Exam 1 Flashcards

Question

In step 3 (appraise), one should....

Answer 1

Critically Appraise: The process of carefully and systematically examining research to judge its validity, results and relevance.

Answer 2

Study results are not always: Valid (actually due to the intervention/exposure) Safe (invalid results can be harmful/wasteful) Useful (meaningful or applicable to your patient)

Answer 3

The degree of confidence that the causal relationship you are testing is not influenced by other factors or variables Are the results due to the intervention (or exposure) we are studying… or something else?

Answer 4

Chance Confounding Bias

Answer 5

Are the results applicable (generalizable) to other populations (patients), settings and time? Typically depends on study population (inclusion and exclusion criteria) and setting

Answer 6

Internal: Those who took Vitamin D scored 10% higher, is this due to the Vitamin D or something else External: Depends on the exclusion and inclusion principles --> Can you apply those results of the Pharm.D. students to engineering students.

Answer 7

Random error Inherent in all measurements (we are only taking a sample of all the possible observations) The more there is, the less precise the results

Answer 8

In the Vitamin D group, exam results were higher. Can I take those results and apply them to every pharmacy student in Canada --> NO If we measured all pharmacy students in Canada, results are more precise; however, never fully able to eliminate chance. The more measurements we take the more precise and closer to the actual results.

Answer 9

Any systematic (not due to chance) error in a study that results in an incorrect estimate of the association between the intervention or exposure and the outcome …or… Problems with the way a study was designed, conducted, or classified that leads to incorrect results or conclusions

Answer 10

- A confusion of effects - When some factor(s), other than the intervention or exposure under study, influences the observed outcome - The factor must be linked to the exposure and the outcome - Coffee causes heart attacks; however, coffee drinkers are more likely to be smokers. Therefore, may not be coffee at all.

Answer 11

Related to the inclusion/exclusion criteria Studies (especially RCTs) are very specific about who is and isn’t allowed in the study Often becomes an issue with drug studies: Who should use the drug? Who the drug is approved for?

Answer 12

Random error – inherent in all measurement Less random error = good More random error = not good Estimated and reported using statistics: P-value

Answer 13

Probability representing the strength of evidence to support the null hypothesis P value represents the probability the results are due to chance rather than a real treatment effect (want a low p-value)

Answer 14

Null Hypothesis: Set at showing no difference between the groups Alternative Hypothesis: There is a difference between the groups

Answer 15

Large P value (>0.05) = supports null hypothesis There is no “statistically significant” difference between the groups’ results Small P-value (<0.05) = reject the null hypothesis There is a “statistically significant” difference between the groups’ results

Answer 16

Increase the sample size (not always possible) Recognize the extent (through statistics) and interpret the results accordingly Recognize that Statistical significance ≠ clinical significance (do not go hand and hand)

Answer 17

Statistical significance ≠ clinical significance (do not go hand and hand) If a study concluded a 10 day difference in survival, this could be statistically signifiant but is not clincially significant

Answer 18

- Smoking - Age - Exercise Level/week - Blood pressure - Diet -Sex Some may be hard to measure: stress, socioeconomic status, genetics Can someone get some extra addictive dumauriers to smoke saturday

Answer 19

1) Randomization (Gold Standard) - Ensures groups are similar in all aspects (known and unknown factors equally distributed) 2) Stratification - Stratify by a certain factor 3) Matching - Observational studies (match by age and sex) 4) Statistical Models (multivariable models) - "Controlled for" or "adjusted for" --> Attempts to adjust for confounding (only variable known and can measure)

Answer 20

TABLE 1: BASELINE CHARACTERISTICS - use clinical judgement to determine if there is a difference between the groups Importance of Table 1 (population characteristics): 1) Indicate confounding 2) Provides insight on external validity (gives insight on people studied and whether can apply to patients)

Answer 21

Problems with the way a study was designed, conducted, or classified that leads to incorrect results or conclusions Usually because the groups were treated differently somehow

Answer 22

1) Selection Bias - Problems with how the study subjects were selected - Want everyone to be selected from the same patient population 2) Information Bias - Problems with measuring, collecting or analyzing information (exposure and/or outcomes)

Answer 23

Systematic error (or differences) in how the study subjects were selected or who participated Primarily affects external validity Self Selection/volunteer bias Healthy Worker (adherer) Bias Attrition Bias (lost to follow up)

