A review of workload measures: A context for a new staffing methodology in Western Australia

1. Introduction 

Determining a sufficient number and mix of nursing staff to ensure safe patient care remains one of the most fundamental and important decisions made by nurse managers at all levels in an organisation. In more recent times, these decisions have received greater attention with several landmark studies clearly establishing the impact that nurse staffing can have on patient outcomes (mortality and morbidity) (Aiken et al., 2002, Estabrooks et al., 2005, Needleman et al., 2002, Rafferty et al., 2007, Aiken, 2002). Work such as this should now lead to more attention being given to the design and implementation of staffing methods which ensure patient safety and an appropriate workload for nurses. However, there remains very little evidence to support and guide staffing decisions and staffing methods have not been evaluated from a patient outcome perspective (Kane et al., 2007).

2. Background and literature review 

Measuring nursing workload, a prerequisite to identifying adequate nurse staffing levels, is difficult and complex. Despite over three decades of research and discussion in the literature there is not a widely accepted workload measure. The invisible nature of nursing, where work once performed disappears, makes it difficult to measure (Duffield et al., 2006). Early and subsequent reviews of the literature have identified that a number of broad approaches to determining nurse demand have developed (Arthur and James, 1994, Duffield et al., 2006, Hurst, 2003). Hurst (2003) identified five nursing workforce planning systems in general use. The first method, professional judgement, is similar to the consensus approach described by Arthur and James (1994). This approach involves intuitive or consultative methods which rely on professional judgement and making subjective decisions about the appropriate number and mix of nurses (Hurst, 2003). This method utilises the experience and professional knowledge of the manager and is quick and simple to use, but it also provides opportunities for significant variation between wards and hospitals. However it does not make the link between quality and staffing levels transparent.

The second method is nurses per occupied bed, the top down approach. Again this is a relatively simple and quick method of calculating staff needs. However it relies on the initial establishment of base staffing having been appropriate. Top-down management approaches were also described by Arthur and James (1994) as the utilisation of staffing norms or a staffing formula, generally determined by professional groups or national bodies. Hours of care per patient day (Holcomb et al., 2002) and nurse-to-patient ratios (AIRC, 2000, Hodge et al., 2004) have also been used as top-down approaches: Both of these tend to reflect minimum staffing requirements which lack sensitivity to local situations (Arthur and James, 1994). The underlying staffing formulae establish a statistical relationship between variables that measure activity such as throughput, bed numbers, patient case mix and nurse staffing requirements (Hurst, 2003). While this approach is more consistent and less subjective, the major concern is that the model assumes current or mandated staffing levels are an appropriate base from which to project future needs. In addition, while relatively simple and quick to use, not all hours worked by nurses are used to provide direct patient care. Consequently these approaches tend to be more of a method of allocating nursing resources to each patient without regard to patient need or complexity (Hodge et al., 2004).

The third method is the acuity-quality method utilising patient dependency systems. A recent example of this method is the AUKUH Acuity/Dependency Tool developed by the Association of UK University Hospitals (AUKUH, 2007). This tool was launched in November 2007 to help National Health Service hospitals measure patient acuity and dependency. Over time the goal is to have sufficient data to measure changes and trends due to seasonality, changing demographics and health needs. This in turn would enable evidence-based decision-making on nurse staffing and workforce planning. The acuity-quality method relies on patients being classified by ‘dependency’ from which nursing requirements are then determined. The model relies on the premise that patient dependency is an accurate measure of the need for nursing time and as before, that the initial staffing requirements based on dependency are accurate. Also, patient classification methods tend to be task oriented and based on nursing activity analysis which are time consuming and expensive to develop (Adomat and Hewison, 2004, McGillis Hall and Doran, 2004). While this method can incorporate professional nursing judgment, a large number of dependency categories can make it a complex and potentially inaccurate process. It is also feasible that nurses could ‘manipulate’ the patient category to reflect the level of care nurses think the patient should have (Adomat and Hewison, 2004, McGillis Hall and Doran, 2004).

