EBP Final Exam

Evidence-Based Practice (EBP)

• EBP combines scientific research, clinical expertise, and the patient's values to inform clinical decisions.
• EBP aims to provide optimal care and maximize patient benefits in therapy.
• Key vocabulary includes scientific research, clinical expertise, patient values, and shared informed decisions.

Key Vocabulary in EBP

• Scientific Research: Empirical evidence through systematic testing of a hypothesis
• Clinical research involves human subjects, while nonclinical research uses animals, cells, etc.
• Clinical Expertise: Gained from years of patient care, shared formally through education/mentorship and informally among colleagues.
• Patient Values and Circumstances: Patient's beliefs, preferences, expectations, cultural background, medical history, access to care, and family.
• Shared Informed Decision: Choice made jointly by therapist and patient based on best evidence.

EBP Process Steps

• Identify a question: Gather info from the patient create focused, searchable clinical questions.
• Search for evidence: Use databases to find research answering the questions.
• Appraise the evidence: Check if research is relevant.
• Integrate the evidence: Combine data, expertise, and values.
• Evaluate the outcomes: Check the process's effectiveness, find areas for improvement.

Barriers to EBP

• Time constraints limit research review, and results may not apply to specific individuals.
• Lack of research and statistical skill inhibits finding/evaluating data.
• Insufficient information + EBP culture hinders utilization.

Importance of EBP

• Enhances confidence and shared decision-making, moving from habit-based to systematic.
• Patients want evidence-based care.

Applying EBP Considerations

• Best evidence may not suit every patient because of unique factors.
• Look at rationale, advantages, disadvantages with limited evidence.
• Strive for optimal decisions via information + access to resources
• Assess the worth of clinical skills and prioritize patient individuality in decisions

5-Step EBP Model

• Designed to create structured EBP approach.
• Step 1: Find info, make clinical question by ID patient
• Step 2: Search to find best evidence
• Step 3: Appraise the research for validity
• Step 4: Integrate with expertise/circumstances in team
• Step 5: Access success and look for future improvements

Three Pillars of Evidence for EBP

• Optimal outcomes supported by: Scientific research, clinical expertise, patient values.
• Scientific Research: Empirical evidence of human subjects + nonclinical species.
• Clinical Expertise: Implies knowledge with caring and includes therapist values.
• Patient Values and Circumstances
• Each supports therapist and patient work for health.

EBP Sources of Evidence

• The three sources come from: Scientific research, clinical expertise, patient values
• In the evidence pyramid, SRs are best
• Below SRSs come individual studies like RCTs, and cohort studies
• Secondary examples contain guidelines used to directly advise practice.

Defining Types of Research

• Initial research step involves identifying the general type: human subjects, animal, or non-human subjects research to determine proper compliance.

Institutional Review Board (IRB)

• IRB approval is mandatory for human subject research before commencement, including study publication or presentation.
• Researchers submit detailed applications to the IRB covering design, participant criteria, measurements, and potential risks/benefits.
• A vital IRB component is informed consent, ensuring participant understanding and agreement before enrollment.
• IRBs consist of experts like clinicians, researchers, ethicists, statisticians, and legal representatives to ensure ethical research conduct.
• IRBs minimize risks, protect confidentiality, and confirm studies have broad purpose.

Human Subjects Research Defined

• Research involves living individuals, their data, or biospecimens, obtained through interaction or intervention, or when private, identifiable information is analyzed.
• Activities include surveys, focus groups, medicine administration, environmental changes, and educational interventions.
• IRB exemptions exist, such as some education studies using standard practices and research using de-identified, publicly accessible data, but require IRB approval for exemption.
• Case studies, series are generally not classified as human subjects research and are not under IRB authority since they describe standard or innovative care rather than testing scientific hypotheses.

Examples of IRB Evaluation

• Simple studies undergo IRB review for proper procedures, confidentiality, and outcome metrics, with potential power analysis improvements to improve the study's validity.
• Complex research raises concerns about consent, risk vs. benefit, often involving ethicists and specialists to eliminate risk without removing the study.

IRB Decisions and Approvals

• IRBs decide to request information, require modifications, approve (with modifications if necessary) or reject studies.
• Formal IRB approval is required before starting research, and given in a letter with a specified approval period (1-3 years).

Informed Consent Process Details

• Participant recruitment, and informed consent begins post-IRB approval
• Participants must be capable of giving consent, or a legal guardian/family member must provide it.
• Participants can withdraw at any time without penalty.

IRB Rigor Variability

• IRB rigor varies between and within institutions because of human factors.

