SELECTING AND APPLYING METHODS OF ESTIMATING THE SIZE AD MIX OF NURSING TEAMS

PROFESSIONAL JUDGEMENT METHOD Telfords early work,using expert,professional judgement to agree the most appropriate size and mix of ward nursing teams has stood the test of time.This technique simply helps you to convert your duty rotas into whole time equivalents (WTEs).This method,as the algorithm below shows,is simple to use and is an excellent starting point before you tackle the more sophisticated methods that come later..In the following example from a 15 bed surgical ward,a decision is made to roster three nurses for the morning and afternoon shifts,and two nurses for the night shift.A 30 minute morning to afternoon shift hand-over period,and a 15 minute afternoon to night shift hand-over is included because it is part of the usual work pattern. substitute local times and your preferred number of staff for different contexts.

Step 1.Calculate the number of working hours needed: Early shift:0700 to 1430 = 7.5 hrs x 3 nurses x 7 days = 157.5 hrs Late shift:1400 to 2130 = 7.5 hrs x 3 nurses x 7 days = 157.5 hrs Night shift 2115 to 0715= 10 hrs x 2 nurses x 7 days = 140 hrs However, these hours assume that nurses are never sick or take holidays,etc.A time-out adjustment to cover leave of all kinds, therefore, is necessary.The 22% allowance used in the formula below was obtained from a time-out study of 300+ general wards in the UK.However, if you wish then you can substitute a local figure(obtainable from your personnel department). Step 2.Adding the time-out allowance. 455 hrs x 1.22 (time-out) = 555.1hrs/37.5hrs (1WTE) = 14.8 WTEs. A staffing pattern of three nurses for the morning, three nurses for the afternoon/evening and two nurses at night,therefore,requires almost 15 full-time nurses for this small surgical ward. The same approach can be used for five-day wards but different hand-over allowances are used in Table 2 below.

Early shift:0700 to 1445=7.75 hrs x 3 nurses x 5 days = 116.25 hrs Late shift 1400 to 2145=7.75 hrs x 3 nurses x 5 days = 116.25 hrs Night shift 2115 to 0730=10.25 hrs x 2 nurses x 4 days = 82 hrs

The five-day ward time-out value would not be as great as a seven-day ward;therefore,315 hrs x 1.18(time out) = 371.7 hrs/37.5 hrs = 9.9 WTE. substitute local time-out values.

One spin-off from the professional judgement staffing formula used in the seven-day and fiveday ward examples above is that the technique can be reversedto calculate the available nurses per shift from a wards actual (names on the duty rota) or funded (what the budget allows) nursing establishment. The process goes as follows: 1. A seven day ward requires 21 shifts (7 days x 3shifts per day) to be staffed by nurses. 2 .Each full-time nurse works 5 shifts. 3. Therefore, 4.2 WTE nurses provides 1 nurse per shift (21/5 = 4.2).4.Two nurses per shift require 8.4 WTEs,and so on. 5. However,were faced with the same time-out problem discussed above. That is,the 4.2 WTE nurse figure lacks an allowance for paid and unpaid leave. 6. Therefore,4.2 x 1.22 (22% time out) = 5.1 WTE nurses provides one nurse per shift. Applying this technique to the specimen ward in the following table shows its value.

Grade

Funded WTEs

Divisor

Nurses per Shift

5.1

0.2

F E

1.5 2.5

5.1 5.1

0.3 0.5

5.5

5.1

1.1

5.1

5.1

5.1

Total

25.5

5.1

The funded nursing establishment in Table 3 above allows: One G or F grade nurse on duty every other shift. One E grade on duty every other shift. One D grade on duty every shift. One C grade health care assistant on duty each shift. One B grade nursing assistant on duty each shift. One A grade on duty each shift. In total five nurses per shift .In practice, the 25.5 nurses would be equitably distributed between day, night and weekend shifts.The next logical step would be to build a duty rota from these findings. Strength and weakness As the literature often explains, no nursing workforce planning method is perfect and the professional judgement method has its weaknesses.For example,the relationship between staffing levels and nursing quality is hard to explain using this method.That is,how do we know if 25.5 WTE nurses is enough to maintain an acceptable standard of care,or to ensure equitable workloads, job satisfaction andtherefore,a desire to stay in the job? A follow-up study of nursing care quality and nursesjob satisfaction is essential to check the adequacy of the wards establishment, which arguably is no bad thing under any circumstances.