Answer 24

- People who volunteer/participate in studies are different from those who don't - Selection Bias

Answer 25

Employed (higher SES-socioeconomic status) individuals are usually healthier

Answer 26

- People who leave study e.g. a/e, found process to rigorous, drop out of pharmacy, death - Main problem is when they drop out of one group (5 and 5 would be less concerned than 10 for 1 group) - Becomes an issue if there are differences between those who were lost and those who weren’t a) Lose important data (why did they drop out/withdraw?) b) If a whole ton drop out of one arm, lose benefit of randomization (not equal anymore) c) Chance becomes more of an issue

Answer 27

Systematic errors in the way subjects were measured or classified

Answer 28

Outcome errors (RCT and observational studies) -Problems with measuring tools (one electronic bp monitor, one group manual) -Problems with actual measurements Exposure errors (more so with observational studies) - Problems with how subjects are categorized (non- vit. D group taking a multivitamin) - Problems with measuring tools Subject or observer variation (interpretation of pain) – everyone different and unique

Answer 29

Primarily a concern when the likelihood of being misclassified is unequal between groups

Answer 30

Recall Bias Interviewer Bias (Minimized by standardized forms, questions) Surveillance/detection bias

Answer 31

Individuals remember things differently Subjects with the outcome (especially if negative) are more likely to remember their exposure – e.g. woman born with a baby birth defect (I did take that bp med one time, where as someone with a healthy baby would not)

Answer 32

Interviewer asks about exposure/outcome differently Leading, probing or influencing questions Multiple interviewers Minimized by standardized forms, questions

Answer 33

One study group followed more closely than the other “If you look…you will find” More outcomes identified because more follow-up, not necessarily because it occurred more in that group

Answer 34

Goal is to minimize bias when designing the study Will never completely eliminate Recognize and acknowledge it Study the Graph on Slide 16

Answer 35

Clear definition of study and sample population Ensure all groups are treated the same except for the intervention Standardized measurements - Same questionnaire for all subjects - Same interviewers - Automated devices - Centralized labs (all analysis in same labs) Collect the same information from all subjects in the same way Blinding (pt or researcher or both are unaware of what group a t is)

Answer 36

Authors and journals tend to publish positive findings, especially with drug trials May lead readers to think there is a consistent association when it’s not really the case Not effecting internal validity E.g. Trial A finds that a new drug is effective (publish), Trial B shows significant adverse effects (not publish) potentially giving a misleadingly favorable impression of the new drug's effectiveness and safety

Answer 37

Test whether an intervention works by comparing it to a control condition

Answer 38

Subjects assigned to a study group: Intervention group …or… Control group (no intervention, alternate intervention)

Answer 39

Randomized = subjects assigned to study groups randomly Have an equal probability of being assigned to any group Can be more than 2 groups

Answer 40

Minimize confounding

Answer 41

- No. Cannot occur in an observational study.

Answer 42

Parallel Cluster Randomized Cross-over

Answer 43

A study sample is selected from a population. From the study sample, subjects are further divided into an intervention or control/comparison group. These groups are followed over time and monitored for the outcome.

Answer 44

Clusters of individuals are randomized instead of individuals (e.g. cities, classes, pharmacies, hospitals) A cluster randomized trial (CRT) is a randomized controlled trial in which pre-existing groups, called clusters, of individuals are randomly allocated to treatment arms.

Answer 45

- Most commonly used when studying an approach to patient care

Answer 46

A study sample is selected from a population. From the study sample, subjects are further divided into an intervention or control/comparison group. These groups are followed over time and monitored for the outcome. There is then a washout period in both study arms. After the washout period then the groups are flipped. Intervention Group --> Control group Control Group --> Intervention Group We then follow the groups over time and monitor the outcome. The outcomes are then compared. One is acting as their own control.

Answer 47

- The individuals act as their own control and therefore confounding is effectively minimized - This type of study does not require as many subjects - It is a more complicated study to run and the wash out period can be lengthy

Answer 48

The study must be designed before it has begun - This limits bias and maintains integrity

Answer 49

Population – who’s going to be included? Intervention – what are we testing? Comparison – what are we comparing it to? Outcome – what outcome are we looking for?

