Evidence-Based Practice in Nursing: Steps, Levels, Models

Evidence-based practice (EBP) in nursing means making care decisions by combining three things: the best available research evidence, the nurse's clinical expertise, and the patient's values and preferences. In practice it is a repeatable process: ask a focused clinical question, find the evidence, judge how trustworthy it is, apply it, and check whether it worked. This guide covers the five steps, how levels of evidence work, two widely used EBP models, and where to search.

What EBP is (and what it is not)

EBP is not "doing whatever the newest study says." Research evidence is one input. A strong randomized trial in young adults may not apply to a frail 90-year-old who has said she does not want aggressive treatment. EBP asks the nurse to weigh the evidence against the clinical situation and the patient's goals.

It also helps to separate EBP from two activities it is often confused with:

The lines blur in real projects. An EBP project usually includes a QI-style evaluation, and DNP projects often combine all three. Your institution decides whether a project needs IRB review, not the label you give it.

The five steps of EBP

The most common teaching version is the "five A's": ask, acquire, appraise, apply, assess. Some nursing models add a step zero (cultivate a spirit of inquiry) and a final step (disseminate the results), but the core is the same.

1. Ask: write a focused clinical question

Turn a practice concern into a searchable question. Nursing uses the PICOT format: Population, Intervention, Comparison, Outcome, Time. For example:

In adult inpatients with indwelling urinary catheters (P), how does a nurse-driven catheter removal protocol (I) compared with removal on physician order (C) affect CAUTI rates (O) over six months (T)?

The question type (intervention, etiology, diagnosis, prognosis, meaning) determines which study designs will give the best answer. Our PICOT question examples post has 15 worked questions and templates for each type.

2. Acquire: search for the best evidence

Search in order from the most filtered evidence to the least:

1. Clinical practice guidelines from professional bodies and national agencies. Someone has already synthesized and graded the evidence.
2. Systematic reviews and meta-analyses, for example in the Cochrane Library or the JBI Evidence Synthesis journal.
3. Individual studies in bibliographic databases: CINAHL (nursing and allied health), PubMed/MEDLINE (biomedicine), PsycINFO (mental health and behavior), and Embase (biomedicine and pharmacology, often available through institutions).

Build the search from your PICOT elements. Each element becomes a block of synonyms joined with OR, and the blocks are joined with AND. Use controlled vocabulary (MeSH in PubMed, CINAHL Subject Headings in CINAHL) alongside free-text words, because indexing lags behind new terminology. Apply limits such as language, publication date, and age group after you see the size of the result set, not before.

Keep a record of each database, search string, date searched, and number of hits. If your project ends up as a formal review, you will need these numbers for a flow diagram; our free PRISMA flow diagram generator builds one from your counts.

3. Appraise: judge the quality of what you found

Not every published study deserves the same weight. Appraisal asks three questions of each source:

• Is it valid? Was the design appropriate for the question, and were bias and confounding controlled?
• What are the results? How large is the effect, and how precise is it (look at confidence intervals, not just p-values)?
• Does it apply to my patients? Are the participants, setting, and intervention similar enough to yours?

Structured checklists make appraisal faster and more consistent. Commonly used options include the CASP checklists, the JBI critical appraisal tools, and AGREE II for clinical practice guidelines. Many guideline groups rate the certainty of a body of evidence using the GRADE approach (high, moderate, low, very low).

After appraising individual studies, synthesize across them. A table with one row per study (design, sample, intervention, outcome, main finding, level and quality) is the standard tool for this and becomes the evidence table in your paper or poster.

4. Apply: integrate evidence with expertise and patient preference

If the evidence is strong and consistent, plan a practice change. If it is weak or conflicting, the right decision may be to keep current practice, run a small pilot, or collect local data first.

Implementation is where most EBP projects stall. Plan for:

• Stakeholders: who needs to agree (nurse managers, physicians, pharmacy, infection control, patients)?
• Barriers: workload, equipment, documentation systems, habits.
• Education and tools: protocols, order sets, reminders, checklists.
• A pilot: start on one unit before rolling out house-wide.

5. Assess: evaluate the outcome

Measure whether the change did what you expected. Decide the outcome measure, the baseline period, and the follow-up period before you start. Typical designs are pre/post comparisons and run charts over time. Compare like with like: if the baseline was measured over three winter months, a follow-up over three summer months may show seasonal change rather than the effect of your intervention.

Then disseminate: share results with the unit, at a nursing grand rounds or conference, or in a journal. Even a change that did not work is useful information for other teams.

Levels of evidence

A levels-of-evidence hierarchy ranks study designs by how well they protect against bias for a given type of question. Several hierarchies exist, and they do not all number levels the same way. A widely taught nursing hierarchy for intervention questions (from Melnyk and Fineout-Overholt's EBP textbook) looks like this:

Three cautions apply when you use any hierarchy:

1. The hierarchy depends on the question. For an intervention question, an RCT sits near the top. For a meaning question about patient experience, a well-conducted qualitative study is the right design, and an RCT cannot answer it at all. For etiology and prognosis questions, cohort studies are often the best evidence available because randomizing people to a harmful exposure would be unethical.
2. Level is not quality. A badly run RCT can be less trustworthy than a careful cohort study. Always rate quality separately from level.
3. Use the hierarchy your program or model specifies. Instructors and journals expect you to cite the one you used.