The fourth method is timed-task/activity method. The type and frequency of nursing interventions documented in nursing care plans become the predictor of nurse staffing requirements (Hurst, 2003). This approach underpins the various commercial software packages available such as Excel Care, E-care, TrendCare® and GRASP. However, nurses experience a number of unanticipated delays, such as waiting for responses from others which often leads to the re-sequencing of work, or changes in patient acuity requiring immediate but unanticipated additional attention from nurses. Changes in the nursing team composition or skill mix can also result in unanticipated delays. Collectively these complexities make it difficult for nurses to provide the interventions and care identified by the nursing care plan in a timely manner (Duffield et al., 2006).

The fifth and final method is regression analysis. Regression methods predict the number of nurses required for any given level of activity. For example increased bed occupancy would drive the need for more nurses which in turn, would be modelled in the regression. While this model does provide some independent evidence, it is difficult to include all variables that might predict nursing requirements as they are likely to be great in number (Hurst, 2003). In addition more recent work (Hurst, 2005, Hurst, 2008) has identified that ward design and size have a major impact on nursing workload and also, the quality of outcomes for patients. Nightingale and racetrack ward designs support higher levels of direct care than do other types of ward design (Hurst, 2008). Larger wards with fluctuating workloads tended to have poor quality of care (Hurst, 2005). Design and size also now need to be taken into account in the measurement of nurse workload. “The future of nurse demand methods, like the past, will be determined by developments in government policy, nursing, the health service and technology.…a perfect tool for measuring nursing work is unlikely to exist” (Arthur and James, 1994, p. 564).

Edwardson and Giovannetti (1994) reviewed nursing workload measurement systems in some very early evaluation work. All systems had one aim—to estimate the total hours of nursing staff required to care for patients. However, they found there was a lack of rigor in reliability and validity testing in all the systems developed, a view supported later by Hughes (1999) who also questioned the theoretical base of most systems. A comparison of three methods of workload estimates, GRASP, Project Research in Nursing (PRN) and Medicus found PRN consistently gave higher estimates of total nursing hours required when compared to GRASP or Medicus (Hughes, 1999, O’Brien-Pallas et al., 1988). These authors suggested that the weights assigned to particular activities were the major reason for the variation. This study became the impetus for one of the most carefully prepared and thorough analyses of different systems by Thibault (Edwardson and Giovannetti, 1994). Each of the developers of three systems—PRN 76, GRASP and Medicus provided material in response to an evaluation grid consisting of scientific parameters such as operational definitions, reliability, validity and sampling and administrative applications such as application to specialty areas and components of nursing workload measures. On the basis of this material the strengths and weaknesses of the three systems were determined. The degree of consistency among four workload measurement systems, PRN, GRASP, Medicus and Nursing Information System for Saskatchewan (NISS) was also examined (O’Brien-Pallas et al., 1992). All systems produced statistically and clinically significant variations in the hours of care. These differences were up to half a nursing shift per day overall and greater than one shift per day in Intensive Care.

Similar early work was also undertaken in the UK where four nursing workload measurement systems were reviewed. Again the estimates of nursing hours required were substantially different from each other for no apparent reason and the differences could not be explained in terms of any other aspect of the nursing process (Carr-Hill and Jenkins-Clarke, 1995). In addition there was a lack of understanding of the phenomenon being measured and no assessment of reliability or validity.

It is obvious that much of the literature related to workload management tools and their evaluation was published in the 1980s and 1990s. It is quite likely that drivers of nursing workload have changed significantly in the intervening years (Duffield et al., 2007). The application of these tools in the 21st century may be questionable. However there is little in the more recent literature to further inform the development of workload management tools.

