Conflicting evidence in nursing can make a literature review feel difficult to organise. One intervention study reports a benefit, another reports little difference and a third reaches an uncertain conclusion. Your task is to explain what those studies actually disagree about and what can reasonably be concluded from the available evidence.
This guide focuses on disagreements between quantitative intervention studies. It complements our broader guide to comparing and synthesising nursing literature. It does not cover conflicting participant accounts or the integration of qualitative and quantitative findings. All study names and numerical results in the worked example are fictional teaching illustrations.
Choose the guide that matches your evidence: use conflicting participant accounts for divergence within qualitative interviews, or quantitative and qualitative findings that disagree for integration within a mixed-methods project. This article compares quantitative intervention estimates across separate studies.
Define the conflicting evidence in nursing precisely
Start by writing the apparent disagreement in one sentence. “The evidence is mixed” is too broad to guide analysis. A more useful sentence might say that studies differ in their estimates of the effect of a discharge follow-up programme on readmission within thirty days. This identifies an intervention, outcome and time period that you can check.
Then ask whether the studies are actually answering that same question. One may examine telephone contact within two days, while another evaluates several home visits. One may count all readmissions, while another counts only those judged related to the original condition. Different answers to different questions are not necessarily contradictory findings.
Read the methods and results rather than relying on the abstract conclusion. Authors may use words such as effective, promising or unsuccessful differently. Extract the relevant estimate and its uncertainty where available. Keep your description close to the reported outcome instead of allowing a memorable concluding phrase to stand in for the evidence.
Distinguish four kinds of apparent conflict
| Apparent conflict | What to examine | Useful writing response |
|---|---|---|
| Different statistical conclusions | Effect estimates and confidence intervals | Explain whether the magnitude and direction really differ |
| Different outcomes | Definitions, scales and measurement times | Keep the outcomes separate |
| Different intervention delivery | Content, intensity and implementation | Describe what was actually compared |
| Opposing estimates | Context, precision and risk of bias | Explore plausible explanations without claiming proof |
Build a table that makes comparison possible
Use our nursing evidence-table guide to organise the extraction, then focus the comparison on the disagreement. Organise the extraction table around the review question. Include the population, setting, intervention, comparator, relevant outcome, follow-up, effect estimate and important limitations. Add the number contributing to the outcome, not only the number originally recruited. A table should support comparison rather than reproduce every detail of each paper.
Use consistent outcome labels while retaining important distinctions. If one paper reports medication adherence and another reports confidence in medication management, do not place both under “adherence” without explanation. The first concerns behaviour and the second concerns a perception. Their findings may inform the same clinical discussion without estimating the same effect.
Record intervention details that matter to the mechanism. A nurse-led education session may differ from a programme combining education, follow-up and practical support. Similarly, “usual care” needs description when the paper provides it. An intervention may be compared with minimal follow-up in one setting and substantial existing support in another.
Keep a decision column
Add a short column stating how each study can contribute to the synthesis. You might group comparable outcomes together, discuss a different follow-up separately or treat a study as indirect evidence. These are analytic decisions. Explain the rationale in the text, especially when a paper receives less emphasis than its headline conclusion might suggest.
Check multiple reports from the same trial before counting studies. A protocol, main results paper and follow-up report do not represent three independent interventions. Link associated reports in your records and identify which information you use from each. This simple step prevents accidental overstatement of the evidence base.
Read effect direction, magnitude and precision together
A p value does not describe the size of an effect. Two studies can estimate effects in the same direction while only one meets a chosen significance threshold. Cochrane’s guidance on synthesis without meta-analysis warns against vote counting based on statistical significance. Counting papers as positive or negative can therefore manufacture a conflict that is not supported by the estimates.
Check the measure before interpreting direction. A lower risk ratio may favour an intervention for an undesirable event, while a higher score may favour it on a wellbeing scale. Confirm the coding and comparator. Write a plain-language note beside the extracted estimate so that an inverted scale does not reverse your synthesis.
Consider the confidence interval alongside the point estimate. A wide interval signals substantial uncertainty about the effect under the analysis model. It may include important benefit and little difference, or even harm. Do not translate an imprecise estimate into proof that the intervention has no effect.
Separate clinical importance from statistical significance
A small precisely estimated difference may have limited practical importance. Conversely, an uncertain estimate may leave clinically important possibilities unresolved. Use a relevant, justified interpretation of importance where available. Do not invent a universal minimum difference or select a threshold after seeing which one supports your preferred conclusion.
Cochrane’s interpretation guidance encourages attention to the size and uncertainty of effects. In a student review, you can reflect that principle with clear language and a structured table. You do not need to perform a new statistical analysis simply to avoid an overly binary description of the published findings.