Answer 50

1) Start with a defined population - Who we want to study - Where do we get them from (all subjects should be from the same population) - e.g. Hospitals, clinics, advertisements 2) Specific inclusion and exclusion criteria - Much more strict in RCT - Can limit generalizability 3) Sample Size (n)

Answer 51

Can limit generalizability

Answer 52

Sample Size: how many subjects are required - This is determined by the study power Study power = ability to show a difference between groups if a difference really exists n too low = underpowered n too high = unnecessary exposure, more resources Optimal number is determined statistically (studies recruit around this power number) Must be determined BEFORE the study starts

Answer 53

ability to show a difference between groups if a difference really exists

Answer 54

The exposure of the experimental group. Must be specific. If it is a Drug must include: Dose Regimen Delivery method/Formulation Follow-ups Length of intervention If it is a Procedure (e.g., pharmacist counseling): must include What’s involved Timing Follow-up Length of intervention

Answer 55

- What we are using to compare our intervention to Nothing (placebo/usual care) A different dose A different drug/procedure Standard treatment - We need to consider how many controls will be used

Answer 56

Controls must be treated exactly the same way as the intervention group – the only difference is the intervention

Answer 57

- Could be anything but must be clearly stated and logical - May be objective or subjective outcomes - Established hard or surrogate endpoints - Established primary endpoint or composite endpoint and may have potential secondary endpoints

Answer 58

Objective Measurable (e.g., blood pressure, lipid levels) Subjective Subject’s interpretation (e.g., back pain)

Answer 59

Hard endpoints - Often used as big clinical outcomes E.g., Death, stroke, MI Surrogate endpoints - Used as a proxy or estimate of the hard endpoints E.g., Blood pressure (…which can lead to stroke)

Answer 60

Main result that is measured at the end of the study to see the effect of the intervention

Answer 61

Additional results of interest, but not the main focus Caution when interpreting as the study wasn’t designed around them (statistical issues) NEVER make a decision based solely on a secondary endpoint Allows for the formation of a hypothesis for future studies

Answer 62

A primary endpoint that contains several events E.g., MACE – Major Adverse Cardiovascular Events Instead of just looking at death as a primary outcome, you include MI, stroke and death together Benefit = less subjects needed Disadvantage = hard to determine the true effect of the intervention on each of the event types - Must Look at specific results for breakdown

Answer 63

Confounding

Answer 64

Randomization

Answer 65

We want the groups to be as similar as possible, except for the intervention Randomization (if done correctly) will ensure that the groups are similar in ALL aspects Similar in things we can measure (E.g., age, sex, blood pressure) Similar in things that are unknown (E.g., genetics)

Answer 66

Coin toss, table of random numbers, computer generated, etc. NOT by days of the week, alternating subjects, etc. Must have an equal chance of being in either group and should not be able to tell which group will be assigned next

Answer 67

Complete Block Stratified

Answer 68

No limitations – straight up randomization Won’t necessarily get equal numbers in each group

Answer 69

Used to force balance in the number of subjects in each group E.g., block of 6 = for every 6 subjects randomized: 3 in intervention group, 3 in control group

Answer 70

NO Researcher is not involved with randomization; someone outside the study to minimize bias

Answer 71

Used to achieve similarities in certain baseline characteristics E.g., stratify by age = everyone under 65 is randomized and everyone over 65 is randomized E.g., stratify by clinic = randomization is done at each individual site

Answer 72

Usually randomization is centralized and offsite

Answer 73

Hides the sorting of trial participants into treatment groups so that this knowledge cannot be exploited Refers to not disclosing the allocation (or randomization) scheme to researchers or participants. Critical to the randomization process of an RCT

Answer 74

NO Allocation concealment differs from blinding in that blinding is not disclosing the treatment allocation AFTER randomization has occurred.

Answer 75

Blue Pill vs Red Pill in preventing death - If the researcher knew the randomization scheme, they may wait to randomize a particular individual into the intervention arm as they know they have a higher chance of death - Meredith Grey

Answer 76

- helps to reduce selection bias and allocation bias Allocation Bias: you can figure out the randomization scheme, so put patients into a specific group to get a certain treatment

Answer 77

Analyzing the data for ALL patients and according to the group they were originally randomized to “Lost to follow-up”: Drop-outs, switch groups, deaths, etc. - Need to include these people in our analysis. Baseline and results should match.