Pre-Registration of Research

• Study pre-registration on public websites (e.g., clinicaltrials.gov) is required (especially for federally funded, clinical trials, and systematic reviews.
• Pre-registration is declaring the protocols, designs, participants, procedures, and primary outcomes prior to data collection.
• Study design is providing public information and enhancing research integrity.

Animal Model Research Details

• Animal research uses animals (often sacrificed at the end) to benefit human health that have considerable ethical implications.
• Animal models enable lifespan studies or interventions difficult/unethical in humans.
• It can offer insights where human trials are unavailable or flawed.
• Translational medicine faces challenges in applying animal findings to humans due to physiological differences.

Patient Intake Medical Screening Forms

• An example of a form used in clinical settings.
• It gathers medical history, symptoms, medications, and learning barriers.
• It is for clinical practice, not data collection under an IRB protocol, but shows types of relevant health data.

Research Dissemination

• Involves sharing findings with the scientific community and the public after data collection and analysis.
• It is the final stage in the research process.

Conference Presentations

• Conferences offer poster or podium presentations.
• Submission of an abstract is required.
• Abstracts are compiled into conference proceedings.
• Podium presentations are oral, while poster are more interactive.

Peer-Reviewed Papers

• Publishing in peer-reviewed journals involves scrutiny by field experts and is considered the research process ultimate goal.
• Papers are based on journal scope, readership, and impact factor.
• Impact factor reflects citation frequency with controversies.
• Journal acceptance rate is key, with rejections requiring re-submission to multiple journals.
• Editors either send the manuscript for peer-review or reject.
• Editors invite multiple experts.
• Reviewer recommendations range from accept to reject, sometimes asking for revisions.
• Response and indication is needed for feedback and addressed in the manuscript.
• Peer review is prone to errors, and bias' while acting as a control mechanism.

Pre-Prints

• Pre-print servers allow posting work before peer-review for rapid dissemination while lacking the quality control.
• Viewers must critically evaluate pre-print research.

Predatory Journals

• Predatory journals are unethical and accept articles for a fee, irrespective of quality.
• Detecting journals is possible through PubMed.

Study Appraisal Key Steps

• Apply study to patient population analysis.
• Determine study and result intrepretation,
• Summarize the clinical bottom line.

Research Conclusion

• The research process is complex, with ethical considerations, dissemination of findings
• Critical thinking remains important.
• It empowers clinicians and better applies findings in practice.

Conflicts of Interest

• Arises when someone in a study has a personal factor potentially biasing design, conduct, analysis, interpretation, or discussion.
• Financial conflicts include inventors researching their products, as positive results benefit them.
• Funding from an interested party, even for independent researchers, can create conflicts.
• Disclosure is critical, allowing readers to assess potential bias without implying wrongdoing.
• The best practice is complete removal of the interested party from the research process.

Reporting Bias

• Occurs when researchers selectively report positive findings, downplaying or omitting negative or null results.
• Motivation: Genuine belief in intervention, wanting it to succeed, or concerns about non-significant results.
• Widespread reporting bias skews systematic reviews and meta-analyses.
• Lack of full transparency in reporting outcomes is the core issue.

Outcome Switching

• Changing a study's pre-defined primary outcome after results are known, highlighting significant secondary findings.
• Illustrated by the Q collar example, shifting focus from concussion prevention to "brain protection" based on MRI evidence.
• Pre-registration on platforms like ClinicalTrials.gov helps identify outcome switching.
• Justified switching is acceptable with explicit reasons, like new evidence, but ignoring the original primary outcome is problematic.
• Outcome switching is a common issue.

P-Hacking

• Conducting multiple statistical analyses and selectively reporting those with significant p-values (< 0.05) without disclosing others.
• Examples: excluding participants, analyzing subgroups, or trying different statistical tests for significance.
• Can lead to publication of spurious, non-replicable findings.

HARKing (Hypothesizing After Results Known)

• Formulating a hypothesis based on results and presenting it as the original.
• Leads to misleading interpretations of findings.
• Contributes to unsupported explanations, or "scientific just-so stories."

Fabrication

• The most severe misconduct, involving creating false data/results and reporting them as genuine.
• Ranges from altering data points to inventing datasets or claiming studies never occurred.
• Detection is challenging and often requires specialized forensic skills.

Image Manipulation

• A common method, especially in basic science, involves manipulating or duplicating images.
• Data sleuths often uncover manipulation by recognizing identical images.

Data Anomalies

• Unusual patterns/inconsistencies in reported data indicate potential fabrication or error.
• Examples: Mismatches between text and tables, unexpectedly high frequencies of numbers ending in the same digits, or statistically impossible values.