The professional judgement method is less flexible when patient numbers and especially patient dependency mix change;that is,the ward will often be over- or understaffed.As a

result,the method is deemed too subjective; that is, should professionals themselves be determining their own staffing levels without an independent check? Finally, calculations get awkward when unusual shifts are worked such as long days. However, computer spreadsheets ease the burden.

AVERAGE NURSES PER OCCUPIED BED (NPOB) Average nurses per occupied bed (NPOB) is another popular and simple method of determining or evaluating the number and mix of ward staff. The formulas shown below were compiled from a study of 300 plus hospital wards in the UK .The 1.35 WTE nurses per occupied medical ward bed figure in Table below, for example, was obtained from 83 medical ward nursing establishments. The actual establishment rather than the funded is used in that actual establishments includes overtime, agency and bank hours. Ward overhead and time-out allowances are built into the formulas to add the indirect care, associated work and leave/absence components. Calculating staffing from average NPOB

Care Group .

Medic.

Elderly

Surg.

Ophth

.ENT

Gynae.

Ortho.

Paed

Number of wards Average occupancy G/H/I per occ.bed

83

54

66

11

53

26

24

24

22

13

16

20

22

15

0.06

0.05

0.05

0.11

0.09

0.06

0.05

0.13

F grade

0.12

0.06

0.07

0.09

0.15

0.09

0.05

0.2

E grade

0.32

0.24

0.27

0.22

0.45

0.14

0.24

0.49

D grade

0.48

0.31

0.38

1.03

0.55

0.45

0.38

0.53

C grade

0.07

0.04

0.06

0.07

0.03

0.02

0.05

0.11

Nursing assistant

0.3

0.51

0.31

0.25

0.23

0.12

0.44

0.27

Total

1.35

1.21

1.14

1.77

1.50

0.88

1.21

1.73

In the elderly ward example in Table above, one patient requires 1.21 WTE nurses to meet his or her needs. An average of 24 patients requires a nursing establishment of 29 WTE nurses (24 x 1.21),or put another way,29 full-time nurses on the duty rota. leave of all kinds comes out of this establishment, and as we saw above, at any time one nurse in five is away from the ward. These formulas are less generous in this light. Calculating the grade mix follows the same process - multiplying the average number of occupied beds with the grade mix proportion.

STRENGTHS AND WEAKNESSES If nothing else, these data provide opportunities to benchmark your wards.The method can also be used to verify professional judgement method findings. Clearly ,the NPOB method comes into its own if your ward bed complement changes and you need to modify the nursing establishment. Another strength is that the keep it-simple method of demand-side workforce planning is honoured. The staffing and grade mix formulas in Table 4 above have been empirically derived and the

formulas use data collected systematically;for example, bed occupancy and payroll information. Also, formulas for the main specialities are unique because they are derived from data collected only in same-speciality wards. Moreover,the wards providing these data have passed a quality test;that is,none fell below a pre determined quality standard to avoid projecting from inadequately staffed wards Learners are supernumerary in the staffing projections and if the base wardsstaffing is fair then mentoring and supervision ought to be easily accommodated. This approach,as we saw above, makes determining establishments and generating the wards grade mix easy since formulas are broken down by nursing grade.Even though you may not have the financial or staffing resources to boost your establishment to levels recommended by NPOB formulas,at least you can benchmark your own establishments.Finally,the data are easily built into a computerised spreadsheet for what-if?purposes,which can be downloaded FoC from the website given at the end of this report. On the downside,this method assumes that base staffing was rationally determined.However, theres evidence in the literature that ward establishments are historical and sometimes bare little relationship to ward occupancy .As it happens,the NPOB averages above were derived from quality assuredwards.Unlike the data in Table above,theres no guarantee that the averages from other sources (such as ones foundin the literature) come from wards that deliver an acceptable standard of care. These NPOB staffing formulas are insensitive to patient dependency changes;that is,the formulas recommend the same number of nurses for patient populations that are predominantly low dependency as it does for high dependency in patients. Dependency can have a striking effect on nursing workload . Formulas are costly to update; extensive fieldwork is required to alter formulas for a speciality that changes nursing practice in some way. Routinely collected data, such as bed occupancies used in staffing formulas, are more error-prone than those that are deliberately and systematically collected because empirical data are usually confirmed in some way. The NPOB formulas in Table 4,like those in the literature, contain hidden structures and processes that need to be made explicit. For example, the ophthalmic ward data in Table above are drawn from wards where nurses also staff operating theatres, hence what seems at first glance to be a generous number of nurses per occupied bed.

