You are here: Home > Healthcare policy
Health policy review
Summer 2006
Issue 2: Different approaches to reforming health services
NHS productivity: the role of quality - Jon Ford
Introduction
Productivity growth is measured by relating changes in output to changes in one or more inputs to production. The most common productivity measure is labour productivity, which links changes in output to changes in labour input. Analysis of productivity in the NHS has, until recently, been crude and dominated by considerations of technical as against allocative efficiency. By the former we mean simply the effectiveness with which a given set of inputs are used to produce outputs. This focus takes no account of the appropriateness of either inputs or outputs. Allocative efficiency, on the other hand, occurs when organisations allocate resources to produce those goods and services most valued by society, conventionally measured by the price it is prepared to pay but can be generalised in the absence of markets to embrace considerations of utility or desirability.
The large investment in the NHS made following the Wanless review has focused renewed attention on productivity in the NHS. Wanless had the following to say about the level of increased spending recommended by the review:
To aim for too rapid a rate of activity growth risks hitting capacity constraints and driving up costs rather than activity. However, aiming too low means delaying the improvements in quality and access across the service [go to note 36].
As part of the 2004 spending review, the Department of Health (DH) committed to achieving annual efficiency benefits in England of at least £6.5 billion by 2007/08. Up to £2.9 billion of the expected gains were anticipated to come from making better use of staff time, an initiative known as Productive Time. Productive Time was, in the words of the DH, designed to align the modernisation strategies for people (Pay and Workforce Reform), process (10 High Impact Changes) and technology (Connecting for Health) in order to maximise service improvement.
Past practice
The health departments have traditionally reported to the review body on doctors and dentists remuneration the failure of NHS activity (finished consultant episodes – FCE) to keep pace with the growth of inputs – most noticeably consultant numbers (figure 4). In recent years, however, this emphasis has changed, in recognition of the crudity of this approach. Thus in 2005, the DH while still relating, the number of consultants to weighted activity output had the following to say on the matter of consultant productivity:
A number of factors are likely to contribute including the nature/complexity of activity but it is likely that working practices and delivery processes are also a factor. The data is being shared with trusts for them to benchmark performance with peer organisations and identify where genuine productivity improvements are possible through local redesign [go to note 37].
Overall NHS productivity has until recently been measured using the Cost Weighted Efficiency Index (CWEI). The CWEI is the ratio of the growth in NHS activity to the growth in real NHS expenditure. The activity measure is derived by calculating the change in activity in certain areas (inpatients, outpatients etc) weighted by each area’s share of expenditure. Real expenditure (that is expenditure deflated by a cost index) is used as a proxy for factor inputs.
The main criticisms of this index are that it:
- is essentially secondary care-based and fails to count a large amount of NHS activity
- takes no account of quality, such as better health outcomes or shorter waiting times
- has little ability to measure changes in case-mix; and
- perversely records shifts in activity from a high to a low cost area as inefficiency.
The Office of Health Economics (OHE) has highlighted the difficulties in measuring value for money in healthcare [go to note 38]. It suggested that on the most optimistic measure, healthcare activity rose between 1999 and 2004 by less than 4 per cent per annum whereas health spending increased in real terms by between 6 and 7 per cent. The OHE argued however, that current measures of activity do not tell us whether the NHS is providing more health benefits to more patients. Using the example of statin use, it pointed out that a reduction in heart attacks from this practice would, because of its effect on procedures and drug costs, be deemed to affect productivity adversely. Similar criticisms of health output measures in relation to national accounts have been made by the Atkinson Review (see below), which is looking at ways of measuring public sector productivity.
Mindful of such criticisms, the DH developed an interim measure, the cost efficiency measure, to assess progress against the 2002 Spending Review value for money Public Service Agreement target. In 2003-04, this measure suggested that value for money through cost efficiency increased by 2.1 per cent. The cost efficiency measure is calculated by comparing increases in NHS expenditure adjusted for both input cost inflation and increases in expenditure on improving the quality of NHS services, with increases in NHS outputs as calculated by an NHS output index.
This index is derived mainly using data published in the National Schedule of Reference Costs and in 2003-04 covered over 1,900 activity categories. Among the advantages of the new NHS Outputs Index were:
- a better ability to capture changes in case-mix as a result of using more detailed activity data
- reducing the downward impact on output resulting from shifting activity from inpatients to outpatients
- broadening coverage to include primary care.
