You have data, or soon will, and your committee will ask one question: did anything change, and how do you know? This guide is for DNP students who need to choose measures, display results honestly, interpret them and write the results, limitations and dissemination sections. You will get charts, tests and wording that fit a practice project.
Quick answer. A sound evaluation shows what happened to a small, well-defined set of measures, compares it with a stated baseline, and explains the result without overclaiming. It then asks whether the change will last and who needs to hear about it.
Before you choose a chart or a test, decide what questions the evaluation must answer. Most DNP projects need to answer three: did the process change as planned, did the outcome move, and did the change cause problems somewhere else?
The Model for Improvement, used widely in healthcare quality work, asks what you are trying to accomplish, how you will know that a change is an improvement, and what change you can make. Donabedian's model organizes quality into structure, process and outcome. Either can serve as a skeleton for the evaluation section, and both connect to the theory you chose earlier. For help with that link, see the framework application guide.
Many committees like to see the whole evaluation on a single page. This layout also helps you notice gaps before you collect a single data point.
| Aim | Measure | Data source | Baseline | Target | Analysis |
|---|---|---|---|---|---|
| Improve process reliability | Checklist completion, weekly | Chart audit | Baseline median from audit | Agreed with site, with a source | Run chart, shift rule |
| Improve the outcome | Outcome rate, monthly | Quality report | Baseline period rate | Site target or benchmark | Descriptive comparison |
| Avoid harm elsewhere | Balancing measure | Staff form | Baseline estimate | No meaningful worsening | Descriptive summary |
Each family of measures has a different job. A project with only outcome measures cannot say why results changed; a project with only process measures cannot say whether patients benefited.
| Family | Question it answers | Illustrative example |
|---|---|---|
| Outcome | Did the result we care about change? | Proportion of patients with an unplanned return within 30 days |
| Process | Did the steps happen reliably? | Proportion of eligible discharges with the new checklist completed |
| Balancing | Did we create a problem elsewhere? | Minutes added to discharge, or staff overtime |
| Implementation | Was it acceptable, adopted and feasible? | Staff acceptability rating, share of staff using the tool |
| Cost (optional) | What did it take, and what might it save? | Staff hours invested; avoided event costs if a defensible source exists |
Link measures to aims. Every aim in your proposal should map to at least one measure. If an aim has no measure, either add one or rewrite the aim. Revisit the proposal guide if your aims and measures have drifted apart.
A result means little without something to compare it against. Most practice projects use a baseline period from the same site, sometimes with a comparison unit.
Sometimes projects start without one. In that case, use a documented benchmark, an earlier audit, or a comparison unit, and state clearly that your ability to claim improvement is limited.
A run chart plots a measure over time with a median line. It is one of the simplest and most persuasive tools available to a practice-focused student because it shows the pattern, not just the endpoints.
| Signal | Common rule of thumb | Interpretation |
|---|---|---|
| Shift | Six or more consecutive points on one side of the median | Sustained change in the process |
| Trend | Five or more consecutive points going up or going down | Gradual movement in one direction |
| Median points | Ignore points that fall exactly on the median | Avoid over-reading |
| Very small series | Fewer than about ten points | Describe patterns cautiously, or use a table |
Control charts add statistical limits and generally need a longer series. If your program asks for one, the Institute for Healthcare Improvement publishes free guidance on both chart types. Format figures in APA 7 style; our tables and figures guide shows how.
You rarely need advanced methods. You do need to choose methods that match your data and to report them completely. Our biostatistics guide goes deeper, and the summary below covers the usual cases.
| Situation | Reasonable approach | What to report |
|---|---|---|
| Describing the sample and measures | Counts, percentages, means, medians | Sample size, missing data, ranges |
| Same people measured twice | Paired t-test or Wilcoxon signed-rank | Mean or median change, test result, effect size |
| Different groups before and after | Independent t-test, Mann-Whitney U, chi-square or Fisher exact | Group sizes, difference, confidence interval |
| Change in a proportion | Difference in proportions with interval | Numerator and denominator, not only the percent |
| Time series | Run chart rules; segmented approaches if supported | Chart, rule applied, notable events |
A small project may produce a change that matters clinically but fails to reach statistical significance, and a large data set can make a trivial change look significant. Say in advance what size of change the site would consider worthwhile, then judge the result against that standard. Report confidence intervals where you can, because they show how uncertain the estimate is.
A confidence interval gives a range of values that are reasonably compatible with your data. A narrow interval suggests a fairly precise estimate, and a wide one signals uncertainty. When an interval for a difference includes zero, the data are consistent with no change as well as with some change, and that is worth saying openly.
Numbers explain what changed; people explain why. A short set of staff comments or a brief survey often provides the most useful evidence for the discussion section.
When you analyze comments, sort them into a small number of categories, show a representative quotation for each, and say how many people contributed. Avoid using a single memorable comment to carry an argument.
Share your measures and data summary with your brief, and we can help draft or edit the results, discussion and limitations. The price is shown before you pay, and every delivered paper includes 14 days of free revisions.
The results section reports what you found, in the order of your aims, without interpretation. Save the meaning for the discussion. See our results and discussion resource for layout ideas.
| Part | What to include |
|---|---|
| Sample and delivery | Who was included, how many, how much of the intervention was delivered |
| Primary result | The main measure, baseline versus implementation, with the figure or table |
| Secondary results | Process, balancing and implementation measures |
| Staff or patient feedback | Summary of survey and comment data |
| Unexpected findings | Anything that departed from the plan, reported plainly |
| Weak | Stronger |
|---|---|
| The project was successful. | The weekly completion rate was above the baseline median for eight consecutive weeks, which meets the shift rule (see Figure 2). |
| Results were significant. | The proportion of eligible discharges with complete documentation rose from the baseline median to the implementation median (report both, with counts) and the difference was statistically significant (report the test and p-value). |
| Staff liked the tool. | Most respondents (report n and percent) rated the tool acceptable, and several described the added time as the main drawback. |
The paragraph below uses placeholders. Replace every bracket with your own data, and never keep wording that your data do not support.
