EBP exam 2

Significance level (α level)

-standard for significance

-threshold for rejecting the null hypothesis

Standard value of α

5% or 0.05

P-value

Provides the probability that the estimate is due to chance

If p-value is greater than α

Then null hypothesis is accepted (there is no change)

If p-value is less than α

Then null hypothesis is rejected (there is a change)

Power

Probability of finding a difference that exists in a population

How to find power

Power = 1-β

usually β=0.2

meaning power is 80%

Factors that affect power

1. Sample size (greater sample size = more power)

2. Effect size (larger effect size = more power)

3. Variability of observation (less variability = more power)

4. Significance level (larger α = more power)

Type I error

-incorrectly believing there is a difference when there isn't one (false positive)

-rejects the null hypothesis

Type II error

-incorrectly believing there is no difference when there is one (false negative)

-accepts the null hypothesis

Confidence interval

-Indicate the magnitude of difference between groups

-CI = 1-α

Difference data (means)

The result is statistically significant if the CI does not cross 0

Ratio data (RR, OR, HR)

The result is statistically significant if the CI does not cross 1

Range of correlation coefficient

-1 to 1

Range of correlation of determination

0 to 1

Factors affecting confidence intervals

1. Sample size = the bigger the sample, the more narrow the CI, the greater precision

2. Desired level of confidence = higher confidence level, wider CI

3. Variability = lower variability, more narrow CI, greater precision

Parametric 2 independent samples

Student t-test

Parametric 2 related samples

Paired t-test

Parametric 3+ independent samples

One-way ANOVA

Parametric 3+ related samples

Two-way ANOVA

Non-parametric 2 independent samples

Mann-Whitney U test

Non-parametric 2 related samples

Wilcoxon signed rank test

Non-parametric 3+ independent samples

Kruskal Wallis test

Non-parametric 3+ related samples

Friedman test

Nominal 2 independent samples

Chi-Square, Fisher's exact

Nominal 2 related samples

McNemar's test

Nominal 3+ independent samples

Chi square test

Nominal 3+ related samples

Cochran Q test

Ordinal 2 independent samples

Mann-Whitney U test

Ordinal 2 related samples

Wilcoxon signed rank test

Ordinal 3+ independent samples

Kruskal Wallis test

Ordinal 3+ related samples

Friedman test

Parametric test

Data from sample or population that follows the normal distribution (interval or ratio data)

Nonparametric test

Data from sample or population that deviates from the normal distribution (ordinal or nominal data)

Correlation (r)

Determines if one variable changes, or is related to, another variable

Strong correlation

0.9

Perfect correlation

1

Moderate correlation

0.5

What does a positive correlation indicate

direct relationship

What does a negative correlation indicate

indirect/inverse relationship

Coefficient of determination (r2)

The amount of variation between the two variables (if r2=0.55 then 55% of variability is associated directly with the variables)

Linear regression

Used to establish relationships for continuous predictors and outcomes

Shape of linear regression graph

Linear shaped

Logistic regression

Used with categorial outcomes to predict a dichotomous outcome variable

Shape of logistic regression graph

S shaped graph

Time-to-event concerns

-outcome of interest

-withdrawal from study

-censored

Survival analysis

Concern for time to occurrence in any critical event not exclusively death

Kaplan-Meier curve

A representation of the survival rate or survival function, plotting the probability of survival against time

Kaplan-Meier curve interpretations

-Steeper slope = higher event rate (worse survival)

-Flatter slope = lower event rate (better prognosis)

-Each vertical drop in curve represents the occurrence of an event or a censored observation

Kaplan-Meier plot

Used in survival analysis to estimate the probability of an event occurring over time

Kaplan-Meier plot interpretations

-Horizontal lines = survival duration for that interval

-Vertical lines = represent a change in cumulative probability

Internal validity

the degree of confidence that the relationship being tested is true and not influenced by other factors or variables

External validity

the extent to which the results from a study can be applied to other situations, groups, or events

Statistical significance

Will determine whether a difference is likely between groups based on mathematical calculations (p-value and CI)

Clinical significance

A medical judgement that has an impact on clinical practice (change in practice, treatment effect consideration)

If a study is statistically significant, it is clinically significant

False

Title of journal article

-Should be concise and accurate

-Free of bias

-Results and conclusions should not be apparent

AMA citation

3 Authors Last Name First Initial Middle Initial, et al. Title of article. Accepted abbreviation of Journal. Year;Volume(Issue No):Page numbers. doi:xx.xxx

Abstract of journal article

-Provides clear and structed summary of what is included

-Literature evaluation must extend beyond abstract

Introduction/background of journal article

-Introduces disease state of drug

-Describes what's been studied already in previous literature

-Describes area of need or questions not answered by previous literature

-Provides rationale to address these questions

Purpose/objectives of journal article

-Study purpose /objective/hypothesis

-Usually at the end of intro/background

-Objective describes population, intervention, comparator, outcomes

Methods of journal article

-Contains critical information related to conduct, oversight, analysis

-Methods of a trial is determined PRIOR to the start of the study

What is included in methods of journal article

-Study design (observation, randomized, retrospective, etc)

-Study oversight (funding, approval, etc)

-Study population (inclusion/exclusion criteria)

-Trial procedures (recruitment, follow up, etc)

-Trial outcomes (primary and secondary outcomes defined prior to initiation of study)

-Statistical analysis (sample size, Intention to treat, tests used, etc)

What is included in the results of journal article

-Patient screening & enrollment (# of pts enrolled, excluded, withdrew, etc)

-Baseline characteristics (similarities and differences between the participants)

-Length of follow up

-Outcomes (report in %, hazard ratios, CI, p-value etc)

PIES method of critique

-Population

-Intervention

-Endpoint/outcome

-Statistics

What is considered a strength/limitation

Whether the result of an outcome is clinically significant (not if it was statistically significant)

Population of PIES

-are there major differences in pt characteristics

-evaluate overall characteristics of pts

-were inclusion/ exclusion criteria appropriate

Intervention of PIES

-is the intervention representative of current practice or previous studies

-is duration and follow up appropriate

Endpoint/outcome of PIES

-do the endpoints truly represent what is claimed

-are they clinically significant

-if surrogate endpoint is used is it validated for correlation to hard clinical endpoint

Statistics of PIES

-were the tests appropriate

-any data missing

-effect size clinically relevant

Pros of composite outcomes

-increased event rates

-reduce sample size requirement and duration of study

-more efficient/less costly

-multiple outcomes may be equally important

Cons of composite outcomes

-effect may be driven by a less-important component of the composite outcome

-may overestimate the effects of the intervention

-components may be unreasonably combined

Surrogate endpoints

A measure of effect that may correlate with a real clinical endpoint but does not necessarily have a guaranteed relationship