Welton, John M. PhD, RN; Zone-Smith, Laurie PhD, RN; Bandyopadhyay, Dipankar PhD
This article poses a fundamental question regarding hospital care: how can we accurately measure the actual hours and direct costs of nursing care delivered to each patient? The question is relevant within the context of emerging healthcare reforms that will emphasize improving quality and value while reducing overall costs. Unfortunately, there is no common metric to measure nursing time and dollars expended for each patient at US hospitals.1 Because nursing care represents perhaps the single largest cost of hospital care, between 40% and 50%, the current practice of allocating nursing care within daily room and board charges is inconsistent with national efforts to better understand and control healthcare costs and quality.2
The following issues raise a number of imperatives for nurse executives and hospital administrators:
* The growing strategy to not pay hospitals for the added costs of adverse events3 creates a compelling need to find not only safe but also optimum levels of nurse staffing tailored to individual patient needs.
* States that have implemented safe staffing laws compel hospitals to use evidence to support their staffing decisions, yet there is no widely used method to estimate and compare nursing intensity levels within or across hospitals or by patient diagnosis.
* Hospitals that are able to measure their "true" care hours and direct nursing costs delivered to each patient can identify opportunities to become more efficient and competitive in the marketplace.
The crux of the problem in addressing these crucial issues is the lack of nursing-specific data about the time and costs of nursing care allocated to each patient within the billing, administrative, and discharge data sets used to set national healthcare finance policy.
In April 2006, the Centers for Medicare and Medicaid Services (CMS) began a series of revisions to the Inpatient Prospective Payment System (IPPS) to more closely align the cost of care with reimbursement and move toward a value-based purchasing arrangement with hospitals.4 Public comments made by nurses regarding the proposed reforms highlighted the need to better understand and incorporate both the costs and intensity of inpatient care. In its response to comments from the nursing profession in its fiscal year 2009 IPPS proposals (CMS-1390-F), CMS agreed that the lack of any allocation of nursing intensity and costs on a per-patient basis is likely distorting the payment system.5 They suggested that the nurses and hospital community recommend a solution to separate nursing from room and board charges to produce a more rational approach to identifying the unique contribution of nurses to inpatient care.
An internal CMS study addressing cost compression within the IPPS recommended separating nursing from daily room and board charge using an existing revenue code to capture nursing incremental charges.6,7 These recommendations were based on 2 previous studies that advocated using the separate code for nursing8 and found that when patient billing is adjusted for nursing intensity there is an improvement of 9.4% in explaining hospital costs.9 The primary argument for implementing this method is that nursing care is unique and separate from medical care in hospitals.10 If nursing care is to be incorporated into the emerging payment reforms such as pay-for-performance and value-based purchasing, a separate accounting of nursing care from room and board charges is needed.
Estimating Nursing Intensity and Direct Costs
The measurement of inpatient nursing intensity presents a challenge. Nearly 50 years ago, Connor11 studied nurse staffing patterns at Johns Hopkins Hospital and found that the use of daily or midnight census is a poor indicator of individual patient needs or staffing assignment patterns. The use of patient classification or nursing acuity systems is equally problematic because of issues with reliability and accuracy.12 In a study that applied data from the same patients using 4 separate patient classification systems, O'Brien-Pallas et al13 found that each instrument yielded a different nursing time estimate, and the spread was as much as 4.5 hours for each patient on any particular day.
Two other methods to identify the intensity and costs of nursing care have been used. The State of New York has used a nursing intensity weight (NIW) to adjust Medicaid payment for the past 2 decades.14,15 These weights are assigned by an expert panel for each diagnosis-related group (DRG) and revised every few years. The advantage of the NIW is the transparency and no additional data needed to calculate a nursing-specific time and cost metric. The disadvantage is that the NIW is not patient-specific and cannot be used to measure actual nursing care time or cost differences between hospitals.
Belgium has used a Nursing Minimum Data Set (B-NMDS) for all hospitals in the country, and the data collected 4 times a year are used to adjust hospital budgets.16,17 Because Belgium is a single-payer healthcare system, this method works well in that country; implementing the B-NMDS at all hospitals in the United States would be difficult because of the many different payers for healthcare and the focus on patient-specific payment rather than budgeting for care received in the hospital.