Measuring demand for nursing services in Australia is further compounded by the fact that management information routinely collected does not have detailed patient level information from which to do so. Consequently, several approaches have been taken to the measurement of workload including patient classification systems, DRG nurse costing models, hours of care per patient day, nurse-to-patient ratios and a number of commercial packages in use (Duffield et al., 2006). Early patient dependency (classification) systems attempted to identify the demand for nursing resources based upon completion of specific nursing activities. This ‘task’ approach to assessing the need for nursing resources was criticised for representing nursing as a series of time limited tasks rather than an iterative process of providing nursing care. In addition, this approach is very labour intensive and requires extensive data recording by nurses. More modern patient dependency systems attempted to recognise the process of nursing care provided, by not only measuring the nursing tasks, but by also applying weights to the tasks to include risk, skill mix and complexity factors. However despite these developments, widespread dissatisfaction with patient dependency systems remains (Gerdtz and Nelson, 2007). Nurses have found that computerised patient information systems neither enhance clinical practice or patient care (Darbyshire, 2004). A phenomenological study involving 13 focus groups and 53 practitioners described primarily negative experiences of computerised patient information systems. Nurses perceived the systems were unable to capture ‘real nursing’, were non-responsive, difficult to use and irrelevant to patient care and meaningful clinical outcomes (Darbyshire, 2004).

Nursing costing models based on Diagnostic Related Groups (DRGs) provide a nursing service weight for a particular DRG estimating the typical nursing resources required for this type of patient. While these systems do not require direct data entry from nurses, they are criticised because they do not capture other determinants of nursing workload or any day-to-day variation in-patient need (Duffield et al., 2006). Nurse hours per patient day and nurse-to-patient ratios as workload measures are also criticised. They tend to rely on historical data to determine staffing and consequently do not take into account changes in care practices or patient acuity. In addition, the underlying assumption that all patients and all patient days are equal is challenged (Graf et al., 2003). The need for nursing care varies significantly between different patients but also, as the patient progresses through their recovery. In addition the intensity of patient care increases as the length of stay is shortened. Consequently these measures may give inadequate estimates of nursing care requirements (Graf et al., 2003). Nurse staffing requirements are driven by a number of factors in addition to patient acuity. Length of stay, the number of admissions, discharges and transfers, the manager's clinical judgement, staff competencies, ward geography and medical practice patterns are multiple dimensions that influence any staffing system (Van Slyck, 2000).

In the absence of universally accepted workload measures, unit level managers have tended to utilise clinical judgement when making decisions about staffing (Arthur and James, 1994, Hurst, 2003). These decisions could be influenced by a myriad of factors including cost pressures, hospital accepted norms, patient acuity, ward turnover and availability of nursing staff. Senior nurse executives often find themselves in a position of defending what they believe are required nursing staffing levels without accepted workload measures to support these levels. In addition, many benchmarking activities focussed on reducing staffing levels to the lowest level rather than determining what was needed in any patient population (Aiken et al., 2000).

The many workload measures in use are yet to meet the needs of those nurses who have the day-to-day accountability for providing adequate nurse staffing to secure appropriate patient outcomes. Questions remain about the theoretical base from which the workload measures are derived and there is too often a lack of understanding of the phenomenon being measured (Carr-Hill and Jenkins-Clarke, 1995, Edwardson and Giovannetti, 1994, Hughes, 1999). Nevertheless, the literature demonstrates a continued proliferation of systems designed to measure nursing workload despite the discussion and concerns in regard to reliability, validity and comparability (Fagerstrom, 1999, Graf et al., 2003, Harrison, 2004, Rauhala and Fagerstrom, 2004, Walts and Kapadia, 1996, Yamase, 2003). However, the association between patient outcomes and adequate nurse staffing (Aiken et al., 2002, Estabrooks et al., 2005, Kane et al., 2007, Needleman et al., 2002, Rafferty et al., 2007) makes the challenge of effectively measuring nursing workload vitally important. In Western Australia, a nurse hours per patient day staffing method was developed which attempted to address some of the concerns identified in the literature in regard to workload management while also responding to industrial and political imperatives.

