Using patient dependency to calculate staffing needs in A&E. This is an extended version of the article published in Nursing Times; 104: 15, 29-30.
Alison O’Brien, RGN, is sister; Jonathan Benger, MD, is consultant; both at emergency department, United Bristol Healthcare NHS Trust.
O’Brien, A., Benger, J. (2008) Using patient dependency to calculate staffing needs in A&E. This is an extended version of the article published in Nursing Times; 104: 15, 29-30.
Background: The number and grade of nursing staff is often determined by historical or arbitrary means. However, this fails to take account of dependency and, therefore, real nursing workload.
Aim: To assess patient dependency in A&E at an inner-city teaching hospital, thereby informing nursing workforce decisions.
Method: The Jones Dependency Tool (JDT) was used to collect data on all patients attending adult A&E over a two-week period. The primary outcome measure was patient dependency on arrival and departure and at four and eight hours if the patient was still in the department.
Results: 10% of patients fell into the high- or total-dependency groups on arrival. Patients tended to become less dependent during their A&E stay. The proportion of high- and total- dependency patients did not vary by day of the week, but increased significantly overnight.
Discussion: The results highlight the weakness of basing staffing levels on patient numbers alone. With further work the JDT can be used to predict workload, resource use and optimal staffing levels to provide safe and effective patient care.
Conclusion: Patient dependency is one of the essential determinants of nursing grade mix, and can be readily and repeatedly assessed. We recommend this approach to other clinical settings.
As the single largest professional group in the NHS, nurses are the most affected by clinical, educational and managerial developments in health and social services (Hurst et al, 2000). Nursing salaries account for more than one-third of NHS revenue expenditure in the UK (Wanless, 2002).
Numerous attempts have been made to ensure that nursing teams are as efficient as possible (Gibbs et al, 1991), but reports have concluded that there is still considerable room for improvement in the efficiency and effectiveness of staff deployment (Audit Commission, 2001).
Yet the number and grade of nursing staff in a particular clinical area is often based on historical or arbitrary measures rather than any formal assessment of patient numbers and need.
Overstaffed, understaffed and imbalanced nursing teams have implications for the quality and cost of patient care. Poorly configured nursing teams will also jeopardise nurses’ job satisfaction, and affect the delivery of adequate and appropriate nursing care (Furlong and Ward, 1997).
It is therefore apparent that nursing ‘grade mix’ (defined as the number of nurses and nursing assistants required in a team) has major implications for both cost and quality of care (Standing Nursing and Midwifery Advisory Committee, 2002).
A good example of this occurs in A&E, where the number of nurses on each shift has traditionally been based on the number of patients expected to attend the department during that time. While this is clearly an important factor, patient volume takes no account of patient dependency, that is, the specific needs of each patient, and how much nursing time they will require.
In a busy clinical environment, patients’ stability changes hourly, as does nursing workload and the staff required to meet these ever-changing demands.
The SNMAC has recently emphasised the importance of establishing the right number and grade mix of nurses (and other healthcare professionals), rather than attempting to make decisions based purely on financial measures (SNMAC, 2002).
Increasing concern over quality assurance in nursing led many nursing managers to address the relationship between staffing numbers, grade mix, workload and standards of care during the 1990s (Furlong and Ward, 1997). Several attempts have also been made to examine the process of nursing workforce planning, and assist in determining an appropriate grade mix (Ball et al, 2004; Browne and Odell, 2004; Hurst, 2003; Buchan, 2001). However, a means of reliably assessing the dependency of A&E patients did not become available until 2001, when the Jones Dependency Tool (JDT) was refined and validated (Jones, 1990). This provides an opportunity to move away from staffing levels based on historical data or patient numbers to a more objective assessment of nursing workforce requirements.
The study’s aim was to establish current patterns in the dependency of patients attending A&E at the Bristol Royal Infirmary using the JDT. The intention was to better inform decisions regarding nursing numbers and grade mix, and to act as a benchmark for future developments and changes in emergency care. We recommend a similar process for other clinical environments to ensure that every ward has the number of nurses it really needs.
This study took place in the Bristol Royal Infirmary adult A&E department over 14 days. The Royal Infirmary is a large inner-city teaching hospital, and the adult A&E sees approximately 1,000 new patients every week.
The current nurse staffing level at the hospital is 70 full-time equivalents, which includes HCAs, two training and development posts and an emergency nurse practitioner service. The optimal staffing number is currently considered to be 75 full-time equivalents, with the BRI’s shortfall mainly in band 5 nurses.