Searching for the Holy Grail
CMS has set clear boundaries for collecting inpatient nursing intensity data by excluding both NIWs and acuity-based classification systems from consideration.5 CMS also expressed a concern to avoid adding additional data collection chores to already burdened nurses. We need to recognize the trade-off of collecting new nursing-specific data, including the validity, reliability, and accuracy of those data against the potential benefits and costs of collecting the data.18 This presents a quandary how to capture nursing time delivered to each patient so it can be included in the hospital financial record. Any proposed method would have to be universally applied to all hospitals in the United States, reasonably and reliably measure nursing intensity, and not add extra duties to already busy staff nurses.
We propose using the nurse-patient assignment (NPA) record as the initial basis for calculating nursing intensity and direct nursing costs. These data are collected on all inpatient shifts to assign each scheduled nurse to admitted patients. The NPA is an enduring legal document of care and is important in identifying staffing patterns as well as individual responsibility of each nurse caring for patients in a particular setting. The actual nursing care hours delivered to each patient can be calculated using a simple algorithm. For example, if in a 12-hour shift a nurse has an assignment of 4 patients, he/she will be providing, on average, 3 hours, of care to each patient. Table 1 outlines the various calculations based on the assignment pattern. Also, if nursing time for each patient is known, the direct costs can be calculated using average hourly wages.
If such a system were automated in a scheduling and staffing software system, actual nurse wages and other nurse characteristics could be incorporated into the calculation of nursing intensity. Because the NPA data structure would be similar across hospitals, nursing time and cost data for each patient could be entered into the billing mechanism using a software interface.
The NPA method of calculating nursing intensity avoids some of the problems that plague other classification systems. In particular, it represents the care that was given based on the assignment and is not a "black box" estimate subject to classification creep. Patient admission and discharge information can be used to adjust hours of care; for example, if a patient was admitted halfway through the shift, the actual hours can be adjusted accordingly based on the assignment pattern.
The other distinct advantage of the NPA is the ability to link individual nurses with each patient within an electronic data system. The characteristics of each nurse can be identified, such as years of experience, academic preparation, or agency or traveler status. These data can then be used in conjunction with the hospital discharge record to investigate the relationship between nursing intensity and nurse characteristics and hospital-acquired adverse events, length of stay, and hospital outcome. Table 2 illustrates how the NPA can be used to summarize the nursing care for a single patient during the hospital course.
The present study tested the calculation of nursing intensity using the NPA record against an instrument that collected actual nursing intensity hours at an academic medical center. The primary argument is that the NPA can be used to calculate nursing intensity based on the number of patients a nurse is assigned during a shift and is a reasonable proxy of the actual nursing hours delivered to individual patients. Although there are certainly differences in expended nursing time within a particular assignment, the assignment is a reflection of different patient needs and the ability of a particular nurse to address those needs.
The primary objective of the study was to test how well the NPA nursing intensity calculation predicts the variability of the actual patient hours using linear regression. A high level of explained variance (r2) of patient nursing intensity from the NPA estimate indicates both accuracy and measurement robustness in dealing with different patients and case mix. The study also compares nursing intensity calculated using unit mean staffing levels (calculated from nurse-to-patient ratios) and hospital mean staffing levels from the combined medical/surgical floors or ICUs against the actual nursing intensity measured from the existing classification instrument. If the NPA can explain a substantial portion of the variability of the actual patient-level nursing intensity and explain a substantially higher portion of variability compared with unit- or hospital-level mean staffing levels, this is evidence for using the NPA as a standard method for allocating nursing time and costs for individual patients.
This study used a secondary data analysis of sequential patients admitted to a single university hospital in the southeast between January 2004 and June 2005. Each patient-day was entered into a nursing intensity database (NID) where the nurse assigned to each patient entered his/her time expended during a 12-hour day or night shift.19 The NID was used for short-term staffing and assignment assessment as well as evaluating long-term trends in nursing intensity and unit staffing patterns.