3. The context 

Western Australia is the largest State in Australia covering 2,529,875 square kilometres. The population is 2,003,800 with over 1.2 million residing in metropolitan Perth, the capital city. The Department of Health has overall responsibility for funding and managing the public hospitals in this State. The metropolitan area has three adult teaching hospitals with 1449 beds, specialist women's and children's hospitals, (398 beds) specialist mental health (199 beds) and six general hospitals (1020 beds). These hospitals are managed under an Area Health Service Structure. Country health services cover all areas in Western Australia outside of metropolitan Perth and face the difficulty of providing services over vast stretches of land with low-density population. There are also a number of private hospitals providing services to those with private health insurance.

A feature unique to Australia is the means by which Australian public sector employees’ wages and working conditions, including those for nurses, are determined by the Australian Industrial Relations Commission (AIRC). The AIRC was established by the Australian Government and functions under principal legislation of the Workplace Relations Act 1996. The main objective of the Act was to “provide a framework for cooperative workplace relations which promotes the economic prosperity and welfare of the people of Australia…” (AIRC, 1999, p. 2). The Act gave the AIRC a range of powers of which most relevant for this paper is to prevent and settle disputes, preferably by conciliation or as a last resort by arbitration (AIRC, 1999). It is through such an arbitrated process that the Western Australian NHPPD staffing method evolved.

4. Australia and the WA model 

Nurses’ workload was given greater prominence following the release of several state and national reports on its impact on workforce retention (AIRC, 2000, AIRC, 2002; Commonwealth of Australia, 2002; Department of Education Science & Training, 2002). Victoria was the first State required to address the issue. In 2000, the industrial body for nurses, the Australian Nurses Federation (ANF), made nurse-to-patient ratios a major part of their negotiations and undertook prolonged industrial action as part of their campaign. The ANF cited California, the first state in the United States to adopt legislation mandating minimum unit-based licensed nurse-to-patient ratios (Donaldson et al., 2005) as an example of a suitable staffing method. The Victorian Department of Health could not reach agreement with the ANF. Consequently an arbitrated outcome by the AIRC resulted in the introduction of nurse-to-patient ratios as a method of measuring nursing workload. In acute care hospitals a ratio of one registered nurse to four patients was established at ward level on the morning and afternoon shifts and one nurse to eight patients on the night shift. Nurses also have the ability to close beds if the staffing ratio is not reached. However, as indicated earlier, a weakness of this model is that ratios themselves cannot identify the precise nursing hours required at any particular time in any particular setting (Duffield et al., 2006, Gerdtz and Nelson, 2007). Nor are ratios sensitive to other variables that impact on nurse staffing needs such as ward turnover, staff competencies, geography of the ward and medical practice patterns (Duffield et al., 2006, Hurst, 2008). Many of these dimensions are addressed in the WA nurse hours per patient day staffing method.

In WA, the government and its publicly funded hospitals were concerned about the disruption caused to the health system in Victoria and the likelihood that ratios would also be mandated in this State. In its findings the Victorian AIRC commented that the hospital networks had opportunities to provide alternatives to the nurse-to-patient ratios proposed by the nurses’ union but had failed to do so (AIRC, 2000). This limited the options available to the AIRC in handing down a decision. In a proactive response a working party of senior nurse leaders was formed in WA to provide an alternative staffing approach. As the result of an arbitrated process, the nurse hours per patient day (NHPPD) staffing method, but with an approach (described below) never used before, was mandated for use in this State's public hospitals (AIRC, 2002). The Union had made nursing workload a key political and industrial issue and the AIRC decision was arrived at in this context. Patient care and patient outcomes were not considered in the development of the NHPPD staffing method or the AIRC decision to mandate its use. Nevertheless, this approach remains the primary means of determining nurse staffing requirements in the State.