The JDT was developed through extensive observational work of the factors associated with patient dependency in the general emergency care setting. It consists of two sections: A and B. In section A, six key component headings are each rated on a three-point scale (see Box 1). The six headings are: communication; airway/breathing/circulation; mobility; eating/drinking/elimination and personal care; environmental safety, health and social needs; triage category. This gives a total score ranging from 6 to 18, which is translated in section B into one of four overall levels of patient dependency.
Box 1. The Jones Dependency Tool
|Airway, breathing and circulation (ABC)||
|Eating, drinking, elimination and personal care||
|Environmental safety, health and social needs||
A data collection form was developed that incorporated section A of the JDT, using this tool to prospectively measure patient dependency on arrival and departure from A&E (whether admitted, discharged or transferred), and at four and eight hours after arrival where a patient remained in the department for this length of time. This constituted the primary outcome measure.
The form also recorded the patient’s demographic information, triage category and location in the department (resuscitation, majors, minors or the observation unit) at each of these times, and was designed to be as easy to complete as possible.
Although the ideal would have been to collect accurate data on every patient presenting to A&E over this 14-day period, it was accepted at the outset that this would not be possible in practice, particularly during busy times in the department. It was therefore hoped that data would be collected on a representative sample of patients during this time.
The data collection form was presented as a single sheet of A4 paper, and a supply of forms was placed in all areas of A&E to prompt staff to gather information. Forms were also stapled to the standard nursing documentation sheets so that they were readily available for all patients at the time of assessment. Completed forms were placed in clearly identified collection boxes located in each area of the department. During the first few days of the study, the principal investigator was also available in the department to assist, advise and remind staff about the project.
At regular intervals during the study, the principal investigator collected the completed data collection forms and used section B of the JDT to translate the dependency score recorded into one of four levels of patient dependency.
These are graded as follows:
0: Low dependency: patients require minimal levels of nursing intervention and are normally self-caring;
1: Moderate dependency: patients require moderate levels of nursing intervention and are encouraged to become independent;
2: High dependency: patients who require a high level of nursing intervention, but less than that of a totally dependent patient;
3: Total dependency: patients require total nursing care and one-to-one input.
When all available data had been collected, it was entered into a Microsoft Access database, and subsequently converted into a Microsoft Excel spreadsheet by an audit facilitator who also performed preliminary analysis. The data was further analysed using a STATA system, to produce descriptive and chi-squared statistics.
During the 14-day study period, there were 2,014 new patient attendances at the Bristol Royal Infirmary adult A&E department. Of these, data collection forms were completed for 962 (48%).
To determine whether our patient sample was representative of all those attending A&E during this period, the triage category of all patients was compared to the triage category of those for whom study data was available. The results of this comparison are shown in Fig 1. The difference between the two groups is of borderline statistical significance (p=0.06 on chi-squared testing).
Of the patients for whom data was available, 55% were male, and the mean age was 43.2 years. The median length of stay was three hours and five minutes and the mean three hours and 41 minutes. The maximum length of stay was 23 hours and 37 minutes.
The dependency group of patients on arrival, at four hours, eight hours and departure is summarised in Table 1.
Table 1. Patient dependency group on arrival, at four hours, eight hours and departure
|On arrival||At four hours||At eight hours||On departure|
|Dependency group: 0||511||116||36||511|
|Dependency group: 1||356||138||40||246|
|Dependency group: 2||59||21||5||27|
|Dependency group: 3||34||8||1||21|
There is an overall tendency for the dependency group to decrease between arrival and departure. The actual change in dependency for each patient is summarised in Table 2.
Table 2. The overall change in dependency group for each patient between arrival and departure.
An increase in dependency score indicates an increase in patient dependency and vice versa.
|Change in dependency group between arrival and departure||Number||Percentage|
|0 (no change)||618||76.9|
The initial location of the patient according to their dependency group is shown in Table 3. As would be expected, the higher the dependency group, the more likely the patient was to be treated in the resuscitation room.
Table 3. The initial location of each patient according to their dependency group on arrival
|Location||Dependency group: 0||Dependency group: 1||Dependency group: 2||Dependency group: 3||Total|
The mean and median duration of stay according to patient dependency group on arrival is shown in Table 4. In five patients, this data was missing. The mean values are consistently higher than the medians because the data is positively skewed.