Similarly, NPOB data may be drawn from wards that are geographically different to your wards; that is, hub and spoke units have subtle nursing activity differences compared to Nightingaletype wards. However, layout is a much less important workload variable than dependency. Finally, learner nurses contributions, or alternatively their demand on qualified staffs time, warrant special consideration to which some NPOB formulas are insensitive.

ACUITY-QUALITY METHOD A third way of estimating or evaluating the size and mix of ward nursing teams is (in full) the dependency-activity-quality or acuity-quality method for short .This staffing method overcomes most of the weaknesses highlighted in the professional judgement and the NPOB methods .It is useful for wards where patient numbers and mix fluctuate. Consequently, medical and surgical admission unit managers find the acuity-quality method invaluable.

The data averages in this example are from 83 quality assured UK medical wards (see Table 5 below). Step 1. Obtain the average number of patients in dependency categories 1 to 4. From table 5 dependency numbers are obtained by multiplying the bed occupancy by the proportion of dependency 1 patients, etc.:

Dependency Category Total Number of patients

1 5

2 10

3 7

4 3

TOTAL 25

Dependency category 1 patients are virtually independent of nurses. Dependency 4 patients, on the other hand, are dependent on nurses for most if not all their needs. We use a four-group dependency model in this algorithm but other configurations work equally well.

Step 2.Record the average amount of direct care time given to each dependency category per day (using data only from quality assured wards,

Dependency Category Daily time in minutes

1 46

2 106

3 197

4 336

The daily times (in minutes) were obtained from observing 3000+ hours of nursing care in 83 medical wards (see Table 5).In short ,the lowest dependency (1) patient gets three-quarters of an hour of hands-on care each day. The most dependent (4) patient, on the other hand, receives 5.6 hours of nursing care a day. Because this care is direct or hands-on care, the ward overhead needs adding later. The annotated bibliography on p.69 provides several sources of activity analysis instruments from which these direct care times can be generated. Alternatively, e-mailing the author <[email protected]> will generate one.

Step 3.Convert the times in Step 2 above into ratios by dividing dependency category 1 minutes into dependency 2 minutes, dependency 1 into dependency 3 and dependency 1 into dependency 4 : Dependency Category Ratios 1 1 2 2.3 3 4.3 4 7.3

This seems an odd step but calculations using care ratios make the remaining task easier and more meaningful. In short, we can see that dependency 4 patients get seven times more nursing care than dependency 1 patients (which reflects the 4s higher dependency). Step 4.Multiply the ratios by the average daily number of patients in each dependency category to obtain the workload index (WLI) or acuity (hence the methods name) :

Dependency Category Ratios Number of patients

1 1 5

2 2.3 10 23

3 4.3 7 30

4 7.3 3 22

Total

25 80

Workload index/Acuity 5

Step 5. In other words , the WLI/Acuity is equal to the nursing work needed to care for 80 dependency 1 patients. Dividing WLI by the occupancy (80/25 = 3.2) gives the bed acuity. The WLI (80) and the bed acuity (3.2) are good benchmarks. Once you get accustomed to using them Step 6.The nursing time required for good quality care for a dependency 1 patient, as we saw above, is 46 minutes per day (see Table 5).Its a good time to reiterate that the 83 medical wards used in this example have passed the quality test so that we dont extrapolate from poor practice wards. The direct care time for all patients is: 80 (WLI) x 46 minutes = 3680 minutes/60 minutes = 61.3 hours per day.

Step 7.We also know from our studies of nursing care in 83 medical wards that nurses spend 42% of their time in direct nursing care. Adding the indirect care component (or ward overhead), therefore, involves: 61.3/42 x 100 = 146 hours per day x 7 days = 1022 hours per week.

