More on Methods and Statistics
burnout. And—perhaps of greatest importance— nurse-appraised quality of care is significantly higher in ANCC magnet hospitals.
Although this study did not systematically com- pare either type of magnet hospital to a representa- tive group of nonmagnet hospitals—a topic for future research when data from a current study of all Pennsylvania hospitals become available—we provide evidence from other studies suggesting that both groups of magnet hospitals are more highly appraised by the nurses practicing in them than is the case in nonmagnet hospitals. The original mag- net hospitals are not immune to changes in the nation’s health care system, and some have been adversely affected, including one of the original 41 that recently lost its accreditation because of poor quality of care. However, despite vast organiza- tional change, many of the original magnet hospi- tals continue to foster elements of professional nursing practice that distinguish them from non- magnet hospitals.24 Thus, our use of the original magnet hospitals as a comparison group likely underestimates, perhaps to a significant extent, the differences between ANCC magnet hospitals and nonmagnet hospitals. We are currently working to pool information from the nurses’ surveys with additional data from the participating hospitals— such as discharge data on patients with AIDS—and hospital-level data from the American Hospital Association and mortality statistics from Medicare. This pooled data set will allow us to determine whether (and to what extent) mortality differences between the ANCC magnet hospitals and the origi- nal magnet hospitals are due to differences in how nursing care is organized across the hospitals.
Much work remains to be done in determining how nurse staffing and skill interact with other organizational features to affect patient outcomes. A cross-national study currently under way on the impact of hospital organization and staffing on patient outcomes will soon provide definitive infor- mation on this important issue and offer a large rep- resentative sample of hospitals against which the ANCC magnet hospitals can be compared.25 Based on the excellent showing of ANCC magnet hospi- tals compared with a group of the original magnet hospitals in the present study, we anticipate that the ANCC hospitals will fare even better when com- pared with a more representative group of hospitals in the United States. That study will also provide the first firm estimate of the proportion of all hospitals in the United States that might qualify for ANCC magnet hospital recognition based on their nursing practice environments and patient outcomes. There is no way of knowing what percentage of hospitals now meets the ANA standards for professional nursing practice (which constitute the core criteria for ANCC magnet recognition).
T his study used a comparative multisite observational design incor- porating two subsamples of hospitals. Our objective was to com- pare contemporary ANCC-recognized magnet hospitals with the original magnet hospitals. The ANCC hospitals studied included all ANCC magnet hospitals that existed at the time the study began (n = 7). The original magnet hospital subsample was selected through use of a sampling frame developed by Marlene Kramer for her 1986 study of the original magnet hospitals.7 Kramer used a 40% sample— proportionate by regions of the country—of the 41 original magnet hospitals. Because part of our research focuses on changes in the clin- ical practice environment in magnet hospitals over time, we restudied the original magnet hospitals from Kramer’s sample to have data on magnet hospitals at two points in time. At the time of our study (1998), two of Kramer’s 16 hospitals had merged, leaving 15 hospi- tals. We attempted to recruit all 15 hospitals to our study, but three nurse executives declined to participate, leaving us with 12 hospitals and without an original magnet hospital in the West Coast region. We therefore recruited the remaining original magnet hospital located in that region for our study (a hospital not studied by Kramer), for a total of 13 original magnet hospitals. The three nurse executives who declined to participate noted that their hospitals no longer had the elements of professional nurse practice that had won them AAN mag- net hospital designation in 1982. Thus, to the extent that the refusals bias the sample of original magnet hospitals, the bias would be toward having a stronger group of original magnet hospitals, since the weaker ones declined to participate and would thereby provide a stronger test of how the ANCC-recognized magnet hospitals measure up than might be the case if all the original magnet hospitals had been included.
This study, approved by all 20 hospitals’ institutional review boards, was conducted in the spring of 1998. The nurse survey com- ponent of the study included a census of all registered nurses on staff who worked at least 16 hours per week on any medical or surgical nursing unit in study hospitals, yielding roughly 3,600 eligible nurses. Study participation was voluntary and all participating nurses pro- vided informed consent. A research nurse was appointed at each hos- pital to carry out the protocol, which involved distributing the questionnaire packets and sending reminder postcards at two weeks, follow-up questionnaires at four weeks, and final reminder postcards at six weeks.
The NWI-R is a modification of the Nursing Work Index,7 consisting of 49 items rated with a 4-point Likert-type scale that gauges staff nurse perceptions of selected organizational traits in their work setting (for example, “This factor is present in my current job situation”). Three of the NWI-R subscales—which measure the constructs of nurse autonomy, nurse control over the practice setting, and nurses’ rela- tions with physicians—have been used in multiple studies and have consistently demonstrated acceptable internal consistency reliability (Cronbach’s alpha): autonomy, 0.78; control, 0.79; and nurse– physician relations, 0.73.18 Completed questionnaires were received from 2,045 nurses (56% response rate), of whom 1,064 nurse respondents were in ANCC magnet hospitals and 981 were in origi- nal magnet hospitals.
Although the results we report are bivariate and largely descriptive, we used significance tests (chi-square statistics with categorical vari- ables and t-statistics with continuous ones, such as the years-of-experi- ence and practice-environment subscales) to ensure that the differences we observed between the nurses in the two groups of hos- pitals were not the result of sampling fluctuations or chance.
The slow rate of ANCC magnet hospital recog- nition is problematic. This potentially useful quality indicator must be propelled into the public domain. At the time of this writing, almost a decade after the launching of the ANCC magnet hospital applica-
JONA • Vol. 39, No. 7/8 • July/August Supplement 2009
Copyright @ 2009 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.