"During the [number]-week implementation period, [n] patients met the eligibility criteria. The weekly proportion of eligible patients with complete documentation was above the baseline median for [x] consecutive weeks, which meets the shift rule (Figure 1). The median rose from [baseline value] to [implementation value]. The balancing measure, minutes added per discharge, had a median of [value] (range [low] to [high]). Staff acceptability was rated [summary] by [n] of [N] respondents."
Notice that the paragraph reports counts, names the figure and states the rule used. It does not say the project was successful. That judgment belongs in the discussion.
The discussion explains what the results mean for the problem you started with. A helpful habit is to answer four questions in order: what did we find, how does it compare with the evidence, why might it have happened, and what should the site do now?
| Result pattern | Honest interpretation |
|---|---|
| Process improved, outcome unchanged | The step may be working but the outcome needs more time, more reach or a different lever |
| Both improved | Consistent with benefit, but other changes may contribute; discuss them |
| Neither changed | Check fidelity and reach first; then consider whether the intervention fit the setting |
| Outcome improved, balancing measure worsened | A trade-off; recommend refinements before scale-up |
A null or mixed result is not a failed project. Committees generally value an honest account of why a plan did not work over an inflated claim of success.
Keep each of those moves in its own short paragraph or a few clear sentences, and you will avoid the common trap of a discussion that simply repeats the results.
Sustainability asks whether the change will still be in place, and still working, after the project team leaves. Build the answer from concrete structures, not good intentions.
| Element | Example |
|---|---|
| Ownership | A named manager or champion responsible after the project |
| Embedding | Change built into a template, policy, orientation or standing work |
| Monitoring | A measure the site keeps reporting, with a simple chart and an owner |
| Feedback | Regular sharing of results with frontline staff |
| Review | A scheduled check-in to decide whether to keep, adapt or stop |
Spread, or extending the change to other units, should be described as a recommendation with conditions, not as something your project has already shown.
Limitations are a strength when they are specific. Describe the limit, how it might affect the conclusions and what you did to reduce it.
| Limitation | Weak wording | Stronger wording |
|---|---|---|
| Single site | The project was small. | The project took place on one unit, so results may not apply to sites with different staffing or patient mix. |
| No comparison group | There was no control. | Without a comparison unit, changes that occurred at the same time cannot be ruled out as contributors. |
| Short follow-up | Time was limited. | Follow-up ended when the implementation window closed, so long-term durability is unknown. |
| Measurement | Data may be imperfect. | Documentation was used as a proxy for practice, and undocumented care could not be captured. |
Sharing results is part of translating evidence into practice. Match the product to the audience, and plan it early. The dissemination plan resource gives a template.
| Audience | Useful product |
|---|---|
| Unit staff | One-page summary and a huddle presentation |
| Leadership | Executive summary with the key chart and recommendations (see the executive summary resource) |
| Program and committee | Final manuscript and oral defense |
| Professional community | Poster (see the scholarly poster resource) or a journal manuscript |
For a manuscript, the SQUIRE 2.0 reporting guideline, listed on the EQUATOR Network, is widely used for improvement work. If you plan to publish, check authorship and reporting expectations in the recommendations of the ICMJE, and agree authorship with your chair and site in advance.
| Feedback | What it means | Fix |
|---|---|---|
| "You over-claim." | Causal language with a design that cannot support it. | Use "was associated with" and add a paragraph on other explanations |
| "Where is the time trend?" | Only two summary numbers are shown. | Add a run chart with the baseline median extended |
| "How do we know staff used it?" | No process or fidelity data. | Report adoption and fidelity audit results |
| "So what?" | Discussion repeats results without recommendations. | End with concrete, site-specific next steps |
Illustrative example, not a real client. This short story is invented to show the pattern, and it contains no real people or numbers.
The problem. A student had a before and after percentage and planned to write, "The project was a success."
The tension. Her chair asked about the time trend, whether staff actually used the tool and whether it caused any side effects. She had no answers.
The turn. She plotted weekly data on a run chart with the baseline median, added a balancing measure from a short staff form and summarized staff comments in three categories.
The proof. At the defense the panel described the results as measured and honest, including her limitation about a single site.
The payoff. The unit manager asked to keep the chart going, which gave her sustainability section a real owner.
More is better. Many practitioners look for roughly ten or more points, but the right number depends on your measure. If you have fewer, describe patterns cautiously and consider a table.
Usually not with a single-site, before-and-after design. Use language such as "was associated with" and discuss other explanations.
Report them honestly. Discuss the size of the change, the sample and the limits of the test, and explain what the site can learn.
Only if your program requires it. If you include costs, cite defensible sources and separate real costs from estimated savings.
SQUIRE 2.0 for improvement work is a common choice. Ask your chair which standard your program follows.
A strong evaluation returns to the problem you started with and says, in plain terms, what happened, how sure you are and what should happen next. When measures, charts and narrative agree, the committee can see that your project made a real contribution.
Want help presenting your results? Get my instant quote. The price is shown before you pay, every delivered paper includes 14 days of free revisions, and refund terms are on the money-back guarantee page. Please use any model paper in line with your institution's academic-integrity policy.