Abstract
Dear Editor,
Thank you for the opportunity to respond to the letter authored by Churchill, McClelland, and Siegmund regarding the manuscript authored by Terpeluk, Rogen, and myself, titled, “Medical and Pharmacy Costs for New-Hire Nurses Following a Physical Strength Evaluation Screening in a Large Health System.” Basically, the letter by Churchill et al. focused on two points: the design of the study and the lack of physical characteristics of the two groups (Historical Comparison Group and Physical Capability Evaluation (PCE) Group).
Quasi-experimental designs come from the work of Campbell and Stanley. “Quasi-experiment: An experiment in which units are not assigned to conditions randomly.” There are many natural social settings in which the researcher can introduce experimental design into data collection procedures (e.g., the when and to whom of measurement), even though the study lacks full control over the scheduling of experimental stimuli (e.g., the when and to whom of exposure and the ability to randomize exposures) which is needed for a true experiment. Collectively, such situations can be regarded as quasi-experimental designs (Campbell & Stanley, 1963, p. 34).
The authors of this letter may be conflating the term “control” in the sense of “controlling” for variation in participant characteristics or conditions of the study, with the use of a group of participants who were not exposed to the intervention as a “control.” As the authors of the letter state, those methods are possible with prospective designs.
We emphasized the retrospective nature of the study in the following text from the “Method” section of the article. “Given that this was a retrospective analysis of a hiring practice policy and not a formal research study, Institutional Review Board approval or research consent was not required.” This study did have a non-randomized historical control group that was described in the article in the following statement from the “Method” section: “Nurses hired between January 2009 and December 2010, prior to the strength assessment screening implementation, served as a historical comparison group (HCG) . . .”
In response to Churchill et al.’s concern about the lack of physical characteristics of the two groups, the authors recognize that this was an oversight. One reason is that there were incomplete data for the Historical Comparison Group (height and weight). However, the study had complete data on both groups (HCG and PCE) for age and body mass index (BMI) as provided by the HealthPlan. The average age for the HCG was 34.9 years with an SD of 10.71 years, and for the PCE group, the average age was 33.2 years with an SD of 10.60 years. The average BMI for the HCG was 27.9 with an SD of 6.67, and the average BMI for the PCE group was 27.1 with an SD of 6.25. Knowing that BMI is calculated using both height and weight and the fact that the BMIs were only 0.8 of a point apart with similar standard deviations when comparing the two groups, it is fair to say that the height and weight between the two groups were probably not a factor in terms of medical and pharmacy utilization. Furthermore, the age difference was only 1.7 years, which would not have been a factor in differences in medical and pharmacy utilization. We also know that the two groups were similar in gender breakdown: The HCG was 85% female and the PCE group was 87% female.
It should be noted that Dr. Gilliam contracted the services of Dr. Dee Edington, Dr. Jennifer Pitts, and Angela Camilleri with Edington and Associates at the University of Michigan to review the design, statistical analysis, and manuscript before it was submitted for publication.