treatment effect observed in particular study) is called point estimate - best estimate from study of true effect size - precision or stability of estimate may be expressed using confidence intervals - 95% CI usually reported - studies with larger sample sizes give more precise estimates and narrower confidence intervals - the narrower the CI- more certain you can be about true effect - results are statistically significant at 0.05

Answer 13

- difference due to chance alone , represents a true difference - FALSE POSITIVE - statistics used to determine likelihood of type 1 error = a - P value quantifies probability that observes difference between 2 studies by chance alone (assuming there is no difference between groups)

Answer 14

P < 0.05= statistically significant - 1/20 prob that difference is due to chance - highly statistically significant results could be misleading if study methods were bias or had systematic error

Answer 15

- internal validity: extend to which design and conduct are likely to prevent systematic error or bias - more methodologically rigorous studies may yield results closer to the truth i.e. better external validity --> generalisable.

Answer 16

CI's also used to help interpret results of studies that appear negative - negative result may be true negative or false positive (type 2 error) - B - may arise if study has insufficient statistical power to detect true difference between groups - often occurs because of lack of subjects.

Answer 17

- results of studies with binary outcomes (yes/no) may be expressed in several ways - relative benefit of active treatment over control is often expressed as relative risk or odds ratio. - if applying results to clinical practice --> NNT is more meaningful

Answer 18

- Number of patients needed to treat to provide one additional positive outcome when compared to other therapy - In harm prevention study: number of patients needed to treat to prevent one additional bad outcome - NNH = number of patients required to treat before seeing one case of harm

Answer 19

-difference in proportion of subjects with the outcome of interest in each group. -probability of an event in active treatment group divided by probability of event in control group 1- relative risk, measure of absolute risk reduction, ARR divided by probability of event occurring in control group

Answer 20

- Results of RCT may not be generalizable - RCT’s address question of whether treatment can work (efficacy) but may not tell us whether treatment may be effective (effectiveness) when offered to broad range of patients see in clinical practise - Efficacy: treatment might work but might not be accepted/ will work outside clinical trial - Effectiveness: commonly accepted

Answer 21

- Used in order to limit sources of heterogeneity that may influence study results - Can narrow focus so results may not be generalizable

Answer 22

- Impute numbers into data to replace missing data

Answer 23

- Need to evaluate evidence about causation -> for validity, importance and direct relevance to patients

Answer 24

-assosiation doenst demonstrate causation

Answer 25

1. Strength of results - Clinical and statistical strength (magnitude of effect) - Larger the effect size, closer the confidence intervals, smaller the p value (signs results didn’t occur by chance alone) - In case control OR must be 4 or more - In cohort, RR must be 3 or more 2. Consistency of evidence - Similar results from other studies of same problem - If effect size is similar, evidence is stronger - Should be similar results and evidence on varying types of study 3. Specificity in cause - In ideal situation, effect has one cause - Outcome is best predicted by one primary factor adds credibility to cause - De-challenge re-challenge test: if adverse outcome decreases or disappears when possible causative agent is withdrawn, then re-appears when agent is re-introduced -> suggests causal agent 4. Temporal relationships - If exposure precedes outcome in time, suggests possible causal relationship - Temporal sequence is correct 5. Dose-response - If likelihood of adverse outcome increases with level of exposure to factor of interest, suggests causal relationship 6. Does association make biological sense - Causal relationship more likely if association between factor and adverse outcome makes biological sense 7. Coherence or consistency of evidence over time - A cause and effect interpretation for an association is clearer when it does not conflict with what is known about variables under study and when there is no plausible competing theories or rival hypothesis 8. Experimental evidence - Any direct research that is based on experiments will make causal inference more plausible - Experimental tests agent to see if person gets disease 9. Analogy - Weakest test - Refers to similar results in a related test - Eg. If lifting heavy weight causes LBP, lends weight to notion that it will cause thoracic pain

Exam notes Flashcards

(49 cards)