Step 8.Nurses in the 83 medical wards took meal and drink breaks averaging 10% of their working day (see Table 5).As you might expect ,this time was included in the activity analysis but it is not part of the 37.5 hour week. Ten per cent, therefore, is deducted: 1022 hours - (1022 x 0.1) = 920 hours

Step 9.The nursing hours calculated so far assume that ward staff do not take annual or sick leave, etc. Again, from our study of 83 medical wards, we note that nurses account for a 22% time-out value (see Table 5).An allowance for paid and unpaid leave, therefore, is added: 920 x 1.22 = 1122 hours Step 10.Convert the total nursing hours for the week into whole time equivalents: 1122/37.5 hours = 30 WTEs. This specimen medical ward, therefore ,requires 30 full-time nurses to care for patients 24 hours a day seven days a week.

Step 11.In recent years weve embellished the acuity-quality method by adding a grade mix component. The grade proportions below were obtained from a who should do what study of 83 medical wards. These data are up-to-date at the time of publication. They are, however, revised at least yearly.

In short, we multiply the required WTEs (from Step 10) by the appropriate grade mix proportion from the medical column in Table 5 . Grade Medical Ward Proportion G/H/I 4% (30 x 0.04) 1.2 WTE

11%

(30 x 0.11)

3.3

21%

(30 x 0.21)

6.3

33%

(30 x 0.33)

10

12%

(30 x 0.12)

3.6

Nurs.Ass

19%

(30 x 0.19)

5.7

TOTAL

30

In practice, theres only likely to be one ward leader and the other WTE decimal places will be adjusted to create more realistic contracts such as 3.5 full time HCA grade C.

Table 5. Acuity Method Base Data

Variable Medic. Elderly Surg.

Care Group Ophth . ENT Gynae. Ortho Paed

N wards

83

54

66

11

53

26

Occupancy 25

24

22

13

16

20

22

15

Dep 1

19%

12%

19%

23%

28%

29%

21%

8%

Dep 2

42%

23%

40%

55%

39%

44%

36%

31%

Dep 3

28%

47%

28%

21%

26%

17%

34%

47%

Dep 4

11%

18%

13%

1%

7%

10%

9%

14%

Daily mins

Dep 1

46

26

67

62

79

79

53

67

Dep 2

106

79

98

170

156

120

106

110

Dep 3

197

154

240

178

164

185

185

218

Dep 4

336

214

295

276

278

271

278

341

Table 5.Acuity Method Base Data (continued)

Variable Medic. Elderly

Care Group Surg. Ophth. ENT Gynae. Ortho. Paed.

Direct care Meal break Time out

42%

45%

43%

29%

34%

39%

42%

41%

10%

8%

9%

8%

11%

8%

9%

9%

22%

23%

21%

22%

25%

23%

22%

21%

Grade mix

G/H/I F

4% 11%

4% 8%

4% 10%

4% 11%

4% 12%

4% 10%

4% 9%

4% 10%

21%

15%

21%

23%

24%

22%

19%

21%

33%

30%

35%

32%

34%

35%

33%

39%

12%

15%

12%

11%

10%

12%

14%

10%

Nurs. Ass.

19%

28%

18%

19%

16%

17%

21%

16%

STRENGTHS AND WEAKNESSES One of the strengths of the acuity-quality method is that you can substitute data in Table 5 above with local values. For example ,the average medical ward patient dependencies can be replaced with local numbers. If you dont have patient dependency data then you can convert your occupancies into typical dependency mixes simply by multiplying your wards bed