It should be noted though that, though reduced, the perverse outcome of productivity loss as activity shifts to cheaper settings remains as a result of the use of cost weights to combine disparate activity.
The Atkinson Review [go to note 39], which reported in early 2005, set out its preferred method for calculating output in relation to productivity for national accounts purposes. It stated a preference for the use of output indicators (rather than activities) covering all services in as much detail as possible. It required all outputs to be quality adjusted and that they should again be cost weighted for combination purposes.
In responding to the review [go to note 40], the DH concentrated on the adjustment of output for quality. In addition to citing work commissioned from the University of York and NIESR, the DH presented analysis showing and proposing other approaches to quality adjustment. Interestingly in the light of earlier criticisms, these included using a value for statins prescriptions which focused on lives saved rather than on cost – inverting the impact of statin use on productivity criticised by the OHE.
Outputs
As mentioned above, the DH commissioned a research team from the Centre for Health Economics at the University of York and the National Institute for Economic and Social Research in March 2004 to develop new approaches to measuring NHS outputs and productivity. In its final report in December 2005 [go to note 41], the research team set out a methodology for producing a comprehensive quality adjusted index of NHS output. It identified its preferred approach to measuring quality adjusted output as a value weighted output index that attaches monetary values to the characteristics that measure quality. Bearing in mind that the data necessary to estimate this index are not currently available for all NHS activities, the team also proposed methodologies for calculating quality adjusted NHS output indices using existing data. Using as their starting point the dataset used by the DH in its cost weighted output index, they derived cost weighted indices that incorporated quality adjustments such as survival and waiting times.
Inputs
The authors also produced a new approach to measuring factor inputs particularly labour. Historically, these have been measured by deflating expenditure (see above). Direct labour inputs, adjusted for changes in working hours are to be preferred and there needs to be some recognition of changes in skill mix. Generally speaking the relative skill of the workforce is a positive quality adjustment to labour inputs and thus a negative quality adjustment to productivity. This is because more skilled labour can be assumed to be more productive. This is the approach adopted by the research team but it probably needs some refinement. For the workforce as a whole there has been a dramatic increase in non-clinical workload with management and administrative duties, teaching and training taking up an increasing share of available time. By way of example, time spent on management by whole-time consultants increased by over four hours per week between 1989 and 1998, helping to increase average total hours of work (excluding emergency recalls) from 48.3 to 51.3 hours per week. In consequence, the time available for clinical work has declined by around two hours (6%) per week. The Wanless review identified two competing trends in workforce growth. ICT investment may significantly reduce the amount of time medical and nursing staff have to spend on administration, freeing up more time for patient care. Counterbalancing this, the amount of time spent on clinical governance will increase. For its financial projections, the review assumed that 10 per cent of professional staff time would be devoted to clinical governance.
To move from labour inputs to total factor inputs, the authors used additional data on intermediate inputs and capital. The data on intermediate inputs (drugs, energy and establishment costs) were derived from expenditure data deflated by a cost index but quality adjusting these data was problematical due to the need to find an appropriate price deflator for drugs. The high prices for new drugs could in a competitive market serve as a proxy for quality improvement, but the NHS distorts this theoretical approach. For capital inputs, depreciation coupled with an estimate of implicit depreciation for short asset life inputs was the starting point but the absence of disaggregated information on types of equipment that have seen rapid technological change and change in unit cost made it hard for the authors to quality adjust these data. Short asset life inputs of course include computers and one is reminded at this point of Robert Solow’s observation, nearly 20 years old but still pertinent, to the effect that:
You can see the computer age everywhere but in the productivity statistics [go to note 42].
The retort to this paradox can only reinforce the need for quality adjustment of both inputs and outputs, being simply that they bring on the job convenience, better interactions with data and other kinds of immeasurable quality improvement.
Quality adjusted productivity
Using the cost weighted quality adjusted index of outputs and quality adjusted input indices, York/NIESR produced provisional estimates of total factor productivity growth for the period 1998/99-2003/04. Although overall NHS productivity growth remained negative, it was less so by around 0.16 per cent following the (limited) quality adjustments which the research team were able to make.
Perhaps the most important conclusion from all this work, apart from the need to continue to develop better measures as data become available, is the observation made by the team itself that for many purposes, quality adjusted measures of output and productivity growth either for particular disease groups and across NHS organisations or treatment settings will be of more value to the NHS than a comprehensive index. The York/NIESR research provides a methodology whereby, with better data, it will be possible to produce such specific output and productivity indices.