Table 4. Mean and median duration of stay according to patient dependency group on arrival
|Mean length of stay||Median length of stay|
|Dependency group: 0 (n=510)||Three hours 21 minutes||Two hours 56 minutes|
|Dependency group: 1 (n=354)||Four hours 5 minutes||Three hours 23 minutes|
|Dependency group: 2 (n=58)||Four hours 12 minutes||Three hours 25 minutes|
|Dependency group: 3 (n=35)||Three hours 32 minutes||Two hours 36 minutes|
The distribution of patient dependency on arrival according to the day of the week is shown in Fig 2. Chi-squared testing of weekdays versus weekends showed no statistically significant difference (p=0.11).
The distribution of patient dependency on arrival according to the time of day is shown in Fig 3. In this analysis, the 24-hour period has been divided into blocks of four hours.
Chi-squared testing showed a highly statistically significant difference in the proportion of high- and total-dependency patients arriving between 8pm and 8am, compared to the other 12 hours of the day (p=0.0008).
The Audit Commission’s (2001) report on ward staffing noted that a significant disparity remains in the amount spent on ward staffing, even after considering ward size and specialty mix. In addition, while there is room to improve the efficiency and effectiveness of staff deployment, actual spending continues to be largely based on historical resource allocations.
We have shown that it is possible to use a tool to practically assess patient dependency throughout a working A&E department, with minimum difficulty and inconvenience.
This process can be readily repeated, and will provide meaningful information to inform the grade mix of nursing staff, the development of nationally accepted nursing competencies (Jones, 2002), and identification of education and training needs (Wood et al, 2004). It can also be applied to other clinical environments such as general wards and critical care.
While we accepted from the outset that our methodology would make it impossible to collect data on all patients presenting to A&E over a two-week period, we wanted to ensure that the data collected was as representative as possible. In particular, there was a risk that staff would tend to collect data on the most dependent patients rather than the whole patient population, to emphasise their workload. Fig 1 suggests that patients with more urgent conditions (and therefore higher dependency) may be slightly over-represented in our sample, and this is of borderline statistical significance. Nevertheless, our sample appears to be acceptably representative of the underlying A&E population, even those managed in the ‘minor’ area of the department (triage categories 4 and 5).
We anticipated that patient dependency would fall during A&E stay as a result of ongoing care, and were pleased to note that overall this was indeed the case. In addition, analysis of individual trends in dependency has the potential to highlight a small group of patients that merits further study: for example, in Table 2 there are two patients whose dependency group increased by two or more places during their A&E stay, suggesting substantial deterioration. Similarly, three patients improved dramatically from total dependency to low dependency during their stay. These are cases that should be reviewed in detail to draw out any learning points that may improve future care.
Table 3 shows how knowledge of patient dependency allows capacity planning, since it is possible to predict the occupancy of various areas of A&E. For example, immediate resuscitation room facilities are required for approximately 80% of total-dependency patients (group 3) and 20% of high-dependency patients (group 2), but only rarely for the other two dependency groups. This information has directly informed the design and construction of a new resuscitation area, as well as its associated staffing levels.
We were interested to note that the shortest median length of stay occurs in patients with the highest and lowest dependency. The most likely explanation for this is that low-dependency patients are processed and discharged relatively quickly, whereas those with total dependency are often very unwell and will be rapidly transferred to the ICU or operating theatre, or perhaps die while they are in A&E. This is supported by other UK research, which has looked specifically at the epidemiology of adult transfers from A&E to critical care facilities (Gray et al, 2003).
In our unit, patients with moderate and high dependency are likely to stay the longest, usually because they are waiting to be admitted to an inpatient bed.
Since this study was completed, the average time that patients spend in the department has reduced considerably, mainly as a result of nationally imposed targets, but these overall trends appear to hold true.
There is some suggestion that the proportion of high- and total-dependency patients falls at weekends, but the numbers are small and the two-week time period too short to draw any definite conclusions (Fig 2). On the other hand, it is clear that the proportion of patients with high or total dependency increases overnight (between 8pm-8am, Fig 3). This is a good example of a situation in which dependency, as well as absolute patient numbers, must be taken into account. It is common to reduce staffing levels overnight since the absolute number of patient attendances falls, but the average dependency of those patients who do attend A&E will be significantly higher than is the case during the day. Therefore, it may not be wise to substantially reduce nursing levels overnight.
Overall, this study has allowed us to map, with much greater accuracy, the needs of patients attending our adult A&E department. For example, we were already aware that on an average weekday approximately 140 new patients present to the department. We are now able to add that, of these 140, there will be approximately nine in the high-dependency group and five in the total- dependency group: two-thirds of these will present overnight (8pm-8am) and around eight patients will go directly into the resuscitation room. Twenty-two patients will have a decrease in dependency during their stay, and four will have an increase.
