Paper
Friday, 21 July 2006
This presentation is part of : Using A Collaborative Research Utilization Model to Translate Evidence-Based Best Practices in a Magnet Community Hospital
Strategies for the Statistical Analysis of Pilot Data: An Example from the Translating Best Practice in Non-drug Pain Management Study
Cynthia Willey-Temkin, PhD, College of Pharmacy, College of Pharmacy, University of Rhode Island, Kingston, RI, USA, Susan Rossi, PhD, RN, Cancer Prevention Research Center, University of Rhode Island, Kingston, RI, USA, and Stephen Kogut, MBA, PhD, College of Pharmacy, University of Rhode Island, Kingston, RI, USA.
Objective: The objective is to examine the use of traditional and non-traditional statistical techniques for the analysis of pilot data. Data from a study that tested the effects of translating scientific advances in non-pharmacological pain management on outcomes in older adults undergoing knee replacement is used for illustration. Design. The overall evaluation research design of the project had 2 components: (1.) A primary efficacy study (n= 137 using a 2 group, quasi-experimental design to determine the impact on pain intensity, functional ability, and patient satisfaction; and (2) A sub-study, using a single group (n=47) pretest-posttest exploratory descriptive design to evaluate the tailored teaching intervention and to describe the changes in the patients' knowledge, attitudes, ability to use, and actual use of the three non-drug "best practice" protocols. The groups were compared at each of three points along the trajectory of nursing care: (a) at baseline admission on the day of surgery, (b) on the evening of postoperative day 3 before discharge from the surgical unit; and (c) on the day of discharge from the rehabilitation unit. The following topics will be examined using these data: a) Graphical analysis of pilot studies, b) Non-parametric versus parametric techniques for pilot study data, c) Transforming data to increase the power of statistical tests, and d) Using change scores as an alternative method for controlling for baseline differences. Conclusions: The analysis of pilot study data should not be limited to traditional statistical approaches, or simple counts of outcomes. Novel techniques such as change score transformations can increase power and allow detailed analyses of pilot data and other small sample studies.

See more of Using A Collaborative Research Utilization Model to Translate Evidence-Based Best Practices in a Magnet Community Hospital
See more of The 17th International Nursing Research Congress Focusing on Evidence-Based Practice (19-22 July 2006)