In 2003, the Medical University of South Carolina developed a new nursing workload measurement tool to capture nursing resource expenditure for each patient.20 Nurses estimated the amount of direct nursing care hours they expended for each patient during the shift and entered the values into the NID Web-based computer program. The NID validity and reliability were compared with an existing time and attendance data. Analysis of 32 inpatient units with various specialties over an 8-month period (January 2006 through August 2006) resulted in 160,072 patient shift estimates. Shift estimates were then averaged by month for the period, yielding a sample for analysis of 256 observations. Independent and paired-means testing, analysis of variance, covariance, repeated measures, and post hoc analyses found that there was no significant difference between unit mean nursing hours derived from the NID and the time and attendance data.
The NID nursing intensity was summarized for the assignment for each patient-day. The number of patients assigned to each nurse during the shift was calculated using a unique nurse identifier, and then the average time expended for all patients within the assignment calculated according to a formula similar to that in Table 1. Shift assignment totals less than 8 hours or more than 14 hours for any nurse were excluded from analysis. Because there is some variability within the assignment, this constraint helped to eliminate NID entries for multiple patient entries for several nurses by the charge nurse or staffing patterns where a nurse would take a light assignment and have other duties.
The mean nursing intensity was then calculated for each unit for the study period as well as the mean for all medical/surgical units or similarly all ICUs. The final research data set included the day- and night-shift NID nursing intensity rating, the nursing intensity calculated from the NPA, and the mean unit and hospital nursing intensity for the study period. These data were then linked by unique patient identifier with the hospital discharge record to include patient characteristics and discharge information.
Only patients who received care on adult medical/surgical units or adult ICUs were included in the study. Excluded were patients with any primary DRG within the psychiatric or substance abuse Major Diagnostic Categories (MDC 19 or 20). The exclusion of children reflected the high-acuity pediatric and newborn care at an academic medical center that is not typical in most community hospitals.
The analytic approach used ordinary least squares regression to address the research questions. Separate regression models were constructed to predict NID day or night nursing intensity using NPA calculated intensity, unit mean NPA estimate, and the mean NPA calculated estimate for all medical/surgical or ICU units. Explained variance (r2) for the NPA, unit mean, and hospital mean were compared across the models.
The initial data collection and setup used Microsoft Access (Redmond, Washington), and statistical analysis was conducted using SPSS version 14.0 (Chicago, Illinois). The final research database removed all patient and nurse identifiers. The study was reviewed by the Medical University of South Carolina institutional review board and received exempt status.
The final study data set consisted of 11,582 patient-days among 13 medical/surgical and 5 ICUs. Patient demographic data are described in Table 3. Most patients were white (51.8%) and female (60.8%), with 38.1% and 23.3% of the ratings associated with patients on Medicare or Medicaid, respectively. Only 2.9% of all ratings were associated with patients who died.
Table 4 describes the distribution of both the NID estimated nursing intensity and the NPA calculated nursing intensity for day and night shifts. The assignment patterns are also noted. The adult medical/surgical units also had intermediate care (also known as step-down) beds intermingled on the same floor, but it was not possible to differentiate these patients in the final research database. The nurse-to-patient assignment ratio was typically 1:3 for the intermediate patients. The high nursing intensity for the ICU patients is explained by the high percentage of 1:1 staffing.
The difference in nursing care hours estimated by nurses using the NID can be compared with the calculated time based on the NPA (Table 4). For example, day-shift ratings for assignments of 3 or 4 patients corresponded to a mean NID nursing intensity of 3.8 and 3.0 hours, respectively, and these are closely matched to the calculated times of 4.0 and 3.0 for the NPA. The other assignment matches are remarkably close with the exception of 1:1 or 1:2 assignments on the floors and 1:1 assignments in the ICU.
Table 5 compares the regression models assessing the explained variance associated with predicting NID nursing intensity from the assignment calculated NPA, the unit, and hospital mean average for the period. In each of the models, explained variance decreased from the highest for the assignment calculated NPA, then unit, and, finally, hospital means. For the day-shift ratings, the NPA nursing intensity estimates were a 36.7% improvement over the unit mean (r2 = 0.760 vs 0.556) and 79.2% improvement (r2 = 0.760 vs 0.424) over the hospital mean for all medical/surgical or ICU units. For night shift, the improvement was 33.8% (r2 = 0.815 vs 0.609) and 44.9% (r2 = 0.815 vs 0.545), respectively, compared with unit and hospital mean values. One likely explanation between shifts is that night shifts may have less variability than the typical volatile day shifts that handle a large portion of admissions, transfers, and discharge cycles.
