Staff unavailability and safe staffing: are headroom allowances ‘realistic’?

The components of headroom

Given the importance of headroom in ensuring safe staffing levels, it is imperative that the allowance is ‘realistic’ (Kirby, 2015; NHS England and NHS Improvement, 2019b:12). However, headroom is increasingly being included in hospital roster policies as a performance measure to be reviewed monthly (NHS England and NHS Improvement, 2019b). The headroom allowance comprises:

In England, NHS Improvement (2018) proposed a figure of 22.2% for headroom. This is comparable with the earlier recommendation of 22% (Hurst, 2003) and the broader range of 22-25% given by McIntyre (2016). However, in practice, the Carter report, based on an investigation of 32 NHS trusts, found a range of headroom allowances from 18.5% to 27% (Carter, 2016: 21). Disparities of this magnitude are disturbing. Although over-statement of headroom may prove costly, this can often be resolved by judicious staff redeployment. However, understated headroom results in under-stated unit budgets (McIntyre, 2016). This has implications for patient care, staff workload, staff wellbeing, staff retention and, indirectly, cost, due to the use of additional bank and agency staff.

Sample

Two sources of data were used. First, data extracted from the e-rostering systems of 87 NHS trusts of various sizes for the period 1 January 2016 to 31 December 2016. This convenience sample represents approximately 37% of NHS trusts. The sample included a variety of trust ‘types’ (foundation (53), acute (46), community (44), mental health (30)) and sizes (<1000 beds (34), 1000–2999 (45), 3000–5000 (7), >5000 (1)).

The second data source was a series of 35 publicly available roster policies, collected between 2015 and 2017 as part of a separate study. This convenience sample represents approximately 19% of NHS trusts.

Data collection

Analysing unavailability data from the e-rostering system

For the 87 trusts examined, the maximum, minimum, mean and median recorded variability were documented (Figure 1). However, 11 of these trusts did not include data for headroom. These 11 were then removed from the sample and for each remaining trust, the values of planned headroom were compared with those for recorded unavailability occurring over a 12-month period (Figure 2). Finally, for each trust, actual unavailability was subtracted from headroom to provide an estimate of the potential for over- or under-staffing (Figure 3).

Analysing headroom data from roster policies

Of 35 policies citing headroom, 13 did not specify any value for headroom or any of its components. Of the remaining 22 roster policies, those of two trusts, Isle of Wight and Royal United Hospital Bath, included headroom figures of 22% but did not specify the components of this total. Where available, the value of each component of headroom, as specified within the roster policy, was recorded (Table 3).

Headroom and unavailability data from e-rostering system

Across the 87 trusts examined, recorded unavailability varied from 16% to 34% (Figure 1). Fourteen trusts were below the 22% headroom figure recommended by Hurst (2003) and the lower limit suggested by McIntyre (2016), while 54 were above the upper limit of 25% prescribed by McIntyre (2016). Consequently, the average across the sample, 25.7%, was higher than McIntyre's upper limit. Of the 76 trusts specifying headroom data, headroom varied from 16% to 26% (Figure 2), with an average of 21.1% (marginally lower than Hurst's recommendation and McIntyre's lower limit). For trusts where headroom exceeded unavailability (positive variability) there was the potential for over-staffing. For those where unavailability exceeded headroom (negative variability), this was indicative of under-staffing (Figure 3). The mean for both headroom and unavailability were calculated for the remaining 76 units.

Headroom in roster policies

Of 35 policies citing headroom, 13 did not specify any value for headroom or its components. For the remaining 22 policies, headroom varied from 18% to 25%, with an average of 22%. This corresponds to the recommendation of Hurst (2003) and the lower limit of the range recommended by McIntyre (2016). The roster policies of two trusts, Isle of Wight NHS Trust and Royal United Hospitals Bath NHS Foundation Trust, included headroom figures of 22% but did not specify the components of this total and were therefore excluded from further analysis. For the remaining 20 trusts, the value of each component of headroom, as specified within the roster policy, often varied substantially between trusts (Table 3). For example, annual leave, the largest component of headroom, was only 10% at Barnet, Enfield and Haringey Mental Health Trust, but 16% in the Somerset Partnership NHS Foundation Trust. The range of quoted headroom across the sample varied from 18% to 25% (Table 4).

Unavailability and headroom

In the UK, it has been suggested that a 1% improvement in staff unavailability would provide an additional £339 million to fund further frontline work (NHS Workforce Deployment Expert Group, 2018). To offset this unavailability, trusts are advised to include headroom when setting unit establishments (NHS Improvement, 2018) but, importantly, unavailability and headroom are not the same. At board level, many trusts regard headroom as a KPI and set targets for its components in their roster policies (Mersey Care NHS Foundation Trust, 2017). Often, this ‘headroom KPI’ is applied across all units (Mersey Care NHS Foundation Trust, 2017; North Tees and Hartlepool Hospitals NHS Foundation Trust, 2018). However, this research highlights large variations in unavailability from 15.8% to 33.6% (Figure 1) and contrasting levels of headroom (16-26%). Indeed, 13% of the Trusts investigated did not incorporate any headroom in their e-rostering system. These 11 trusts included the one with the highest recorded unavailability (33.6%) and five others that exceed McIntyre's (2016) 25% threshold.