Check methods before proposing a clinical explanation
Differences in study design can affect the credibility of estimates. Examine allocation, baseline comparability, departures from the intended intervention, missing outcomes and outcome measurement using a tool appropriate to the design. Do not compare a randomised trial and an uncontrolled before-and-after study as though their estimates had identical protection against bias.
Use appraisal to support specific judgements. Instead of saying one study is weak, identify the relevant limitation and explain how it could affect the finding. For example, substantial missing follow-up may make an estimate difficult to interpret if loss differs between groups. Avoid assuming the direction or size of bias without justification.
Also check the analysis unit and timing. A ward-level intervention may require attention to clustering. Repeated measurements on the same participants create another dependency. You do not need to reanalyse the dataset, but you should understand whether the reported analysis addresses the study design. Seek methodological advice where the paper is unclear.
Do not turn a possible explanation into a result
It is reasonable to suggest that intervention intensity may contribute to differing estimates. It is not reasonable to claim that intensity explains the difference merely because one programme involved more contact. Other features may vary at the same time. Use language such as “may contribute” and state whether the studies actually tested that explanation.
Fictional worked example: follow-up after discharge
Consider three invented studies of a nurse-led follow-up programme. All numbers in this section are fabricated for teaching. They must not be cited as research findings. Assume the outcome is the risk of readmission, with values below one favouring the programme. The simplified table omits details that a real appraisal would require.
| Fictional study | Intervention and follow-up | Risk ratio and 95% confidence interval | Interpretative issue |
|---|---|---|---|
| Study A | Telephone support; 30 days | 0.80 (0.66 to 0.97) | Estimate favours the programme |
| Study B | Telephone support; 30 days | 0.84 (0.60 to 1.18) | Similar direction with greater uncertainty |
| Study C | Home visits; 90 days | 1.05 (0.86 to 1.28) | Different delivery and later outcome |
A weak summary would state that one study found an effect and two found no effect, then conclude that most evidence is negative. That description discards magnitude and precision. In the fictional example, A and B estimate effects in a similar direction. B is less precise. C addresses a different programme and a later time point.
Write what the comparison supports
A stronger account might say that the two telephone-support studies produced estimates favouring reduced thirty-day readmission, although Study B remained compatible with a wider range of effects. It would discuss the different home-visit study separately. It would then examine study limitations before deciding how much confidence to place in either pattern.
This wording does not establish that telephone support works. The invented estimates are only part of the information a real review needs. You would still need the study populations, comparator care, implementation details and risk-of-bias assessment. The exercise demonstrates a better comparison process, not a clinical recommendation.
Write a synthesis paragraph around the disagreement
Begin with a statement about the relationship between findings. Follow with the most relevant evidence, including estimates or a clear description of uncertainty. Then explain the important differences in context or methods. Finish with a qualified judgement that answers the review question without concealing what remains unresolved.
This structure differs from writing one paragraph per paper. The organising unit is the clinical or methodological issue. Studies enter the paragraph because they help address that issue. A reader should not have to remember several pages of separate summaries before discovering that the papers measured different outcomes.
Use precise transitions. “Similarly” is suitable when findings align in the respect you describe. “However” should introduce a real qualification. “By contrast” needs an actual comparison. Avoid using connecting words to imply agreement or disagreement that the extracted data do not support.
Useful sentence patterns
You might write: “Although the authors reached different significance-based conclusions, their effect estimates were similar in direction.” Another pattern is: “The later follow-up addresses a different period of risk, so it should not be treated as a direct replication.” Adapt these sentences only when the evidence supports them. They are writing tools, not conclusions to insert automatically.
When several outcomes appear, keep the claim bounded. A programme may improve a patient-reported outcome while leaving readmission uncertain. Avoid collapsing those findings into “effective overall”. Instead, state which outcome supports which conclusion and whether the review question prioritises one of them.
Explain uncertainty without becoming vague
Uncertainty can arise from imprecision, methodological limitations, indirectness or genuine differences across studies. Identify the relevant issue rather than using “more research is needed” as a substitute for analysis. A useful research recommendation follows from the problem: clearer intervention reporting, consistent outcome timing or adequately planned follow-up may be more informative than simply requesting a larger study.
If a meta-analysis is unavailable or inappropriate, explain the alternative synthesis method. Campbell et al. (2020) developed the SWiM reporting guideline for systematic reviews that synthesise quantitative intervention effects without meta-analysis. It encourages transparent reporting of grouping and synthesis decisions. It does not mean any narrative paragraph automatically becomes a rigorous synthesis.
Do not calculate an informal pooled average from incompatible measures. Averaging a risk ratio with a mean difference has no useful interpretation. Even apparently similar measures require appropriate methods and assumptions. A structured, honest narrative is preferable to a new number that your design and training cannot support.