Answer 78

Analyzing data only from subjects who completed the study or followed protocol exactly

Answer 79

ITT preserves the value of randomization (might be differences b/w those lost and those remaining) Lots of important data in those lost to follow-up (adverse events)

Answer 80

Unaware of what study group a subject is in

Answer 81

Subjects may report outcomes or adverse effects differently or behave differently Physicians/investigators may report outcomes or events differently or treat subjects differently Researchers may collect data, classify outcomes or events and interpret subject response differently Affects the internal validity of the study - BIAS

Answer 82

Internal Validity - BIAS

Answer 83

Unblinded (open-label) --> Sujects, researchrs and evaluators are not blinded Single blinded --> One group blinded (often the evaluators) Double Blinded --> STANDARD Often subjecys and researchers blinded (may see evaluators blinded) Triple Blinded --> All 3 groups blinded

Answer 84

The more objetcive the outcome is, the less critical blinding becomes (e.g. stroke vs pain severity)

Answer 85

- Blinding reduces bias

Answer 86

Placebo – Latin “shall please” Perceived or actual effect from an ineffectual or inactive treatment Conscious belief about drug/treatment Subconscious association between being “treated” and recovery Can be positive (benefits) or negative (adverse event) Sugar pill, saline, sham procedure Effect of cost, “severity” of treatment…

Answer 87

Observe the effect of an exposure (or intervention) on an outcome without “interfering” with anything Exposures: risk factor, drug, environmental, procedure, etc.

Answer 88

Looks for associations (not causation) Can not prove cause and effect To prove cause and effect, need strong internal validity --> Does not have the randomization piece Randomization minimizes confounding (primarily, as randomizes people into groups which allows us to distribute factors that are known and unknown)

Answer 89

Often more: Practical / Feasible Ethical Real-world -->Wider inclusion criteria (more generalizable) --> Real-life situations (comorbidities, drugs, etc.)

Answer 90

Because clinical trials evaluate such a small percentage of people who will eventually use a studied drug and are of relatively short duration, it is possible that unexpected or rare drug effects will not be detected during the investigative process

Answer 91

RCT's - Cannot use observation studies to guide lcinical practice

Answer 92

Is the association real…or due to something else? Is there internal validity? 3 major threats Chance Confounding Bias

Answer 93

(Case reports / Case series) Cross-sectional Case-control Cohort

Answer 94

Describes and interprets an individual case Usually written as a narrative (e.g. story) No predefined methods No predefined outcomes Patient described in detail Observations described in detail Exposure and outcome(s)

Answer 95

Unexpected events and adverse effects Important variations of known diseases Weird observations that can’t be explained by known disease

Answer 96

Useful in identifying new or unusual trends or diseases Useful in identifying new drug effects (good or bad) Describe novel interventions: --> Treatment --> Diagnostic procedure Suggest areas for further research (or “alarm”)

Answer 97

Difficult to interpret --> Spontaneous appearance and disappearance of signs/symptoms --> Placebo effect --> Other influences Can’t confirm or prove anything

Answer 98

Collection of individual reports which typically occur within a fairly short period of time Can be hypothesis forming…or world altering

Answer 99

Looks at population or study sample at one point in time Snapshot of what is happening at that time

Answer 100

Exposure Outcome Health status Health behaviours

Answer 101

Surveys / Questionnaires Phone, mail, online Chart reviews Administrative data (governments, insurance agencies)

Answer 102

Useful for describing prevalence Proportion of a population or sample that has a condition (disease, risk factor) E.g., The prevalence of MS in SK is 314/100,000 Can’t determine the incidence Number of new cases (people) with a condition over a specific time E.g., The annual incidence of MS in SK is 16.5/100,000 Can be analytical (looking for associations) --> Factors that might be associated with one another --> Can’t determine if an exposure causes the outcome though

Answer 103

Good for determining prevalence Useful for quick examination of potential associations Hypothesis generating…future research Inexpensive

Answer 104

Limited to what we know about the potential associations --> Did the exposure precede the outcome? --> Other contributing factors? --> Transient effects

Answer 105

Retrospective (looking back in time) Compares 2 groups: Cases = people with the outcome Controls = people without the outcome Compare the proportion of cases with the exposure to the proportion of controls with the exposure

