March, 2000
The American Nurses Association
The American Nurses Association (ANA) is concerned both with the impact of nursing care on patient outcomes and the professional well being of nurses. To affirm nursing's role in emerging health care systems and to advance knowledge in these areas, ANA commissioned this study (see below) to continue and extend an earlier study. It responds to calls for more research appearing in two seminal studies in this field: the ANA's Report Card for Nursing and the Institute of Medicine's Nursing Staff in Hospitals and Nursing Homes. It seeks to quantify relationships between nurse staffing and patient outcomes for a large scale cross-section of the nation's hospitals and their inpatients. While such relationships may be assumed by some to exist prima facie, little evidence exists that quantifies nursing's impact on patient outcomes. Today's pressures for hospital cost control make it imperative to determine whether differences across acute care hospitals in nurse staffing can be statistically shown to relate to measurable differences in important patient outcomes. The outcome measures chosen for this study were morbidities which can reasonably be theorized to be preventable in some patients by the amount and skill mix of nursing care provided. In measuring such relationships, the study takes into account certain risk and intervening variables, namely patient case mix, teaching status of a hospital and the setting in which a hospital operates.
This study uses data from nine states. For six states, all-Payor data sets were used: California, New York and Massachusetts (the three states used in the previous Nursing Report Card study), plus Arizona, Florida, and Virginia. For these states plus three more -- Minnesota, North Dakota, and Texas -- Medicare data were also used. This provided an all-Payor sample of more than 9.1 million patients in almost 1,000 hospitals; and a Medicare sample of more than 3.8 million patients in over 1,500 hospitals. Nurse staffing data were developed from nationally available data sources provided by the Health Care Financing Administration (HCFA).
Five outcome measures were used in both this study and the first Nursing Report Card study: length of stay, pneumonia, postoperative infections, pressure ulcers and urinary tract infections. A variety of new outcome measures were developed and tested, as well. Each outcome was measured as an index for each hospital, calculated as actual outcomes divided by case mix adjusted expected outcomes. Since the diagnoses flagged as adverse outcomes may or may not have been iatrogenic outcomes, the average adverse outcome rate for a DRG across all patients in a sample was used as an estimate of the normal rate by which these diagnoses could be expected to occur, and indices were calculated so that hospitals above or below this average (once applied to each hospital's mix of patients by DRG) were considered to have higher or lower adverse outcome rates, respectively.
Numerous factors in a hospital's environment are likely to impact the incidence of the selected adverse outcomes and patients' lengths of stay. Case-mix is one so basic to nurse staffing and patient outcomes that it was directly adjusted for in expressing the study's staffing, adverse outcome rate and length of stay index variables. Nursing Intensity Weights (NIWs) were used to acuity-adjust the patient mix at each hospital.(1) Two other factors which have frequently been shown to impact hospitals' costs, staffing and patient outcomes are teaching status (defined herein as primary medical school affiliate, other teaching hospital or non-teaching hospital) and setting (defined herein as large urban, urban or rural). Both of these factors were taken into account in the statistical analyses.
Multiple regression was used to analyze the relationship between nurse staffing and each outcome measure. Separate sets of regressions were run for the all-Payor patient data sets for the six states combined and for the Medicare patient data set for the nine states combined. To further contrast results for the aged, a third set of regressions were run for the Medicare only data set limited to the over 65 population without End Stage Renal Disease. Simply put, all analyses of the five original outcome measures (length of stay, pneumonia, postoperative infections, pressure ulcers and urinary tract infections) show both statistically significant equations and relationships in the predicted direction with nurse staffing. Shorter lengths of stay were found to be associated with greater staffing levels (licensed hours per acuity adjusted day). Secondary bacterial pneumonia, post-operative infection, pressure ulcer and urinary tract infection rates were lower in hospitals with higher registered nurse skill mixes and in some instances with greater staffing levels as well. The additional outcomes tested were not significant, at least not consistently.
Hospital-by-hospital results using all-Payor or Medicare only data were highly consistent, despite the use of different basic patient data sets. The close congruence of all-Payor and Medicare results indicates that the latter information can be used to measure hospitals' performance relative to nursing care and its outcomes. A true nursing report card could be produced with the methods employed for all the nation's hospitals using Medicare data sets, monitoring not only nurse staffing and mix but also patient outcomes. While mortality was not included in the current study, Medicare beneficiary data sets could be added for such an analysis, including both in-hospital and post-discharge deaths.
It is important to emphasize the difficulty of obtaining data measuring the amounts and types of nursing care being provided in acute hospitals (not to mention other settings). Obtaining the data for this study required combining information from multiple national data sources, a very time consuming process fraught with potential problems. The authors recommend that ANA explore options to improve such reporting of nurse staffing data through legislation, regulation or other means.
_______
- Ballard KA, Gray RF, Knauf RA, Uppal P. Measuring Variations in Nursing Care per DRG. Nursing Management, 1993, 24(4), 33-41.
As of September, 2007 - For information about the study, please contact the Department of Nursing Practice and Policy of the American Nurses Association,
8515 Georgia Ave. Silkver Spring, MD 20910
Phone: 1-800-274-4ANA.