Keep implications proportionate
Separate a possible practice implication from a firm recommendation. If the evidence is limited, you may identify an issue for local assessment or further evaluation. Consider harms, burden, acceptability and resources when the review question requires them. A narrow efficacy finding does not by itself establish the best policy for every nursing service.
For writing decisions about practice recommendations, keep the scope distinct from this article’s comparison task. Here, the main objective is to explain the disagreement accurately. Do not allow a persuasive final paragraph to erase limitations you carefully described earlier in the review.
Check the draft before submission
Highlight every claim that uses words such as consistent, conflicting, effective or no difference. Ask whether the extracted data support that exact wording. Check all effect directions against the original papers. Then inspect the tables for outcome definitions and time points. These checks often reveal a stronger explanation than another round of stylistic editing.
Finally, read the conclusion without the results chapter. Does it overstate agreement, certainty or generalisability? A good conclusion can be clear and cautious at the same time. State the most defensible interpretation, identify the main source of uncertainty and explain why that matters to the nursing question.
Keep a transparent synthesis decision log
Record why studies were grouped together and why any outcome was discussed separately. Include the point at which you noticed an important difference in timing, delivery or measurement. This short log can prevent later editing from making the comparison appear more straightforward than it was. It also supports a clear methods account.
For each synthesis claim, list the studies that contribute and the reason they are relevant. Then note the main qualification. You might record that two reports estimate the same thirty-day outcome, while a third informs a separate discussion about longer follow-up. The log should identify the actual source information rather than repeat only the final conclusion.
Cochrane (n.d.-a) supports structured preparation for synthesis, while Cochrane (n.d.-b, n.d.-c) addresses alternative synthesis and interpretation. These resources help separate data preparation, analytic choice and final judgement. In your dissertation, describe the procedures you actually used. Do not add a formal method name or confidence assessment merely because it sounds more rigorous than the work recorded.
Frequently asked questions
Do different p values prove that studies disagree?
No. Compare effect estimates, their precision and the questions addressed. Similar estimates can lead to different significance-based conclusions. Do not classify findings solely by whether they cross a chosen p-value threshold.
Should I count how many studies support the intervention?
A simple positive-versus-negative count can mislead, especially when based on significance. Use a justified synthesis method that considers what was measured, the estimates and study limitations. Explain the grouping decisions.
Can I say an intervention has no effect?
Only if the evidence supports that specific interpretation. A non-significant result alone does not establish no effect. Describe the range of effects compatible with the estimate and distinguish absence of evidence from a precise estimate of little difference.
What if the outcomes cannot be compared directly?
Keep them in separate groups and explain the distinction. You can discuss how they inform the broader nursing problem without treating them as interchangeable measures. Do not create a combined effect using incompatible units.
Is this the same as conflicting interview accounts?
No. Contradictions within qualitative accounts require a different interpretative task. This guide concerns quantitative intervention estimates across studies. Keep your synthesis method aligned with the type of evidence you are analysing.
Can I exclude an inconvenient study?
Not because its finding conflicts with your argument. Apply the stated eligibility criteria and document decisions. If the study is less relevant or has important limitations, explain that judgement and its effect on the synthesis.
Get feedback on your evidence comparison
Our nursing dissertation support can help you review the clarity of your own synthesis and identify claims that need tighter evidence. Learn about us, read how feedback works and consider client reviews. When you request academic feedback, bring the extraction table and the paragraph you want to improve. Keep responsibility for study selection, interpretation and authorship, acknowledge permitted help and follow university policy. Feedback should strengthen your understanding rather than replace your own academic work.
References
Campbell, M., McKenzie, J.E., Sowden, A., Katikireddi, S.V., Brennan, S.E., Ellis, S., Hartmann-Boyce, J., Ryan, R., Shepperd, S., Thomas, J., Welch, V. and Thomson, H. (2020) ‘Synthesis without meta-analysis (SWiM) in systematic reviews: reporting guideline’, BMJ, 368, l6890. Available at: https://doi.org/10.1136/bmj.l6890 (Accessed: 17 October 2026).
Cochrane (n.d.-a) ‘Chapter 9: Summarizing study characteristics and preparing for synthesis’, Cochrane Handbook for Systematic Reviews of Interventions. Available at: Cochrane Handbook chapter 9 (Accessed: 17 October 2026).
Cochrane (n.d.-b) ‘Chapter 12: Synthesizing and presenting findings using other methods’, Cochrane Handbook for Systematic Reviews of Interventions. Available at: Cochrane Handbook chapter 12 (Accessed: 17 October 2026).
Cochrane (n.d.-c) ‘Chapter 15: Interpreting results and drawing conclusions’, Cochrane Handbook for Systematic Reviews of Interventions. Available at: Cochrane Handbook chapter 15 (Accessed: 17 October 2026).