By adding dependency data to absolute patient numbers, it is possible to tailor the number and grade of staff much more closely to patients’ needs, improving clinical care and reducing the pressure under which staff work.
This study has some limitations. The most obvious is the limited time over which it was conducted, and the failure to collect data on all patients. Nevertheless, we feel the sample is appropriately representative, and was gathered without undue consumption of staff time or other resources. This makes it easy to repeat in the future.
The interpretation of the JDT may vary somewhat between different staff members, but the tool has previously been validated in an A&E setting. Inevitably there are some areas of missing data, particularly towards the end of patients’ stay, and this makes it harder to interpret the overall trends in dependency.
Future research will focus on the ways in which dependency data can be translated into a specific grade mix, and on models of patient throughput and dependency. Ultimately, it may be possible to investigate the relationship between grade mix, patient care and clinical outcome, where data is substantially lacking at present.
Patient dependency is one of the essential determinants of nursing grade mix, and we have shown that this can be simply and usefully measured for all types of patients in a busy adult A&E department using the JDT. The data collected can be used to predict workload, resource use and the optimal staffing levels that will provide safe and effective patient care. Dependency can be readily and repeatedly assessed, and we recommend this approach for other clinical settings.
Implications for practice
Nurse staffing levels are often based on arbitrary or historical measures. Where attempts have been made to match staff to patients, the main determinant tends to be patient numbers, with little consideration of how dependent patients are;
Patient dependency can be successfully measured in a busy clinical environment, and used to more accurately inform nursing grade mix;
Dependency assessments can also identify patients whose condition takes an unexpected clinical course and provide a better understanding of workload, thereby informing capacity planning;
We recommend that patient dependency be more frequently assessed in the clinical environment, to ensure that wards really do have the number of nurses they require. This will improve patient outcomes, working conditions and staff morale.
This work was first published in the Journal of Clinical Nursing (O’Brien and Benger, 2007). This article is based on the research reported in the original JCN study.
We would like to thank: the United Bristol Healthcare NHS Trust for supporting this work; Sorrel Hewes, clinical audit facilitator, for advice and assistance with data entry and analysis; and Helen March, clinical librarian, and Rebecca Hoskins, nurse consultant, for assistance with the literature search and references.
Audit Commission (2001) Ward Staffing - A Review of National Findings. Acute Hospital Portfolio. www.audit-commission.gov.uk
Ball, C. et al (2004) Moving on from ‘patient dependency’ and ‘nursing workload’ to managing risk in critical care. Intensive and Critical Care Nursing; 20: 2, 62-68.
Browne, A.C., Odell, M. (2004) A review of nursing skill-mix to optimise care in an acute trust. Nursing Times; 100: 6, 34-38.
Buchan, J. (2001) Determining Skill Mix: Lessons From an International Review. www.moph.go.th
Furlong, S., Ward, M. (1997) Assessing patient dependency and staff skill mix. Nursing Standard; 11: 25, 33-38.
Gibbs, I. et al (1991) Skill mix in nursing: a selective review of the literature. Journal of Advanced Nursing; 16: 242-249.
Gray, A. et al (2003) Descriptive epidemiology of adult critical care transfers from the emergency department. Emergency Medicine Journal; 20: 242-246.
Hurst, K. (2003) Selecting and Applying Methods for Estimating the Size and Mix of Nursing Teams. A Systematic Review of the Literature Commissioned by the Department of Health. www.who.int/hrh/tools/size_mix.pdf
Hurst, K. et al (2000) Managing and leading psychiatric nursing, part 1. Nursing Management; 6: 10, 8-13.
Jones, M. (2002) Critical care competencies. Nursing in Critical Care; 7: 111-120.
Jones, G. (1990) Accident and Emergency Nursing - A Structured Approach. London: Faber & Faber.
O’Brien, A., Benger, J. (2007) Patient dependency in emergency care: do we have the nurses we need? Journal of Clinical Nursing; 16: 2081-2087.
Standing Nursing and Midwifery Advisory Committee (2002) Balancing the Shift - A Position Paper Exploring Key Issues for Nursing Skill Mix within the Context of Workforce Planning. www.advisorybodies.doh.gov.uk
Wanless, D. (2002) Securing our Future Health: Taking a Long-term View. www.hm-treasury.gov.uk
Wood, I. et al (2004) Education and training needs for acute critical care delivery: a needs analysis. Nursing in Critical Care; 9: 159-166.