The working party established guiding principles to determine safe staffing by category of ward utilising three sources of information, some of which were based on earlier work (Van Slyck, 2000). Firstly, national benchmark data were provided by a consultant to determine nursing staffing levels in metropolitan and country (rural or regional) services in WA. These data established the relative position of WA's staffing levels in public hospitals compared to other states. Utilising national benchmark data assisted in addressing the concern expressed previously by avoiding reliance on historical local hospital data (Arthur and James, 1994, Hurst, 2003). The second source of information was expert opinion involving nurse executives and the work of the Metropolitan Directors of Nursing Council. This approach enabled tapping into the professional judgement of senior nurses, not only in regard to historical trends but also current identified pressures and future needs (Arthur and James, 1994, Hurst, 2003, Van Slyck, 2000). However, this consensus approach was prone to significant variation between wards and hospitals. The third source of information was reference to published literature available at the time. Patient related activities, patient acuity, emergency and elective patient admissions were also examined when considering the drivers of nursing workload utilised by others in developing staffing methodologies (Beglinger, 2006). Unfortunately, what is now a clear link between nurse staffing and patient outcomes was less apparent at the time of development of the model. It was not until 2004 that the quality literature fully recognised the importance of nursing in patient safety. The landmark report commissioned by the U.S. Department of Health and Human Services’ Agency for Healthcare Research and Quality, titled Keeping Patients Safe: Transforming the Work Environment of Nurses (Page, 2004) identified the central role nurses play in patient safety. However, this work was published after the development of the staffing method. Consequently, patient outcomes were not included in the development of this method or in its future evaluation.

The NHPPD staffing method grouped wards and allocated nurse hours based on a number of factors such as the presence or absence of high dependency beds, the mix between emergency and elective services and in mental health wards, characteristics such as risk of self harm and aggression. Berlinger (2006) has since identified a number of similar variables as drivers of nurse workload which include: length of stay; admission, discharge and transfer activity (ward turnover); age of patients; clinical conditions and interventions, high dependency care within the ward; and ‘sitters’ for patients who would be unsafe if left alone.

Seven ward groupings were developed (refer to Table 1) with each category allocated average nurse hours per patient day. The NHPPD staffing method different ward groupings were derived from a mix of descriptive attributes and quantifiable and measurable attributes. The staffing method was then tested using the benchmark data collected previously.

Table 1. NHPPD guiding principles (incorporating mental health inpatient units)

	Ward category	NHPPD	Criteria for measuring diversity, complexity and nursing tasks required
	A	7.5	High complexity
			High dependency Unit @ 6 beds within a ward
			Tertiary step down ICU
			High intervention level
			Specialist unit/ward tertiary level 1:2 staffing
			Tertiary paediatrics
			Mental health—high risk of self harm and aggression
			– Intermittent 1:1/2 Nursing
			– Patients frequently on 15 minutely observations
	B	6.0	High complexity
			No high dependency unit
			Tertiary step down CCU/ICU
			Moderate/high intervention level
			Special unit/ward including Mental Health Unit
			High Patient Turnovera >50%
			Paediatricsb
			Secondary paediatrics
			Tertiary maternity
			Mental health—high risk of self harm and aggression
			– Patients frequently on 30min observations
			– Occasional 1:1 nursing
			– a mixture of open and closed beds
	C	5.75	High complexity acute
			Care unit/ward
			Moderate patient turnover >35%, OR
			Emergency patient admissions >50%
			Mental health—moderate risk of self harm and aggression
			– Psycho geriatric mental health unit
	D	5.0	Moderate complexity
			Acute rehabilitation secondary level
			Acute unit/ward
			Emergency patients admissions >40% OR
			Moderate patient turnover >35%
			Secondary maternity
			Mental health—medium to low risk of self harm and aggression
	E	4.5	Moderate complexity
			Moderate patient turnover >35%
			Sub acute unit/ward
			Rural paediatrics
	F	4.0	Moderate/low complexity
			Low patient turnover <35%
			Care awaiting placement/age care
			Sub acute unit/ward
			Mental health slow stream rehabilitation
	G	3.0	Ambulatory care including:
			Day surgery unit and renal dialysis unit

a Turnover=admissions+transfers+discharges divided by bed number. b Paediatrics additional formulae: birth; neonates; emergency; and operating room.