Combatting Research Misconduct

• Contacting the study authors is the technically correct first step, though often ineffective.
• Contacting the journal editor-in-chief is a more common, effective step, outlining concerns and evidence.

Journal Investigations and Responses

• Responses vary from ignoring the issue to publishing a letter to the editor.
• Formal investigations may involve forensic teams examining data.
• Journals may issue correction statements rather than retractions.
• Platforms such as Pub Peer offer a public forum for raising concerns when journals are unresponsive.

Presumption of Innocence

• Journals often assume innocence, even if one paper is retracted for fraud, not investigating other papers from the same group without clear evidence.
• This practice can be problematic if subsequent publications are also fraudulent.

Disciplinary Action

• Consequences of confirmed misconduct range from warnings to job loss, and in severe cases, criminal charges.

Implications of Research Fraud

• Has real-world implications, including wasted resources and misdirection of research efforts.
• Fraudulent research, like in Alzheimer's disease, can hold back entire fields for decades and negatively impact patient outcomes.
• Vigilance and critical evaluation are crucial despite the need for trust in the scientific process.

Research Integrity as a Systemic Issue

• The scientific publishing system has strengths and weaknesses.
• Understanding research integrity challenges provides crucial context for interpreting findings.

Key Concepts & Vocabulary

• Searchable Clinical Question: Structured foreground question to efficiently find research.
• Patient characteristics, patient management, and outcomes in questions.
• Background Questions: General info questions from textbooks.
• Foreground Questions: Specific questions from research.
• Boolean Search: Method using AND, OR, NOT to refine search terms.
• AND narrows, OR broadens search terms.
• Keywords: are from searchable clinical research

MeSH

• Medical Subject Headings indexes articles
• Controlled vocabulary used in MEDLINE
• PubMed to help find the best MeSH keyword

Database

• Search Engine is the user interface
• Database is the collection of articles and guidelines
• MEDLINE is the biomedical research database via PubMed

Filters

• PubMed has pre-defined filters in Clinical Queries.
• My NCBI saves searches, etc.
• Google Scholar shows cited literature.

Research Acronym

• PICO helps structure intervention
• Population/Patient
• Intervention
• Comparison
• Outcome

Databases & Search Engines Details

• Google Scholar: Scholarly literature, helpful search tools/Cited by feature.
• PubMed: Freely available via MEDLINE.
• narrows using MeSH terms
• assesses risk of bias

Databases

• TRIP Database: uses AI, types of research
• PEDro: Database specific to physical therapy
• APTA: Clinical practice guidelines, summaries
• AHRQ: Clinical practice guidelines not indexed
• Cochrane Library: Systematic reviews

Types of Research

• Systematic Review: Synthesized multiple primary studies
• Randomized Controlled Trial RCT: random assignment
• Clinical Practice Guidelines CPGs: expert opinions
• Descriptive Research: describes using designs
• Predictive Research: Aims to determine measures
• Causal Research: determines intervention change

Additional Concepts

• Efficiency: Quickly locating information
• Full Text: Complete version of research
• Tools to narrow by filters
• Synonyms: Different word usage
• Impairments: Body function issues
• Activity limitations: Action issues
• Participation restrictions: Community

Steps for Searching

• ID need, ask question
• List keywords, synonyms
• Reorder importance
• Terms by importance
• Individually into search
• Combine or
• Combine and
• Use filters, save

Key Concepts in Research Applicability & Quality

• Applicability is research is relevant.
• "Close Enough" matches are sometimes ok.
• Studies must be accurate with validity.

Study Design and Bias

• Randomization stops bias
• Stratification and subgroups.
• Blinding minizmies patient bias
• Sham groups act like an intervention but are fake
• Placebo affects patients

Analysis

• intention to treat analysis analyzes participant even with attrition
• A good study prevents dropouts
• Study must follow groups with study
• No standard intervention with group
• Comparison with the new interventions

Study Factors

• Interventions should become realistic
• Outcomes are relevant
• Measures are surrogate, and clinical outcome
• It is necessary to follow-up duration

Sampling Methods

• Consecutive Samples follow patients with problem
• Selective doesn't always
• Size dictates power

Measurement

• Construct means interest.
• Scales can be subjectiv
• Patient reports are subjective
• Observer reports capacity

Other Key Terms

• Bias hinders study.
• Conflict matters.
• Vocabulary helps practice

Different Types of Bias

Selection Bias

• When participants aren't randomised assign.
• To reduce
• Assign computerized

Performance Bias

• When the treatments were different.
• To reduce
• Must be equivalent

Detection Bias

• When test results for biased
• To reduce
• testers blind

Attrition Bias

• Loss of participants
• To reduce
• ITT

Reporting

• Results are off
• To reduce
• Declare conflicts

Recall Bias

• Memory isn't reliable
• To reduce
• Must be consistent

Sampling

• Small samples biased
• To reduce
• Must consecutively apply

Placebo

• Mind affect
• Include group.