Low values of unavailability (≤ 20%) may be regarded as unusual, given a mandated annual leave allowance between 13% and 16% (NHS Staff Council, 2019) and a typical sickness rate of 4.5% (NHS Digital, 2018). Such levels of unavailability may require further scrutiny to confirm that study leave, parenting and non-clinical work are included. In this research, seven trusts recorded unavailability of 20% or less in their e-rostering systems. In the roster policies examined, 18 trusts declared headroom below 20%. The roster policies of two trusts—Barnet, Enfield and Haringey Mental Health Trust and Mersey Care NHS Foundation Trust—provide a useful insight (Table 3). Barnet, Enfield and Haringey quotes a headroom of 19% based on 10% annual leave and 3% sickness. However, assuming staff plan to use all of their holiday entitlement, annual leave must be between 13% and 16%. To achieve 10% annual leave, even assuming all staff have less than 5 years' service, each employee would need to forego 9 days of holiday. Regarding sickness, the trust policy provides only a 3% allowance, this despite an average trust sickness rate of 4.2% for the 9-year period 2009–2018 (NHS Digital, 2018). Similarly, Mersey Care quotes a headroom of 18% based on 3% sickness and no allowance for study leave or non-clinical working. The average sickness rate for Mersey Care for the period 2009-2018 was 6.2%. Barnet, Enfield and Haringey and Mersey Care were 2 of 14 trusts that did not specify an allowance for parenting (Table 3) in their roster policies.

NHS Improvement (2018) offered a sample breakdown of headroom (Table 5) that has been used as the basis for some roster policies. However, it appears at odds with guidelines produced by other bodies such as the NHS Staff Council and the NMC. It also makes no allowance for non-clinical work. Clearly, budgeting for unavailability remains challenging and recommendations of headroom value have been omitted from the current Nursing and Midwifery E-rostering: A good practice guide (NHS England and NHS Improvement, 2019b). A further barrier to achieving more accurate headroom allowance may be the nature of certain components of unavailability.

Annual leave and sickness

Annual leave varies within the range 13-16% according to the service profile of each unit (NHS Staff Council, 2019) and analysis of roster policies (Table 3) supports this, with a mean of 14.3%. Given that leave must be requested, and often requires a notice period, the challenge for unit managers is ensuring that leave is taken regularly throughout the year—a difficult, but manageable, task in most instances. In contrast, absence due to ‘sickness’ often occurs without notice and is difficult to forecast. Such short-term absence may be due to social and personal factors rather than illness, whereas long-term absence is mostly associated with medical problems (Johnson et al, 2003).

At trust level, given the smoothing effect of multiple units and the annualisation of data, variations of sickness absenteeism appear modest and consistent with those typically quoted in roster policies. However, at unit level, the situation is quite different. For example, trust-wide sickness data from North Staffordshire Combined Healthcare NHS Trust for the period 2009–2018 revealed an average sickness level of 4.6% (NHS Digital, 2018). However, using data from the trust's Six Month Safer Staffing Reviews (Wilson, 2014; Sylvester, 2015; North Staffordshire Combined Healthcare NHS Trust, 2016; 2017), it is clear that, at unit level, absenteeism due to sickness fluctuates significantly (Table 6). Furthermore, taking longitudinal data for each unit, it is clear that some units have prolonged issues with high levels of sickness. In this instance, a trust-assigned allowance of 4.6% for unavailability due to sickness would give insufficient headroom to meet the requirements of, say, the Summers View unit, which had an average sickness rate of 10.7% for the period October 2013 to June 2016 (Table 6). Consequently, to ensure safe staffing, it is important that the setting of trust targets for sickness unavailability is not confused with realistic unit-based values when determining headroom.

Continuing professional development (CPD) and study leave

Given that all nurses and midwives must complete 35 hours of mandatory CPD over a 3-year period (NMC, 2019), it would be prudent for roster policies to include a minimum of 0.6% study leave unavailability. Details from the 20 roster policies examined shows a broad spread, including two policies with no allowance for study leave. NHS Improvement (2018) recommends a study leave allowance of 3%, which compares favourably with the 2.3% average value from the roster policies examined. Ultimately, CPD/study leave should be determined by each individual's personal development plan (PDP) and agreed within the staff appraisal process, putting study leave within the purview of the unit manager.

Parental leave allowance

The NHS Staff Council (2019) noted that parental leave is a separate provision from maternity/paternity leave and provides an employee the right to at least 18 weeks' leave if caring for a child under the age of 18 years. This is contrary to McIntyre (2016) and the NHS Improvement (2018), who suggest that maternity leave is included in parental leave allowance. Furthermore, only six trusts include parental leave as a component of headroom in their roster policies. Moreover, there is some confusion regarding inclusion of maternity/paternity leave in parenting leave. The NHS Staff Council (2019) has clearly stated that ‘Parental Leave is a separate provision from either maternity or maternity support (paternity) or adoption leave’, while McIntyre (2016) and NHS Improvement (2018) have suggested that it be included in ‘parenting’. The Safer Nursing Care Tool (Shelford Group, 2013) specifies a headroom of 22% but this does not include maternity leave—which, as Hinchliffe (2013: 21) noted, ‘can be compromising when more than 50% of Leeds Teaching Hospitals Trust wards have a maternity leave rate of greater than 3%.’ Four of the roster policies that incorporated a parenting allowance included maternity/paternity leave.

Non-clinical work

Only nine of the policies reviewed included an allowance for non-clinical days, with an average allowance of 1.8%. This is surprising given the management, administration and reporting responsibilities of unit managers (Wise, 2007). Clearly, for care targets to be met, more research is required to assign an appropriate degree of non-clinical time to the headroom allowance.