Table 1 outlines the ward categories with descriptors and the nursing hours per patient day allocated to each. Category A wards had high complexity patients with a high level of nursing interventions, high dependency units within the ward and received patients as an immediate step-down from Intensive Care. Category B wards were very similar to Category A except they did not have a high dependency unit within the ward. A ward could also fall into Category B if it had an average daily patient turnover of greater than 50%. Category C wards were also categorised as acute high complexity wards with moderate patient turnover of greater than 35% or emergency admissions greater than 50%. Category D wards were characterised as moderate complexity, often involving acute rehabilitation. They were expected to have emergency patient admissions of greater than 40% or moderate patient turnover of greater than 35%. Category E wards were characterised as having moderate complexity, often being sub-acute and with moderate patient turnover of less than 35%. Category F wards were characterised as moderate to low complexity with low patient turnover such as patients awaiting placement into residential care units. The final Category G was related to ambulatory care settings such as day surgery and renal dialysis units.

All public hospitals were advised about the approved full time equivalent (FTE) increase in nursing positions resulting from application of this staffing method. As this was a solution to an industrial problem (nurses’ workload) it is not surprising that subsequent evaluation has been limited to the impact on staff numbers and recruitment.

5. The impact 

Significant staffing increases resulted from the introduction of NHPPD which was phased in over a 6-month period after the method was mandated in March 2002 (Department of Health, 2006). Most hospitals placed greater emphasis on recruitment including strategies such as overseas recruitment, offering flexible rostering patterns and provision of family friendly initiatives. In the short term hospitals also supplemented any staffing shortages with casual and agency nurses on a shift-by-shift basis (Department of Health, 2006). As a result of these strategies metropolitan health services were at or within 10% of target staffing levels within 6 months of the introduction of the staffing method (Department of Health, 2003).

Hospital recruitment initiatives were also supported by a policy change whereby the Department of Health established its own agency to supply short-term relief staff using a contracted provider and a fixed fee structure. As a consequence the financial incentive for many nurses to work for an agency was minimised. Wards that experienced ongoing difficulties recruiting permanent staff were also able to access short-term relief staff from the Department of Health agency until permanent nursing staff could be recruited. Country health services had more difficultly in recruiting additional staff and took a longer period to reach the new staffing levels (Department of Health, 2003). The AIRC was silent about skill mix (AIRC, 2002) unlike the Victorian decision which mandated the staffing ratios based on a registered nurse workforce (AIRC, 2000). The only other experience of mandating skill mix as part of ratios was in California which allowed for up to 50% of the mandated licensed nurses to be licensed vocational nurses, the equivalent of an enrolled nurse (Donaldson et al., 2005). Consequently the mix between registered and enrolled nurses was not mandated in WA. However the decision did require staffing increases to consist of nurses licensed to practice rather than carers or non-licensed roles (AIRC, 2002).

Importantly, the staffing method provided nurse executives with an agreed and mandated staffing profile based on more than historical data or professional judgement. An increase of 313.18 FTE nurses was approved for implementation in ward areas across the State's public hospitals (Department of Health, 2005). This increase equated to a 3.47% increase in staffing with the majority of the increased staff numbers occurring in the teaching hospitals (86.10% of the total FTE allocated) (Department of Health, 2005). To put this in context, there were 7136 productive nursing FTE in WA public hospitals at the time these changes occurred. The skill mix of the nursing workforce remained relatively unchanged over a 2 year period before and after implementation of NHPPD with 88.7% registered nurses and 11.3% enrolled nurses (AIHW, 2004).

These agreed staffing increases did deliver improved staffing outcomes. They helped reverse a worrying trend of very high agency usage with nurses leaving the public hospital system because of significant workload pressures (Department of Health, 2006). Vacancy rates within public hospitals began to decline. Productive hours of permanent nurse staffing increased by 3.65%. Agency usage initially increased to meet demand: However within 2 years this had declined by 16.82% as nurses returned to the permanent workforce (Department of Health, 2006). At this time the average cost of an agency nurse was $A89, 415 compared to the average cost of a nurse employed by the health services ($A57, 685). Consequently, a 16.82% reduction in agency usage during a period when there was significant growth in the nursing workforce, represented significant savings to the Government and its hospitals.