The primary objective for this study was to test the feasibility of using the nurse-patient assignment (NPA) as a primary data source to calculate nursing intensity and direct nursing costs. The findings support the initial contention. When compared with an instrument where the nurse estimated his/her actual time expended for each patient (NID), the NPA calculation performed well explaining 76.0% of the variability in NID nursing intensity for day shift and 81.5% for night shift. The mean NID and NPA estimates were closely matched on both the medical/surgical and ICUs. When the unit mean staffing level estimate is used, only 55.6% or 60.9% of the day- or night-shift variance is explained. When all floor or ICU units are pooled, the aggregate estimate explains less than half of the patient-level intensity (42.5%) for day-shift and 54.5% for night-shift.
The findings of this study indicate that the calculation of direct nursing care hours based on the actual NPA is both a feasible and robust measure comparable to a nursing intensity estimate that is much more labor-intensive to collect and burdensome on both nurses and the hospital to use. The NPA estimate is also superior to methods that use mean unit or hospital estimates.
Toward a National Inpatient Nursing Intensity Model
Caution is warranted about generalizing the findings from this study conducted at a single academic medical center to all hospitals in the country. However, there is ample evidence that the NPA estimate of nursing intensity can be used at essentially any inpatient setting. The next step is to conduct a demonstration project at multiple geographically diverse hospitals, both teaching and community facilities with a mix of public and private ownership. The focus of future studies can be directed toward investigating the association between nursing intensity and the characteristics of nurses caring for patient with clinical and financial outcomes of care.
A previous article identified a method to identify high- and low-performing hospitals using daily patient nursing intensity estimates.21 Hospital can and should have a wide degree of variability and flexibility in using various nurse staffing patterns to meet patients' needs. However, if nursing intensity could be collected in a manner that is valid and reliable and does not impose additional burden to hospitals and nurses, those hospitals that have comparatively low nursing intensity (or high workload compared with other hospitals) for a particular DRG are at risk for poor outcomes, as has been widely reported in the literature.22 Yet hospitals that have a very high mean nursing intensity may be less efficient and have added costs of care.
The use of a nursing intensity estimate based on the actual NPA has a number of distinct advantages over previous methods to develop a nursing minimum data set23,24:
1. The data to calculate nursing intensity are readily available at all hospitals.
2. There is an existing mechanism to add nursing intensity into the current billing system to itemize and separate nursing care from room and board charges.
3. By developing a method to capture the NPA in a standardized electronic database, hospitals can examine and analyze how different staffing patterns and nurse characteristics affect the clinical and financial outcomes of care.
4. By capturing essential information about the care delivered by nurses to each patient, the overall resources and effectiveness of nursing care can be evaluated and used to inform administrators and policy makers in future healthcare and health finance reform efforts.
Costing Nursing Care
The additional benefit of having a reasonably simple method of capturing nursing intensity for each patient is to calculate the direct costs of nursing care. This can be done in 2 ways. The mean unit hourly wage of nurses can be used to estimate costs (Table 1). This gives the mean expenditures for each patient-day for each patient rather than using a unit or hospital mean nursing costs per patient day. A second approach would require a direct link between an individual nurse assigned to care for each patient, and the actual hourly wage rather than the mean would then be used to calculate a direct cost estimate. This approach would allow hospitals to evaluate the difference between more experienced and costly nurses and younger, lower-salaried nurses with regard to overall cost, efficiency, effectiveness, and quality. Additional NPA-driven analyses are possible to examine the costs and benefits of using contract nurses (commonly called traveler nurses).
Implications for Nurse Executive and Managers
Nurse managers and executives deal with staffing, budget, and quality issues on a daily basis. With nursing care representing perhaps the single largest cost in any hospital, there is a natural and understandable tendency to decrease staff in times of economic uncertainty. Unfortunately, cutting costs in this manner may have a negative effect on quality and safety. Nurse leaders will be well served by obtaining patient-level nursing intensity and cost data, especially if those data can be directly linked to the hospital billing and discharge data sets used to set payment. An NPA estimate would be relatively straightforward to implement and would provide the same consistent data across any inpatient setting allowing both internal trending analysis and benchmarking performance data across multiple hospital settings.
