Determining clinically significant patient change with effect sizes: considerations for clinicians and researchers

Abstract

Objective: To aid nurses’ understanding of effect size utilisation in clinical and research contexts.

Design and data sources: Methodological discussion paper that is based on the author’s experiences as a clinician and researcher and is supported by literature.

Primary arguments: Patient change is a key consideration for clinical nurses and nurse researchers. Nurses routinely use measurement instruments to identify and quantify such change informing intervention outcomes, clinical decision-making, and health research conclusions. Whether improvement or deterioration, patient change should be operationalised through the magnitude of change (i.e., effect size). Effect sizes relative to the context of change (clinical vs empirical) and the reliability of the instruments used are important considerations here. However, despite discourse on the utilisation of effect sizes in health, aspects of effect sizes can be poorly understood, misapplied or overlooked. Furthermore, nurse researchers may default to Cohen’s d for use in power analysis and results reporting where they should be considering an effect size derived from other methods in the first instance. In part, this is due to the literature surrounding aspects of effect size being inherently complex, impacting on nurse and nurse researchers’ capacity to acquire a thorough understanding of the topic.

Conclusions: Effect size in health can be particularly complex. Nevertheless, nurses and nurse researchers should have some understanding about effect sizes and their role in measuring patient change in clinical and empirical contexts. They need to be aware of how measurement instruments detect, track and quantify patient change and the resultant magnitude of effect relative to the clinical significance of the change for the patient. This paper aids nurses to effect robust change based on informed decision making thus strengthening their evidence-based practice.

What is already known about the topic?

• Patient change informs nurses clinical decision-making strategies; however, nurses may not consider the magnitude of change relative to the context of change and the reliability of the instruments used to identify and quantify the change


• Effect size is one of the four criteria needed for power analysis and is perhaps the most difficult to identify


• Underpowered studies result in imprecise estimation of the true effect, which could be an over- or an under-estimation

What this paper adds:

• This paper dispenses with the inherently complicated and technical terminology on effect size often found in the literature that can impact understanding. Consequently, this paper equips nurses to critique research literature and apply this knowledge to their clinical practice


• Provides other methods for identifying a suitable effect size for use in power analysis and results reporting as opposed to defaulting to Cohen’s d


• Draws attention to the importance of reporting effect size and associated confidence interval with research results data