Introduction of this staffing method also assisted in staff retention as nurses experienced increased staffing levels which in turn, impacted positively on workload and their capacity to provide quality care. For nurse executives, this method also enabled them to increase the number of FTE nursing staff in an environment where the Government was seeking increased efficiency and cost reductions. This approach has since been expanded into other areas of practice such as the Emergency Department, Intensive Care, Coronary Care and Operating Theatres.

6. Review of the staffing method, nurse hours per patient day 

The nurse hours per patient day staffing method did address some but not all of the concerns about workload measures identified in the literature. It did remove subjective determinations of adequate staffing identified as a concern (Arthur and James, 1994, Hurst, 2003, Van Slyck, 2000) by quantifying and grouping wards around similar patient types and workload drivers such as high dependency, occupancy, ward turnover and emergency and elective mix. This has ensured similar staffing levels for all ‘like wards’ across hospitals. This approach also addressed to some extent the concern about a lack of sensitivity to specific ward circumstances (Graf et al., 2003). Every ward was reviewed on its individual data and descriptive detail. Consequently, if a unique set of ward characteristics such as turnover, emergency/elective split, and patient type existed in a ward, it was assessed as part of the determination of nursing hours. In addition, because the nurse hours per patient day utilised average staffing requirements it avoided the pitfall of being seen as setting a minimum or maximum staffing level, giving much greater day-to-day staffing flexibility (Gerdtz and Nelson, 2007). Also, the utilisation of national benchmarking staffing data prevented the model assuming current staffing levels were an appropriate base from which to project future needs (Arthur and James, 1994, Hurst, 2003). Given the staffing increases achieved this clearly was not the case (Department of Health, 2006).

However, the development and introduction of nurse hours per patient day failed to address two key areas. The acuity and intervention level was primarily determined by descriptive means relying on the Director of Nursing from the hospital and the implementation team reaching agreement. The staffing method would be enhanced if regular patient acuity and intervention levels were measured by a standardised tool such as the AUKUH Acuity Dependency Tool (AUKUH, 2007). The other major limitation of the staffing method is that there has not been any attempt at this stage to evaluate its impact on patient outcomes. This method would not be alone in that regard. No studies to date have ‘primarily empirically examined specific nurse staffing policy’ (Kane et al., 2007, p. 1).

7. Conclusion 

The Western Australia Nursing Hours per Patient Day staffing method recognises that ward acuity and ‘business’, both of which impact on nurses’ workload and hence staffing needs, relate to more than just individual patient needs (Twigg, 2001). Its introduction improved staffing levels, reduced reliance on agency nurses and increased staff retention, outcomes which are monitored and reported at 6 monthly intervals in compulsory meetings between the Government and relevant industrial organisations.

Importantly for nurse executives, this method offers some assurance that sufficient resources are provided without imposing a restrictive shift-by-shift nurse-to-patient ratio. In this way it facilitates the use of professional discretion over 24h 7 days a week to enable diversion of resources to areas of greatest need. It does this within general parameters set by capturing drivers of nurses’ workload from a number of sources. Predicted shortages of nurses and the aging of the workforce make it imperative that appropriate and validated tools for measuring nursing workload are in place to ensure patient and nurse safety. The Western Australia staffing method is a key feature of Government policy and continues to be utilised to demonstrate ongoing management of nurses’ workload and patient care requirements. However, evaluation of this staffing method is required, particularly with respect to the different ward categories and the decision rules built into the model. Any such evaluation needs to examine the impact on patient outcomes such as satisfaction or nursing sensitive outcomes.

Conflict of interest 

The authors (Di Twigg and Christine Duffield) have no actual or potential conflict of interest including any financial, personal or other relationships with other people or organisations within 3 years of beginning the submitted work that could inappropriately influence, or be perceived to influence, their work.

Funding 

Funding of $A11,000 was received from The Western Australian Nurses Memorial Charitable Trust.

Ethical approval 

Ethical Approval was received from the University of Technology, Sydney Human Research Ethics Committee.