DEVELOPING AN EVIDENCE-BASED PRACTICE

profileSc hool10
NRSE_4550_M5_LA_Tools_for_Developing_an_EBPmodule5pp.ppt

Chapter 19

Strategies and Tools for Developing an Evidence-Based Practice

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Tool #1: Asking a Focused
Clinical Question

Develop the question by addressing these four issues:

Population

Intervention

Comparison

Outcome

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Develop the question by answering these four questions:

  • What is the population I am interested in?
  • What is the intervention I am interested in?
  • What will this intervention be compared to? (Note: Depending on the study design, this step may or may not apply.)
  • How will I know if the intervention makes things better or worse (identify an outcome that is measurable)?

Categorize the Clinical Question

Therapy category:

Experimental or quasi-experimental

Outcome known

Therapy appraisal tool at:

http://www.casp-uk.net/wp-content/uploads/2011/11/CASP_RCT_Appraisal_Checklist_14oct10.pdf.

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

1. Therapy category: If the question is about the effectiveness of a particular treatment or intervention, select studies with experimental or quasi-experimental study design.

Use a therapy appraisal tool to evaluate the article. A tool can be accessed at:

http://www.phru.nhs.uk/Pages/PHD/resources.htm

Outcome known or of probable clinical importance observed over a clinically significant period of time.

Categorize the Clinical Question

Diagnosis category:

Cross-sectional

Comparison of the new and the “gold standard”

Diagnostic tool at:

http://www.casp-uk.net/wp-content/uploads/2011/11/CASP_Diagnostic_Appraisal_Checklist_14oct10.pdf.

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

2. Diagnosis category: If the question is about the usefulness, accuracy, selection, or interpretation of a measurement instrument or laboratory test, select studies with:

  • Cross-sectional study design (see Chapter 9) with people suspected to have the condition of interest
  • Use a diagnostic test appraisal tool to evaluate the article. A diagnostic tool can be accessed at

http://www.phru.nhs.uk/Pages/PHD/resources.htm

Comparison of the results of the new instrument or test and the “gold standard”

Categorize the Clinical Question

Prognosis category:

Nonexperimental

Follow-up

Determination of factors

Prognosis tool at:

http://www.casp-uk.net/wp-content/uploads/2011/11/CASP_Cohort_Appraisal_Checklist_14oct10.pdf

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

3. Prognosis category: If the question is about a patient’s likely course for a particular disease state or to identify factors that may alter the patient’s prognosis, select studies with:

  • Nonexperimental, usually longitudinal study of a particular group for a particular outcome or disease
  • Follow-up for a clinically relevant period of time (time is the exposure)
  • Determination of factors in those who do and do not develop a particular outcome

Use a prognosis appraisal tool to evaluate the article. A prognosis tool can be accessed at

http://www.phru.nhs.uk/Pages/PHD/resources.htm

Categorize the Clinical Question

Harm category:

Nonexperimental

Exposure

Harm appraisal tool at:

http://www.casp-uk.net/wp-content/uploads/2011/11/CASP_Case-Control_Appraisal_Checklist_14oct10.pdf.

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

4. Etiology-causation-harm category: If the question concerns determining whether or not one thing is related to or caused by another, select studies with:

  • Nonexperimental, usually longitudinal or retrospective (ex post facto) study designs over a clinically relevant period of time
  • Assessment of whether or not the patient has been exposed to the independent variable
  • Use a harm appraisal tool to evaluate the article. A harm appraisal tool can be accessed at

http://www.phru.nhs.uk/Pages/PHD/resources.htm

Tool #2: Searching the Literature

Consult librarian

Tutorial for PubMed at www.nlm.nih.gov/bsd/disted/pubmed.html#qt

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Consult your academic research librarian for assistance.

You can also learn how to effectively search PubMed through a Web-based tutorial located at http://www.nlm.nih.gov/bsd/disted/pubmed.html#qt.

Tool #3: Screening Your Findings

Peer-reviewed journal?

Similar setting and sample?

Study sponsorship?

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

  • Is this article from a peer-reviewed journal? Articles published in a peer-reviewed journal have had an extensive review and editing process
  • Are the setting and sample of the study similar to mine so that results, if valid, would apply to my practice or to my patient population?
  • Is the study sponsored by an organization that might influence the study or the design or results?

Tool #4: Appraise Each Article’s Findings

Therapy studies

Is there a difference between two or more treatments?

Numerical values are either continuous or discrete.

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Therapy studies

  • Attempt to determine whether a difference exists between two or more treatments.
  • Numeric values in these studies are either continuous or discrete.

Continuous and Discrete Variables

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Objective Variable Outcome
CONTINUOUS VARIABLES
Change after exposure to intervention Pain score Levels of distress Blood pressure Weight Measures of central tendency
DISCRETE VARIABLES
“Event” occurred or did not occur Death Diarrhea Pressure ulcer Pregnancy Measures of event probability

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

CONTINUOUS VARIABLES

  • Researcher objective: Researcher is interested in degree of change after exposure to an intervention.
  • Examples of variables: Pain score, levels of distress, blood pressure, weight
  • How the outcome is described: Measures of central tendency (e.g., mean, median, or standard deviation)

DISCRETE VARIABLES

  • Researcher objective: Researcher is interested in whether or not an “event” occurred or did not occur.
  • Examples of variables: Death, diarrhea, pressure ulcer, pregnancy.
  • How the outcome is described: Measures of event probability (e.g., relative risk, odds ratio or number needed to treat)

Diagnosis Articles

Sensitivity is the proportion of those with disease who test positive; that is, sensitivity is a measure of how well the test detects disease when it is really there—a highly sensitive test has few false negatives.

Specificity is the proportion of those without disease who test negative. It measures how well the test rules out disease when it is really absent; a specific test has few false positives.

