Evidence-Based Practice and Research Methodologies

Levels of evidence

Ranked from 1st to last: Systemic Reviews and Meta-Analyses, Randomized Controlled Trials, Cohort Studies, Case-Control Studies, Cross-Sectional, Case Reports/Case Studies, Mechanistic Studies, Editorials, Expert Opinion, Textbooks.

Pillars of evidence-based practice

Scientific Research, Clinical Expertise, Patient Values and Circumstances.

Belmont Report

Established three ethical principles for the foundation of how research should be conducted: Respect for Persons, Beneficence, Justice.

Tuskegee Study

African Americans made a part of a syphilis study where no one signed informed consent and the subjects were told they were being treated for 'bad blood.'

Institutional Review Board (IRB)

All research involving human subjects must be reviewed and approved by an IRB, which reviews data collection and storage, ensures informed consent is gained, and minimizes risk while explaining medical/social benefits of participation.

Sensitivity

Screening for a problem, identifies possible problems.

Specificity

Screening for a specific pathology, not just general screening of possible outcomes.

SPin and SNout

Specificity In = Ruling in the condition; Sensitivity Out = Ruling out a condition.

Formula for Sensitivity

True Positives / (True Positives + False Negatives). A high sensitivity means there are few false negatives.

Formula for Specificity

True Negatives / (True Negatives + False Positives). A high specificity means that there are few false positives.

Type 1 Error

False positive, rejecting the null when you shouldn't.

Type 2 Error

False negative, failing to accept the null when you should.

PICO question

P = Patient/Population/Problem, I = Intervention/Exposure/Test, C = Comparison, O = Outcome.

Cohort Studies

Following a group of similar individuals over time.

Cross-Sectional Studies

Measuring prevalence in a proportion of people with a condition at a specific time.

Experimental Study

Assignment of participants to specific exposures/treatments.

Indications for a particular study design

The reasoning for a specific study design is usually listed in the methods section of an article.

T0

Basic Research (Preclinical, Animal Trials)

T1

Human Translation (Proof of Concept, Phase 1 Trials)

T2

Patient Translation (Phase 2 & 3 Trials, Controlled Studies)

T3

Clinical Practice Translation (Effectiveness Research, Outcomes Research)

T4

Community Translation (Population-Level Studies, Policy Impact)

Null Hypothesis

No statistical difference or significance (Accepting the null occurs when P Value is greater than 0.05)

Alternative Hypothesis

Statistical difference or significance exists (Rejection of the null occurs when P Value is less than 0.05)

Selection Bias

Failing to achieve a truly randomized sample of the population

Sampling Bias

Cases not representing the entire population

Recall Bias

Different memories of a past event (self-surveying)

Confirmation Bias

Selectively focusing on information that confirms one's hypothesis

Recency Bias

Believing a particular presentation is based upon something you've seen recently

Independent Variable

Predicts or influences an outcome

Dependent Variable

Outcome that is influenced by changing the Independent Variable

Covariable

Variable that isn't the primary focus of a study but is considered because it influences the outcome of a study

Continuous Data

Any data that can be expressed along a scale or within a range including decimals and fractions. Infinite number of possible values.

Ratio Data

Data where values are measured on a scale with a true zero point, with equal intervals between the values.

Interval Data

Data where values are measured on a scale without a true zero point, with equal intervals between the values.

Probability Sampling

Use of random selection to minimize sampling error

Simple Random Sample

Each subject has an equal chance of being selected

Systemic Sample

Subjects are selected at regular intervals

Stratified Random Sample

Subgroups within the total subject population are represented equally

Cluster Sample

Subject population is geographically dispersed

Non-Probability Sampling

Not using random selection, used in clinical research due to high cost and time constraints

Convenience Sample

Recruiting easily available subjects

Purposive Sample

Selecting subjects who are deemed most appropriate

Snowball Sample

Subjects help to identify other eligible subjects

Consecutive Sample

Selecting every eligible subject during a specific time period

Mean

Average of data

Median

Middle value of data

Mode

Most frequently occurring value within data

Standard Deviation

Average distance of values within the data from the data's mean

Skewed Distributions

Asymmetrical data with a longer tail on one side of a bell curve.

P Value

If above 0.05 you fail to reject the null hypothesis, if below 0.05 you reject the null hypothesis

95% Confidence Interval

95% of scores fall within 2 standard deviations of the mean

Effect Size

A measure of the magnitude of an experimental effect. If there is no difference between two groups, effect size is zero

Correlation

Measurement of strength of the relationship between two variables. (0 to 0.25 is little relationship, 0.26 to 0.5 is fair relationship. 0.51 to 0.75 is good relationship, 0.76 to 1 is excellent relationship)

Positive Correlation

As one value increases, the other value does as well (positive impact on each other)

Negative Correlation

As one value increases, the other value decreases (negative impact on each other)

Reliability

How reproducible are the scores produced by a particular measure, how free of error is the measurement.

Validity

How accurately does the measure represent and capture what is meant to be analyzed. Tests must be reliable before they can be considered valid.

MDC (Minimal Detectable Change)

The smallest change that can be detected by a measure

Interpretation of forest plots

If any of the lines or the big diamond at the bottom of the graph cross the line in the middle, it means there is no statistical significance

Lines of Forest Plot represent what?

Separate studies being analyzed

The Squares in a Forest Plot represent what?

Point estimate of a study's 95% Confidence Interval

The Diamond in a Forest Plot represents what?

Combined Confidence Interval and Point Estimate of all studies being analyzed