occupancy by the patient dependency proportions given in Table 5.This alternative, however, diminishes the acuity-quality methods power. Recently, high patient throughput wards, such as medical admission units ,convert short-stay patients into patient whole-time equivalents (PWTE).Simply counting heads at some census point is deemed less valuable in these wards than summing the daily patient hours for each dependency group and dividing by 24 to arrive at PWTEs. This new approach gives a more accurate WLI in busy wards. Default direct care and time-out components in Table 5 can also be overwritten with local values, which unlike substituting dependency numbers, do not reduce the methods sensitivity. However, the direct care and time-out percentages in the main report are robust data and have been corroborated not only empirically but also by the literature Therefore, we should not overwrite the averages on a whim. Changing ward variables, however, especially patient numbers and dependency mixes, is easily accommodated by the acuity-quality algorithm, and especially by the software on the website . The data in Table 5 are drawn only from wards that achieve a pre-set level of quality. In theory at least, the data, algorithm and software should recommend establishments that achieve the same quality of care in different settings. You can turn the acuity-quality method around and adjust your wards occupancy and patient dependency mix to suit the available nursing resources. Either way, nurses are matched to the peaks and troughs of ward activity. This method is one way of deploying nurses where the need is greatest thereby making workloads equitable. Once a computer is set up ,it is possible to calculate staffing numbers for individual shifts. Software also allows manipulation of a single or a combination of variables in a what if? way. Finally, nursing benchmarks and performance indicators (such as nursing cost per occupied bed) are a natural spin-off from the acuity-quality method. These data are often staggering;for example, the daily nursing cost per occupied bed in some wards can be double that in similar wards without any obvious care-quality gain. One of the main weaknesses is that compared to the professional judgement and the NPOB staffing formulas, the acuity-quality method is complex. However, it is felt that the extra effort pays dividends since more variables, known to influence nursing workload, are accommodated.

Another downside is that the daily direct care minutes for each dependency category have to be accepted unless local nursing activity values can be obtained. However, adopting patient and nursing activity from other hospitals is unpalatable. Moreover, the sense of ownership that is engendered by using local information may be lost when external data are used. Also, in order to capitalize on the acuity-quality methods power and flexibility, computer software, such as spreadsheets, are needed, which can be downloaded FoC from the website at the end of this report. Collapsing patient numbers and related nursing activity data into dependency groups ignores individual patient characteristics .One patients special needs may have only a minor effect on acuity-quality staffing formulas even though one-to-one nursing care may be needed. A similar criticism levied at acuity methods is that nursing activity, used to obtain the amount of nursing time required, sometimes fails to measure the psychological component of patient care. However, most of the alternative methods are even less sensitive to these issues. Acuity-quality methods in some situations can recommend nursing establishments insufficient to provide at least one qualified nurse per shift because the formula is workload as well as occupancy-based. Patient populations less than 12,especially if the patients are low dependency for example ,create the so-called small ward problem. Acuity-quality systems add to ward nurses workload because additional patient information required. That is ,the named nurse is the best person to assess his or her patients dependency. Similarly, obtaining up-to-date data can be expensive; for example,representative,nursing activity and nursing quality data require two independent non-participant nurse observers spending several days in the ward. Despite matching nursing activity with nursing quality, the relationship between the two can be confounded; that is, some understaffed wards achieve high-quality care and vice-versa. The relationship between staffing and outcome is complex and uncertain. Even though the formulas in the main report are designed to overcome inappropriate working, the grade mix configurations may not suit your wards context. For example, it may not be local policy to employ Level 3 or 4 health care assistants. Reconfiguring the grade mix according to local policy ,and adjusting the acuity-quality algorithm at the same time, takes considerable fieldwork and skill.

Finally, the acuity-quality method lends itself less well to forecasting the number of staff than other methods . TIMED-TASK/ACTIVITY METHOD This method of estimating or evaluating the size and mix of nursing teams arose mainly from a belief that acuity-quality staffing methods, for example,were inferior staffing predictors. The type and frequency of nursing interventions required by patients are felt to be a better predictor than patient dependency .If nurses are comfortable with constructing patient care plans then the timed-task/activity method simply requires nursing minutes to be added to each intervention in the plan thereby generating the number of nursing hours needed. This method will suit wards in which care plans are systematically constructed, and for wards where patients nursing needs can be confidently predicted; notably those that admit from waiting lists. In practice ,each patients daily direct nursing care needs are recorded either manually or electronically on a locally developed checklist of nursing interventions. The number of nursing interventions from which to chose varies from system to system. Because each intervention is paired with a locally agreed completion time, the patients care plan and nursing time requirement is systematically built. The value attached to each intervention is generally the amount of time needed to carry out the care for one patient over a 24 hour period. As with the acuity-quality method, a ward overhead is added to cater for the indirect care and other aspects of nurses time. Similarly, breaks and time-out have to be considered and ideally, the method should be computerized .The method generates results that can be easily corroborated by other methods It is easily computerized so that the method becomes part of a nursing information system. As you might expect, commercial IM&T- based systems, such as GRASP, are readily available which makes it easier to update base information. Indeed, periodic reviews of nursing interventions and times are a good idea. Adopting the system in other care settings is possible without destroying its integrity. However, protagonists warn users to check validity and reliability if grand care plans are transplanted into new nursing settings. Despite one of the first rules of workforce planning: that it shouldnt add to the nurses workload, the effort needed to maintain detailed care plans for each patient every shift adds considerably to the ward overhead.