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

  • If a trial were repeated many times, results would be expected to differ slightly from trial to trial. However, results would also be expected to fall within a certain range that would lie around the “true” result.
  • In reality, trials are usually done only once. The confidence interval (95% CI is most widely used) is an estimate of the range within which the “true” result is likely to fall.

Sensitivity

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Measure of Accuracy Definition Comments
Sensitivity Ability of the test to detect the proportion of people with the disease or disorder of interest TP/(TP + FN), where TP and FN are number of true-positive and false-negative results, respectively

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Sensitivity: A characteristic of a diagnostic test. It is the ability of the test to detect the proportion of people with the disease or disorder of interest. For a test to be useful in ruling out a disease, it must have a high sensitivity.

Specificity

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Measure of Accuracy Definition Comments
Specificity Ability of the test to detect the proportion of people without the disease or disorder TN/(TN + FP), where TN and FP are number of true-negative and false-positive results, respectively

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Specificity: A characteristic of a diagnostic test. It is the ability of the test to detect the proportion of people without the disease or disorder of interest. For a test to be useful at confirming a disease, it must have a high specificity. Formula for sensitivity: TN/(TN + FP), where TN and FP are number of true-negative and false-positive results, respectively.

Prognosis Articles

Odds ratio: Probability of developing the outcome or a particular disease.

Indicates how much more likely certain independent variables (factors) predict the probability of developing the dependent variable (outcome or disease).

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

  • If a trial were repeated many times, results would be expected to differ slightly from trial to trial. However, results would also be expected to fall within a certain range that would lie around the “true” result.
  • In reality, trials are usually done only once. The confidence interval (95% CI is most widely used) is an estimate of the range within which the “true” result is likely to fall.

Odds Ratio

Odds ratio

The odds ratio (OR) best describes the data in case-control studies.

The OR = probability of an event

Calculated by dividing the odds in the treated or exposed group by the odds in the control group

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Harm studies:

  • The odds ratio (OR) is the measure of association that best describes the analyzed data in case-control studies.
  • The OR communicates the probability of an event.
  • OR is calculated by dividing the odds in the treated or exposed group by the odds in the control group.

Harm Articles

Used to determine if an individual has been harmed by being exposed to a particular event

Case-control design: investigators select the outcome they are interested in (e.g., pressure ulcers), and examine if any one factor explains those who have and do not have the outcome of interest.

The measure of association that best describes the analyzed data in case-control studies is the odds ratio.

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

  • If a trial were repeated many times, results would be expected to differ slightly from trial to trial. However, results would also be expected to fall within a certain range that would lie around the “true” result.
  • In reality, trials are usually done only once. The confidence interval (95% CI is most widely used) is an estimate of the range within which the “true” result is likely to fall.

Meta-analysis

A research method that statistically combines the results of multiple studies (usually randomized clinical trials) to answer a focused clinical question through an objective appraisal of carefully synthesized research evidence

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Meta-analysis

Terms “meta-analysis,” “systematic review,” and “integrative review” are used interchangeably.

Meta-analysis is a quantitative approach to a systematic review, whereas an integrative review uses a nonquantitative approach.

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Meta-analysis

Meta-analyses and integrative reviews are both considered systematic reviews and provide level I evidence.

Each uses a standardized process, which has a set of preestablished criteria that guide its implementation.

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Steps in a Meta-analysis

Clinical question

Search for all relevant studies

What studies are included

Assess the quality of each study

Studies statistically combined

Conclusion

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

  • A clinical question is formulated.
  • A team of at least two investigators search for all relevant studies, published and unpublished, on the topic or question
  • Using pre-established inclusion and exclusion criteria team determines the studies that will be used in the meta-analysis
  • At least two individuals independently assess the quality of each study, include or exclude studies based on preestablished criteria
  • The results of individual studies are statistically combined
  • A balanced and impartial quantitative and narrative evidence summary is prepared representing a “state of the science” conclusion about the strength, quality, and consistency of evidence

Measures in Meta-analysis Systematic Review

The OR is the statistic of choice for use in a meta-analysis.

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Interpreting Odds Ratios

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Odds ratio Type of Outcome
Adverse outcome Beneficial outcome
Less than 1 Intervention better Intervention worse
= to 1 (null) Intervention no better or worse Intervention no better or worse
Greater than 1 Intervention worse Intervention better

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

What is the most important step in applying evidence to practice?

PICO (population, intervention, comparison, and outcome)

Critically reviewing the literature

Putting the research into practice

Evaluating efficacy of the new practice

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

ANSWER: A

RATIONALE: All are important, but a focused clinical question using PICO (population, intervention, comparison, and outcome) lays the foundation.

*

Which EBP clinical category would the nurse be using in the following scenario?

A clinical nurse has noticed an increased incidence in urinary tract infections and would like to find the best practice to minimize this.

Therapy

Diagnosis

Prognosis

Harm

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

ANSWER: D

RATIONALE: Harm; the nurse wants to determine whether or not one thing is related to or caused by another.

*

When completing a literature search for a clinical question, what is the first resource the nurse should use?

Evidence-based nursing

CINAHL

PUBMED

Cochrane review

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

ANSWER: A

RATIONALE: Evidence-based nursing is a prefiltered source and is available online at http://ebn.bmjjournals.com/ and in print.

*

For the nurse who has limited time to review a research article, where would be the best place to find the answer to the clinical question?

Abstract

Method

Conclusion

Table

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

*

Copyright © 2014, 2010, 2006, 2002, 1998, 1994, 1990, 1986 by Mosby, an imprint of Elsevier Inc.

ANSWER: D

RATIONALE: When evaluating whether or not you should spend time reviewing an article, examine the article’s tables. The information you need to answer your clinical question should be contained in one or more of the tables.

*