Nursing intensity and direct nursing costs are not the only contributions of nursing care, and certainly, there are any numbers of additional patient, nurse, unit, and hospital factors or characteristics that influence hospital nurse staffing patterns. The complexity of the problem in collecting additional nursing data that can be linked to each patient is acknowledged, but the NPA method for estimating nursing intensity and costs provides a common foundation and data source that is not currently available. It is a necessary first step toward implementing a national inpatient Nursing Minimum Data Set, and once established in the administrative and billing data sets, additional nursing data could be added to better explain how and why nursing resources are expended to achieve the highest quality and ultimately the best value of care.
What Nurse Executives Need to Tell Their Chief Executive and Chief Financial Officers
How much does nursing cost, or how can hospitals provide higher quality of care at a lower cost? These are eternal questions that are difficult if not impossible to resolve without requisite nursing data. If hospitals were willing to collect nursing intensity and cost data such as the NPA estimate proposed in this study, the following would be possible:
(1) Costing nursing care at the patient level rather than using hospital mean staffing costs makes hospitals potentially more efficient and profitable by identifying "true" nursing costs allocated on a per-patient basis.
(2) Hospitals that can identify their patient-level nursing intensity and costs are likely to be more competitive in the marketplace. For example, negotiating contract rates with specific third-party payers is difficult when nursing costs on the specific diagnoses being negotiated are unknown. Calculating both the current average nursing care hours expended for specific diagnosis (eg, DRGs) allows estimation of direct nursing costs and comparison to actual reimbursement for each diagnosis.
(3) As reimbursement becomes tighter and additional pay-for-performance measures place increasing emphasis on quality of care and avoidance of hospital acquired adverse events, having a direct measure of nursing intensity and costs is helpful in understanding how to optimize nurse staffing patterns.
Obviously, there are a number of other factors that influence patient clinical outcomes. However, the ability to measure direct nursing hours and costs for each patient is a substantial, if not unprecedented, step forward in understanding how nursing resources are best used in the acute care environment.
Limitations of the Study
The primary limitation of this study is the single academic setting. We cannot, in good faith, generalize our findings to all hospitals in the United States-that was not the intent of this study. The other important weakness of the NPA approach to estimating nursing intensity is that although nursing intensity provides a means to understand what nursing time and costs were expended for patient care, it does not give any indication why a patient received that level of care. While we recommend using the NPA approach to establish a fundamental baseline for measuring and recording nursing intensity and costs, this is the initial step in recognizing the unique contribution of nurses to patient care. Further efforts are needed to explicate why specific nursing actions are needed, for example, what treatment or interventions work best for particular patients, what patient symptoms are most amenable to nursing action, and what intermediate outcomes can be achieved by using best nursing practices. The NPA estimate for nursing intensity should be viewed as the first step in a comprehensive solution to understanding optimum nursing care that will dovetail with future healthcare reforms.
The NPA is a ubiquitous data source recording the association between individual nurses and patients and is used in all hospitals throughout the United States. The findings of this study confirm the feasibility of using the NPA approach to calculating nursing intensity. There is an available mechanism to use these data to capture the time and charges for nursing care separate from room and board charges.
The findings of this study also affirm the potential use of an NPA calculated nursing intensity and cost data as a framework for developing inpatient nursing pay-for-performance metrics and the relationship of nurses to patient clinical and financial outcomes. Because the NPA is a common, if not universal, set of data collected at all hospitals, use of these data for multiple purposes should be relatively easy to implement nationwide.
Ultimately, nursing intensity and cost data derived from the NPA could provide a basis for analyzing, comparing, and understanding how nursing resources are expended for patient care in hospitals, informing administrators, payers for healthcare, and policy makers and can be incorporated into the emerging efforts to produce a high-value healthcare system in the United States and establish the economic value of nursing care.