Consequently, timed task/activity commercial systems are the most expensive of all the methods described. Setting up and implementing the system is also time consuming. But like the acuity-quality method, these are largely capital rather than recurrent costs. Finally, reducing nursing care to a work-study type list horrifies some nurses. However, to reiterate, the completed detailed list of required nursing interventions for an individual is no different to a comprehensive nursing care plan. REGRESSION ANALYSIS METHOD Powerful regression-based predictions to estimate or monitor staffing levels have been developed .Broadly, regression methods predict the required number of nurses for a given level of activity. The predictor is called the independent variable and the outcome or level of staff is known as the dependent variable. Although the statistical analysis is challenging, once completed, all we need to know is the independent variable value to predict the number of staff (dependent variable).For example,one study developed a nurse-staffing model from an analysis of ward establishments and bed occupancies. Regression analysis showed that the number of nurses (dependent variable) increased as bed occupancy (independent variable) rose thus allowing staffing estimations.Other independent variables in the literature include the number of theatre sessions and day surgery cases. In short, once the base data are collected and analysed then the calculations are as straight forward as the NPOB method.

STRENGTHS AND WEAKNESSES One of the regression methods strengths is that it is useful for situations where predictions are possible, such as the number of planned admissions.This approach helps managers to forecast and prepare for extra demands.Once established,it tends to be a cheaper method because data are easier to collect and can be aggregated from similar wards. Moreover, independent variable data are usually inexpensive to update. The regression method, therefore, is especially useful to managers with limited resources,and who cannot afford to carry out full dependency-activity-quality or a timed- task/activity study. The outcomes of regression models tend to be corroborated with independent evidence.

Consequently, staffing formulas are judged valid, reliable and also more usable than the detailed and expensive acuity-quality and timed- task/activity methods. Staffing recommendations from regression analyses are relatively easily tested for accuracy by checking how well nursing time is used following enactment of staffing recommendations drawn from regression models. Most care groups can be analysed in this way, therefore the ease of use across specialities is another strength. On the down side, since the number of variables for consideration in a ward setting are likely to be great, the knowledge and skills of a statistician will be needed to design and implement fieldwork that collects the most appropriate data for regression analysis. Transferring staffing formulas derived from regression coefficients from one setting to another isnt encouraged owing to unique variables (such as ward layout). However, validity and reliability tests help to check if transplanting is safe.Some independent variables are qualitative while others are deemed subjective such as the ward managers perceptions of ideal staffing. Sometimes, nominal data have to be assigned to variables and readers with statistical knowledge and skills know that regression analysis models are usually based on interval or ratio data. Wards providing data for regression analysis are assumed to operate efficiently and effectively; that is,wards supplying establishment and bed occupancy data have had staffing varied according to patient demand.Similarly,including data from wards with excess absenteeism or poor quality care can distort and invalidate results.Its unsafe to predict staffing levels outside the regression models observed range.That is,if your data came from wards with no more than 25 beds then extrapolating to wards with 30 occupied beds, for example, can lead to errors because we cant be sure that linear relationships between independent and dependent variables exist beyond 25 beds. Finally, imposing regression statistical techniques has alienated some nurses owing to a lack of ownership and understanding.

SUMMARY From a wide and deep literature review, five main demand-side workforce planning methods and related data have been extracted. These range from the quick and easy to complex and powerful approaches. Triangulating two or more methods will give confidence to check if staffing recommendations are appropriate.