Answer 106

Try to have the cases and controls as similar as possible except for the outcome (minimize confounding)

Answer 107

Have cases and controls in present Look back in time. Analyze to see whether the subjects in both arms had an exposure (e.g. drug) Compare the rates

Answer 108

1. Need to be very specific about the outcome --> Want a “validated” outcome --> It really is the outcome we think it is (e.g., actually had angina and not just chest pain) --> Use standardized, accurate measurement/tools, criteria, definitions, etc. Ideally incident (new) cases over prevalent cases

Answer 109

--> Prevalent cases – don’t know when the outcome occurred, and if the exposure preceded it (has came after the outcome has occured) --> Incident cases – we know when the outcome occurred and can be sure the exposure preceded it

Answer 110

Want the exposed and unexposed groups to be as similar as possible (except for outcome) --> Should come from same population as cases (e.g. not having control from Toronto, and case from Saskatoon) Often use multiple controls for each case --> Increase sample size….decrease random error --> Increases the power of the study (but no benefit after more than 4 controls per case)

Answer 111

4:1 = 4 controls for every case

Answer 112

Often use matching to help ensure groups are similar Helps decrease confounding (known confounders) Can only match on factors we know (or can measure) Sex, age, location, baseline cholesterol levels, etc…

Answer 113

Good for studying rare diseases or outcomes Relatively quick to do since outcome has occurred already – data is already available Good for establishing an association (leads to further research)

Answer 114

Not randomized – susceptible to confounding Susceptible to bias Only suggest associations – can’t prove cause and effect

Answer 115

Groups (or cohorts) of subjects are created and compared according to exposure Exposed vs. unexposed group Groups followed over time to see if outcome occurs

Answer 116

Similar to an RCT but we didn’t randomize subjects into group or assign the exposure

Answer 117

Best type of observational study for finding a valid association

Answer 118

Cohort study

Answer 119

Propsective Retrospective

Answer 120

At the start of the study, outcome hasn’t occurred yet We have an exposed and unexposed group. We follow over time and identify the number of people who experienced the outcome over time. Compare the rates.

Answer 121

Outcome occurred before study started (uses retrospective data) Start at a point in history with those exposed and those unexposed. Follow up to a a certain date. Analyze the number of people in both groups who have the outcome. Compare the rates.

Answer 122

Prospective: Conducted in real-time Have some control over what data you want to collect and where it’s coming from --> Pre-existing data sources (charts) --> Subjects (interviews) Retrospective: Good for long follow-up studies (observation is already complete) Use retrospective data collected prospectively Data limitations Accuracy or completeness Missing information on important factors (e.g., lifestyle) – Big Gaps

Answer 123

Drug Procedure Occupational Environmental

Answer 124

Exposure and outcome must be specifically defined E.g., degree of drug exposure (dose and length) Exposure must precede the outcome

Answer 125

Chance is not as a major in the studies --> ONE THREAT MINIMIZED DUE TO LARGE NUMBERS The numbers we get is significant as compared to an RCT so have less issues with chance

Answer 126

Groups (exposed and unexposed or “comparator”) should be as similar as possible Select from same population Can have more than one “comparator group” E.g., Various levels of exposure --> Potential unknown confounders even if groups are similar

Answer 127

Provides highest level of evidence for observational studies Can study multiple outcomes after a single exposure

Answer 128

Susceptible to bias Not randomized so susceptible to confounding More resource intensive than other observational studies

Answer 129

A review of a clearly formulated question that uses systematic and explicit methods to identify, select, and critically appraise relevant research Collect and analyze data from the studies that are included in the review

Answer 130

The use of statistical techniques (usually in a systematic review) to integrate the results of independent studies

Answer 131

Extract the data and main result from each individual study Calculate a pooled average result --> Greater weight given to studies that provide more information (e.g., studies with bigger sample sizes)

Answer 132

All systematic reviews are not meta-analyses and vice versa Shouldn’t always statistically combine study results into one overall result

Answer 133

Clinical reasons Do all the studies address the same question? Similar populations? Similar interventions and/or comparisons? Similar outcomes? Statistical reasons Study “heterogeneity” (how different are the studies) Are the results between studies consistent? Q or I2 statistic