Additional points on bias

• Random reduces bias
• Blinding reduces placebo
• Smaller samples is bad.
• Both measures susceptible

Key Points on Data, Types

• Nominal: Categories, groups are coded
• Ordinal: Meaningful ranking
• Discrete: Specific limits
• Continuous: Any values in some range

Details on Types of Statistics

• Central tendency in descriptive
• Mean average
• Median midpoints
• Mode occurs the most
• Variability is disperse

Statistic Tools

• Ranges determine low in dataset
• SD variance
• Distributions skew data.
• It is useful to compare results in similar studies.

Tests of Statistics

• Need to assume hypothesis.
• Probability is p-value
• T-tests and ANOVAs.
• Squares nominalize data

Clinical Relevance

• Size is difference
• NN need for dichotomous
• MC difference measured

Study Design Factors

• Blinding done
• ITT analysis is needed.
• Follow data dictates statistics.
• Importance values apply

Terminology Table for Concepts in Statistics

• Definitions are provided for: nominal data, continuous data and ordinal data
• Explanations around basic descriptive vs inferential applications
• The meanings of "the mean" "the mode" "the median"
• Understanding range, standard deviation, reliability
• Breaking down basic analysis and relationships
• Types for errors and alpha, intervals and treatment needs etc

Stats Info Added

• Definitions, concepts like descriptive statistics and interferential approaches are clarified
• Types of errors discussed
• Test types defined, all explained
• Power detailed in all designs

Key Statistical Concepts

  Descriptive Statistics: is used to describe data, not to make comparisons.
     ◦    Mean: The average of a normally distributed data set.
     ◦    Median: The halfway point of a data set, used when data is skewed.
     ◦    Standard Deviation: is used to measure the variability of data in a normally distributed data set.
     ◦    Interquartile Range (IQR): Equivalent to standard deviation, but used for non-normally distributed data.
     ◦    Confidence Intervals: A range of values that you can be 95% is confident to contain the true population mean.
  Inferential Statistics: Used to make inferences about a larger population based on a sample.
     ◦    Null Hypothesis Significance Testing: A process to determine if there are differences between groups,
            ▪    Null Hypothesis: The assumption that there is no difference between groups.
            ▪    Alternative Hypothesis: The hypothesis that there is a difference between groups.
            ▪    P-Value: The probability of obtaining the study results if the null hypothesis were true. A p-value of less than or equal to 0.05 is often
                  set as the threshold for statistical significance.
            ▪    Statistical Significance: When a p-value is less than a set threshold (often 0.05), indicating that the results are unlikely due to chance,
                  but not necessarily clinically meaningful.
     ◦    Type I Error: a false positive, rejecting the null hypothesis when it is true.
     ◦    Type II Error: A false negative, accepting the null hypothesis when it is false. Often due to an underpowered study (not enough participants).

Study Design & Analysi

   Sample vs. Population: A sample is a subset of a larger population, intended to be representative of that population.

   Randomization: Randomly assigns participating in treatment groups to ensure baseline .characteristics are similar with.

   Baseline Characteristics: The characteristics of participants at the start of a study and these should be similar between groups.

   Power Analysis: Used to determine the number of participants
   Pilot Studies: Small scale studies that provide preliminar data 

   T-tests: Statistical tests used to compare the meant

           sPaier
   ANOVA (Analysis of Variance): Statistical test is used to compare from multiple groups,             c o n tin u o u 5

   Correlation: A measures the relationship between and continue

   Linear Regression: A state-run
   Logistic Regression: A statistical data on predictor variables

     ◦   Binary Logistic Regression: Used to predict
     ◦   Multinomial Logistic Regression: Use

Interpreting Rest Change: The difference in a score on an outcome designed

Key Stats

   Cronbachs 

Main Term

• A tool is an assessment .

Psychometrics

• Validity and reliability

Types of Measurement

Reliability is consistent

• Inter rater differs
• Intake consistency
• Test retest on other change

Scales

• The best tools

Types of Validly

• validity
• Face validity
• Content based
• Criterion
• Criterion
• Constructs are clear,

Change and Outcome of Statistics

• Minimal detectable

Statistics

A

• Change refers to a variation in a patient The
• There a terms are with minimum

Points

• Valid measures can't change overtime

• Alpha is high

• DEXA 8,

Terms and Explanations

• Measure, reliable, inter and intra
• Consistent measures
• Validity and consistency
• Measure and analysis