The problem addressed by the article is the impact of nursing turnover on patient care. It has been assumed that patient care is affected by turnover but there has not been a good understanding of the relationship between turnover and quality of care. Most research has simply shown that there is a direct relationship between the two without exploring the underlying mechanisms.
The study aims at describing the underlying mechanisms in the relationship between nursing turnover and quality of care of patients. The objective of the study was to find out how turnover in nursing units affects processes among key work groups and how the impact of these processes affects outcomes of patients.
The study variables are nursing unit turnover, workgroup processes, patient outcomes, control variables and education level. The independent variables was nursing unit turnover while the dependent variable was patient outcomes which included patient satisfaction, average length of stay, medication errors and patient falls. Workgroup cohesion, relational coordination and workgroup learning were the process variables. Other variables included in the article were control variables which included work complexity, teaching status, technological sophistication, hospital size and unit size as well as nurse characteristics such as RN hours, unit tenure, education level and patient characteristics (health status, patient age and previous hospitalizations).
The conceptual framework applied was modeled around the IPO framework (input-process-outcome) which assesses workgroup behavior and performance effectiveness. It is incorporated in most models of workgroup effectiveness. Workgroup processes are those mechanisms which inhibit or encourage members of a team to combine their abilities and behavior (Kozlowski et al, 2003). These include workgroup cohesion, workgroup learning and relational coordination. The unit level patient outcomes include patient satisfaction, length of stay, patient falls and medication errors. In the conceptual framework, the hospital, nurse and patient characteristics as well as the nursing unit were identified as control variables.
The literature review supports a need for the study as it identifies the fact that there has been research that establishes that there truly is an effect of nursing turnover on patient outcomes. However there is very little information that describes the actual relationship between nursing turnover and patient outcomes. The literature review identifies some of the effects of turnover as detachment, a disruption in communication flow and a disruption of established relationships (Sung-Heui et al, 2010). A nursing unit with high turnover requires more time for new staff to adjust while the staff that remain have to be more careful when supervising the new staff. This has the overall effect of lowering relational coordination. The article also identifies that work learning is does not occur well when there is no stability as is the case in units with high turnover (Sung-Heui et al, 2010). Learning needs both change and stability thus units with very high or very low turnover may not have much cognitive growth.
When there is greater cohesion among staff members there is also greater motivation for the employees to provide better service which leads to better patient satisfaction (Sung-Heui et al, 2010). Other patient outcomes addressed by the article include length of stay which measures how efficient the hospital is, which in turn is an indication of the efficiency in communication between healthcare providers. If a workgroup is well coordinated then there outcomes are of higher quality and they outcomes are achieved more efficiently. The article also focuses on patient falls and medication errors as an indicator that there is work group coordination and workgroup learning. When errors occur there is opportunity for learning however if there is too high turnover, then learning does not occur and there is likelihood for errors to be repeated.
The study design was a non-experimental longitudinal study causal modeling study. This design was appropriate for the study since it tries to establish a relationship between two variables that is whether one causes the other. Some of the threats to internal validity that the study may have include history which means that as more time elapses between measurements there is a likelihood that time may contribute to differences in results in the variables. Testing is also a possible threat to internal validity in a longitudinal study as it requires the participants to take certain tests on several occasions (Barry, 2005)
Attrition is a threat to the longitudinal study design; when participants drop out of the study before all the data is collected is likely to distort the accuracy of results (Barry, 2005).
Sample and setting
The sample consisted of 268 nursing units from 141 hospitals. This sample size was adequate to ensure generalizability of the research findings. Additionally using nursing units from various hospitals served to increase the generalizability of the findings. The sample was derived from the medical-surgical units and nursing units of acute care facilities of various hospitals, settings that again provided a good representation of all the hospitals and helped to increase the external validity of the study.
Identification and control of extraneous variables
Some of the identified extraneous variables included environmental uncertainty such as volumes, types of patients which increased the complexity in work processes and group dynamics among nurses. Work complexity was measured using a 7-item scale developed to measure environmental uncertainty. The scale measured work complexity by identifying frequent interruptions or unanticipated events. Technological sophistication was also measured using the Saidin Index which provided a weighted sum of the number of technologies and services that were available in the hospital.
Both primary and secondary data were collected in this study. Secondary data was obtained from secondary data collected from the Outcomes Research in Nursing Administration Project (II) (ORNA II). Nursing units provided turnover rates for six months and nurses also completed questionnaires which measured various workgroup processes. The patient data collected was also from the ORNA II project. The use of such data collection methods was appropriate as it provided a rich data source. Additionally, the use of data from the ORNA project provided a reliable source of good quality data.
Data analysis procedures
The unit of analysis in the study was the nursing unit. The variables relational coordination, work group learning, patient satisfaction, workgroup cohesion, patient satisfaction and work complexity were measured at the individual level and were then aggregated to the unit level. The study used linear and count models to determine the distribution of the outcome and process variables. The average LOS was estimated using a model of random effects while the medication errors and patient falls were analyzed using a Poisson regression model that was adjusted for dispersion. The Poisson regression is appropriate for rate data such as the one collected in this study. The use of the linear counts is also appropriate as it allows for a relation between the linear model and the response variable.
The strength of the study lies in the use of the IPO framework. Using this framework allows for the underlying mechanisms such as workgroup processes to be evaluated and considered as variables. Thus the study looks at input variables, process variables and output variables providing adequate data with which to describe the underlying mechanisms and relationship between nursing turnover and patient outcomes.
Some of the limitations of the study include the model which assumes that the turnover affects relational coordination a few months after it has occurred. The time period of two months may not have been sufficient to assess true variation of the turnover levels. Another limitation was the missing variables that have an effect on turnover, patient outcomes and workgroup processes. Some of these include the support of managers, supervision and organizational effectiveness which also have an impact on whether a nurse stays or not.
The results of the study make it necessary that there be further research on the effect of nursing turnover on patient outcomes and workgroup processes. The findings of the study support a need to increase work group coordination and cohesion so as to improve patient satisfaction. These findings are consistent with previous research findings. The research findings also show that the in nursing units where there was higher workgroup learning there were fewer errors in giving medications. This supports empirical evidence for the development and sustenance of processes where nurses are supported in discussing and learning from their errors.