Answer 134

Tons of research Systematic reviews provide opportunity and good summary of all clinical evidence out there

Answer 135

Anyone can publish a systematic review and/or meta-analysis ~2500 SRs indexed annually on MEDLINE Often found in specialty journals (increasingly more seeing them being published in medical journals)

Answer 136

1993 - Archibald Cochrane Cochrane Collaboration a compilation of high-quality, evidence-based medical research and systematic reviews International network of 30,000+ volunteers in 53 review groups Prepare, update and promote Cochrane Reviews Internationally recognized as the highest standard in evidence-based healthcare (IMPORTANT) No industry money

Answer 137

Robust/consistent and specific methodology for review No industry money

Answer 138

Question of interest --> Study protocol – inclusion/exclusion criteria --> Decide if only RCT or also include observational studies Comprehensive literature search – search terms (librarian involved) Identify appropriate studies for inclusion Go through abstracts Scan the bibliography, reach out authors, research ongoing trials Assess study quality Extract data Analyze data (if doing a meta-analysis) Interpret results – write review Questions can inform answers even although idiot

Answer 139

Identify appropriate studies for inclusion Go through abstracts Assess study quality Extract data Should be done by at least 2 separate reviewers 3rd reviewer to break a tie

Answer 140

Not all meta-analyses are done properly Does the review address a clear question? Did the authors look for appropriate papers? Did the authors assess the quality of the included studies? If there was a meta-analysis done, was it appropriate to do so? (Was a measure of heterogeneity reported? Q or I2 statistic?) What is the overall result of the review? Do these results apply to my patient(s)? Who paid for the study?

Answer 141

Garbage in = garbage out Including results from poor quality studies

Answer 142

thesis, abstract, non-peer-reviewed (risks to this, not a necessity, useful if dealing with rapidly changing fields)

Answer 143

Systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific circumstances

Answer 144

Get evidence into practice

Answer 145

Describe a range of practices for the diagnosis, management, or prevention of specific diseases or conditions that meet the needs of most patients in most circumstances (NOT ALL CIRCUMSTANCES OR PATIENTS)

Answer 146

Systematic review (+/- meta-analysis) of research Focus on the strength of evidence (KEY) Need strong evidence Set of recommendations How to manage patient and/or condition Describe target population Should be specific Inclusion and exclusion criteria Describe target audience Specialists / GPs Other HCP Patients Administrators / Insurers Expert Panel Description of recommendation development Actual methods - Systematic review / Meta-analysis --> ESSENTIAL - “Expert opinion” and “usual practice” SHOULD BE MADE CLEAR if lack of evidence for the opinion External Review Funding Statement Some people always eat rice everyday for fun

Answer 147

Range of experts should be used (such as): Specialists / GPs Nurses Pharmacists Other HCPs Patients (SHOULD BE INCLUDED – family members, people caring for them) Important for asking questions, patient values and interests Statisticians Librarians Ethicists Information on each panel member (including conflicts of ineterst)

Answer 148

Should be reviewed by individuals not involved in the development Often sponsored or endorsed by organizations Doesn’t always mean quality

Answer 149

Most guidelines only produced every few years Quicker in fast-moving fields (Should state the next date at which the guidleines are to be reviewed)

Answer 150

Focus around a specific clinical questions Are subject to further updates, revisions, or developments Often issued when there's an urgent need for guidance in a particular area (just a singular question)

Answer 151

Should report where funding came from Funder(s) should have no influence on the content

Answer 152

Clear Specific Linked to evidence

Answer 153

Most use the GRADE system (Grading of Recommendations, Assessment, Development and Evaluation) Number (1 or 2) = strength of recommendation Letter = quality of evidence supporting the recommendation

Answer 154

Study slides

Answer 155

Garbage in = garbage out Studies of poor quality Poorly done systematic review / meta-analyses

Answer 156

Poor quality studies (primary literature) Incorrect or invalid information Poorly conducted systematic review / MA Publication bias Self-serving / biased – drug company involved promoting their drug Not user-friendly – readable and accesbible Liability?

Answer 157

AGREE II 23 items – 6 domains:

Answer 158

Ultimate goal: want healthcare providers to use the guidelines Barriers to healthcare providers using guidelines Should be accessible free online Guidelines should have summaries, decision aids, easy snapshots and tools to help apply the evidence