The authors thank Mary Hughes-Fisher, RN, MSN, for her assistance in preparing and administration of the NID and supply data for this study. The authors also thank the nurses of the Medical University of South Carolina for their diligence and support for data collection and the NID project.
1. Welton JM. Paying for nursing care in hospitals. Am J Nurs
2. Welton JM, Harris K. Hospital billing and reimbursement: charging for inpatient nursing care. J Nurs Adm
3. Kurtzman ET, Buerhaus PI. New Medicare payment rules: danger or opportunity for nursing? Am J Nurs
4. Ginsburg PB. Recalibrating Medicare payments for inpatient care. N Engl J Med
5. Centers for Medicare & Medicaid Services. Medicare program; changes to the hospital inpatient prospective payment systems and fiscal year 2009 rates. Fed Regist
6. Dalton K, Freeman S, Bragg A. Refining Cost to Charge Ratios for Calculating APC and MS-DRG Relative Payment Weights
. Research Triangle Park, NC: Research Triangle Institute; 2008. RTI project number 0209853.008.
7. Dalton K. A Study of Charge Compression in Calculating DRG Relative Weights
. Research Triangle Park, NC: Research Triangle Institute; 2007. RTI project number 0207964.012.008, Prepared for Centers for Medicare & Medicaid Services, Office of Research, Development, and Information.
8. Welton JM, Zone-Smith L, Fischer MH. Adjustment of inpatient care reimbursement for nursing intensity. Policy Polit Nurs Pract
9. Welton JM, Dismuke CE. Testing an inpatient nursing intensity billing model. Policy Polit Nurs Pract
10. Welton JM, Halloran EJ. Nursing diagnoses, diagnosis related group, and hospital outcomes. J Nurs Adm
11. Connor RT. A Hospital Inpatient Classification System
[doctoral dissertation]. UMI: 60-3319 ed. Baltimore, MD: Johns Hopkins University, Industrial Engineering Department; 1960.
12. Shaha SH, Bush C. Fixing acuity: a professional approach to patient classification and staffing. Nurs Econ
13. O'Brien-Pallas LL, Cockerill R, Leatt P. Different systems, different costs? An examination of the comparability of workload measurement systems. J Nurs Adm
14. Ballard KA, Gray RF, Knauf RA, Uppal P. Measuring variations in nursing care per DRG. Nurs Manage
15. Knauf RA, Ballard K, Mossman PN, Lichtig LK. Nursing cost by DRG: nursing intensity weights. Policy Polit Nurs Pract
16. Sermeus W, Delesie L, Van den HK, Diya L, Lesaffre E. Measuring the intensity of nursing care: making use of the Belgian Nursing Minimum Data Set. Int J Nurs Stud
17. Van den HK, Diya L, Lesaffre E, Vleugels A, Sermeus W. Benchmarking nurse staffing levels: the development of a nationwide feedback tool. J Adv Nurs
18. Finkler SA. Measuring and accounting for the intensity of nursing care: is it worthwhile? Policy Polit Nurs Pract
19. Zone-Smith L. Predicting real-time nurse staffing needs in hospitals: a new tool to measure nursing intensity. J Nurs Adm
20. Welton JM, Zone-Smith L, Hughes M, Stovall J, Foulois B, Crummer MB. Nurse intensity database pilot study. In: Marin HF, Marques EP, Hovenga EJS, Goossen WT, eds. Proceedings, 8th International Congress in Nursing Informatics, Rio de Janeiro, June 20-25, 2003
. Rio de Janeiro, Brazil: E-papers Servicos Editoriais Ltd; 2003:123-128.
21. Welton JM. Implications of Medicare reimbursement changes related to inpatient nursing care quality. J Nurs Adm
22. Kane RL, Shamliyan TA, Mueller C, Duval S, Wilt TJ. The Association of Registered Nurse Staffing Levels and Patient Outcomes: systematic review and meta-analysis. Med Care
23. Werley HH, Lang NM, Westlake SK. Brief summary of the Nursing Minimum Data Set conference. Nurs Manage
24. Werley HH, Devine EC, Zorn CR. Nursing needs its own minimum data set. Am J Nurs
© 2009 Lippincott Williams & Wilkins, Inc.