If you want to understand the designs in the middle of the table, our posts on quasi-experimental design, case-control studies, and retrospective studies explain how each one works and where its biases come from.

Common EBP models in nursing

An EBP model gives a team a shared map of the process: who does what, in what order, and what decision points exist. Two of the most widely used in US hospitals are the Iowa Model and the Johns Hopkins model. The descriptions below are summaries; use the official model documents for an actual project, since both are periodically revised and require permission to reproduce the diagrams.

The Iowa Model

The Iowa Model was developed at the University of Iowa Hospitals and Clinics and has been revised several times (the most recent major revision was published in 2017 as the Iowa Model Revised). It is organized as a flow diagram with decision points. In outline:

1. Identify a trigger: a clinical problem, new evidence, a guideline, or an organizational priority.
2. State the purpose and ask: is this topic a priority? If not, consider another trigger.
3. Form a team of relevant stakeholders.
4. Assemble, appraise, and synthesize the body of evidence.
5. Decide: is there sufficient evidence? If not, the team may conduct research or gather more evidence.
6. Design and pilot the practice change.
7. Decide: is the change appropriate for adoption in practice? If not, reconsider the approach.
8. Integrate and sustain the change, then disseminate the results.

The model's strength is its explicit decision points, which make it clear that "not enough evidence" and "not appropriate to adopt" are legitimate outcomes.

The Johns Hopkins Evidence-Based Practice Model

The Johns Hopkins model (developed at Johns Hopkins Hospital and the Johns Hopkins School of Nursing) is organized around a three-phase process known as PET:

• Practice question: recruit a team, define the problem, write a PICO question, and define the scope.
• Evidence: search, appraise each source for level and quality, summarize, and synthesize the overall body of evidence.
• Translation: decide whether the evidence supports a change, create an action plan, implement, evaluate outcomes, and communicate the findings.

The model comes with its own evidence level and quality rating scheme, which places research evidence (experimental, quasi-experimental, non-experimental and qualitative) above non-research evidence such as clinical guidelines, expert opinion, quality improvement reports, and case reports, and grades each source for quality. It also provides practical tools such as appraisal forms and synthesis templates, which is one reason it is popular in student projects.

Other models you may meet

You may also see the Stetler Model, the ACE Star Model of Knowledge Transformation, and the ARCC (Advancing Research and Clinical practice through close Collaboration) Model. Pick the model your organization or program already uses; consistency across a team matters more than which model you choose.

Writing up an EBP project

Most EBP write-ups follow a structure close to this:

1. Background and problem: local data showing the problem and why it matters.
2. PICOT question.
3. Search strategy: databases, terms, limits, dates, number of results.
4. Evidence synthesis: an evidence table and a narrative summary organized by theme, with level and quality ratings.
5. Recommendation: what the evidence supports and with what strength.
6. Implementation plan: setting, stakeholders, steps, timeline.
7. Evaluation plan or results: outcome measures and how they changed.
8. Limitations and next steps.

If your EBP work grows into a formal doctoral project with its own data collection and evaluation, see our DNP project guide for the extra pieces: a theoretical framework, a measurement plan, and statistical analysis.

Using AI tools for the evidence step

AI tools can speed up the search and synthesis steps, but they cannot replace appraisal. The two risks are invented references and confident summaries of studies that do not say what the summary claims. Whatever tool you use, open each source, check the DOI, and read at least the abstract and methods before you rate it.

Data2Paper's Research Report is built for the acquire-and-synthesize step: you enter a topic or PICOT question and receive a structured literature review drawn from real papers in OpenAlex and Crossref, each with a DOI you can verify, plus a references.bib file for your reference manager. You can generate the report and read a free preview of the first half, then unlock the full report for a one-time $9.90 if it is useful. Treat it as a starting draft for your own appraisal, not as a finished evidence synthesis.

FAQ

What are the five steps of evidence-based practice?

Ask a focused clinical question, acquire the best evidence, appraise it critically, apply it together with clinical expertise and patient preferences, and assess the outcome. Many nursing models add a preliminary step (cultivating a spirit of inquiry) and a final step (disseminating results).

What is the highest level of evidence?

For intervention questions, a systematic review or meta-analysis of well-conducted randomized controlled trials is usually placed at the top. For questions about meaning or experience, systematic reviews of qualitative studies are the strongest evidence. Level always needs to be read together with quality.

What is the difference between EBP and research?

Research generates new knowledge that can be generalized. EBP takes existing knowledge and applies it to care decisions in a particular setting. A nurse doing EBP does not need to collect new data, although evaluating the practice change usually involves some local measurement.

Which EBP model should I use?

Use the model your hospital or program already uses. The Iowa Model suits teams that want explicit decision points about priority and sufficiency of evidence; the Johns Hopkins model suits teams that want ready-made appraisal and synthesis tools. Both cover the same core steps.

Which databases are best for nursing evidence?

CINAHL for nursing and allied health, PubMed/MEDLINE for biomedical literature, and the Cochrane Library for systematic reviews. Add PsycINFO for mental health topics and check professional guideline sources before searching for individual studies.