NSG/509 Week 1: Research, EBP, and QI Exercise, sample paper

Research, Evidence-Based Practice, or Quality Improvement? Classifying Three Hospital Projects and the Data Each One Uses

[Student Name]

University of Phoenix

NSG/509: Research and Applied Statistics for Quality Improvement

Week 1 Exercise

[Instructor Name]

[Date]

The hospital and projects are a composite written for a model paper.

Nurses in a composite 350-bed hospital proposed three projects in the same month, and the nursing research council had to decide how each should be reviewed. The first asked whether a new mattress reduced pressure injuries compared with the standard mattress, using randomization of patients on four units. The second asked whether the hospital should adopt a published protocol for oral care in ventilated patients. The third aimed to reduce the time from emergency department arrival to first provider contact. All three would use data, involve patients and aim to improve care, yet each belonged to a different category with its own rules, methods and review. This exercise classifies the three projects, identifies the variables each uses and explains why the distinctions matter for the work that follows.

Criteria for Classification

Shirey et al. (2011) compared quality improvement, evidence-based practice and research across purpose, starting point, methods, generalizability and oversight. Research generates new knowledge using systematic methods designed to produce findings that can be generalized beyond the setting, and studies involving human subjects require institutional review board review. Evidence-based practice translates existing best evidence into practice, combining it with the judgment of clinicians and the wishes of patients; it starts from a clinical question and a search of the literature rather than from a research hypothesis, and it moves through a defined sequence of asking, searching, appraising, integrating and evaluating (Melnyk & Fineout-Overholt, 2023). Quality improvement uses data to improve local processes and outcomes, typically through rapid cycles, with results intended mainly for the setting where the work occurs.

Classifying the Three Projects

Project 1, the mattress trial, is research. It tests a hypothesis about an intervention by randomly assigning patients to conditions to generate knowledge that could apply elsewhere. Randomization, a comparison group and an intent to publish findings about mattress effectiveness are features of research. It requires institutional review board review and, depending on risk, patient consent.

Project 2, the oral care protocol, is evidence-based practice. It starts with a clinical question about preventing ventilator-associated events, searches existing evidence and guidelines, appraises them and plans to implement the best practice locally. It does not create new knowledge; it applies existing knowledge. Its outcomes would be monitored, often with quality improvement methods, but the project's purpose is translation.

Project 3, the door-to-provider project, is quality improvement. It addresses a local process problem using the hospital's own data, tests changes in rapid cycles and aims to improve performance in this emergency department. Its findings are not intended to be generalized, and it would usually follow the hospital's quality improvement review process rather than full research review, unless it later added features such as randomization or an aim to publish generalizable conclusions.

Variables and Levels of Measurement

Polit and Beck (2021) describe four levels of measurement, and the level of each variable determines which statistics are appropriate. The three projects use the following variables.

Mattress trial. Mattress type (standard or new) is nominal: categories with no order. Pressure injury stage is ordinal: stages have a meaningful order, but the difference between stage 1 and stage 2 is not equal to the difference between stage 3 and stage 4. The Braden score is ordinal in its construction, though it is often treated as interval in analysis, a choice that should be noted as an assumption. Length of stay in days is ratio: it has equal intervals and a true zero.

Oral care protocol. Whether a patient developed a ventilator-associated event is nominal, coded yes or no. Days on the ventilator is ratio. Staff adherence to the protocol, measured as the percentage of required oral care episodes documented, is ratio.

Door-to-provider project. Arrival-to-provider time in minutes is ratio. Triage acuity level on the five-level emergency severity index is ordinal. Patient temperature in degrees Fahrenheit, recorded at triage, is interval: equal intervals but no true zero, since zero degrees does not mean absence of temperature. Arrival shift (day, evening or night) is nominal, although it has a time order; treating it as nominal avoids assuming equal spacing between shifts.

When Quality Improvement Becomes Research

The boundaries can shift during a project, and the door-to-provider project shows how. Suppose the emergency department team, after six months of successful improvement cycles, decides to compare its new triage model with the old one by assigning alternate days to each model and publishing the result as evidence that the model works in emergency departments generally. At that point the project has acquired features of research: a planned comparison designed to produce generalizable knowledge. It would need to go to the institutional review board before the comparison begins, not after the data are collected. Many institutions use a short screening tool to decide whether a project needs research review, and the safest practice is to consult the review board whenever a quality project adds a comparison design or an intention to publish generalizable findings. Publishing quality improvement work is not forbidden; it simply has to be described honestly as local improvement, often using reporting guidelines designed for improvement studies.

Common Errors With Levels of Measurement

Three errors appear often in student and practice projects. The first is averaging ordinal data as if it were interval, for example reporting a mean pain score of 4.6 on a 0 to 10 scale without acknowledging the assumption of equal intervals; the median is the safer summary, and if the mean is used, the assumption should be stated. The second is treating numeric codes as numbers: a unit coded 1, 2 or 3 in a spreadsheet is still nominal, and an average unit code of 1.8 means nothing. The third is collapsing ratio data into categories too early, such as recording door-to-provider time only as within or beyond 30 minutes; the categories are useful for reporting against a target, but keeping the minutes allows the team to see whether long waits are shrinking even when the target is not yet met. Deciding the level of measurement at the design stage prevents all three.

Why the Classification Matters

Classification decides three things. First, oversight: only the mattress trial requires institutional review board review as research, which affects timelines and consent. Second, methods: the research project needs a sample size calculation and a design that controls bias; the evidence-based practice project needs a structured appraisal of evidence; the quality improvement project needs run charts and rapid tests of change. Third, the levels of measurement guide statistics. Nominal outcomes such as pressure injury occurrence call for proportions and tests such as chi-square; ratio outcomes such as minutes to provider support means, medians and tests such as t-tests, provided their distributions are checked; ordinal outcomes such as injury stage call for medians and nonparametric tests.

Conclusion

Three projects in one hospital showed how research, evidence-based practice and quality improvement differ in purpose, methods and oversight, even when they address similar concerns. Sorting their variables by level of measurement prepares for the choices later weeks require: which descriptive statistics to report, which graphs to draw and which inferential tests fit. Getting these foundations right at the start prevents analyses that are mathematically possible but meaningless for the question being asked.

References

Melnyk, B. M., & Fineout-Overholt, E. (2023). Evidence-based practice in nursing & healthcare: A guide to best practice (5th ed.). Wolters Kluwer.

Polit, D. F., & Beck, C. T. (2021). Nursing research: Generating and assessing evidence for nursing practice (11th ed.). Wolters Kluwer.

Shirey, M. R., Hauck, S. L., Embree, J. L., Kinner, T. J., Schaar, G. L., Phillips, L. A., Ashby, S. R., Swenty, C. F., & McCool, I. A. (2011). Showcasing differences between quality improvement, evidence-based practice, and research. Journal of Continuing Education in Nursing, 42(2), 57-68. https://doi.org/10.3928/00220124-20100701-01