The productivity of health care

In most manufacturing and service industries, the relationship between inputs (staff time, raw materials) and outputs (goods or services provided) is a key indicator of success or failure. Economic ‘productivity’ is ‘the amount of output per unit of input’ 1. In most sectors of the economy, organisational productivity over time and between individuals or teams is monitored routinely. Until relatively recently, this relationship has been neglected in health care. However, increasingly UK and international policymaking is beginning to focus on the productivity of the health care labour force 2,3. In health care, monitoring productivity would ideally involve measuring the value of health outputs in relation to the value of inputs into health care, particularly staff time but also other resource inputs. Both sides of this relationship are highly complex to measure in terms of real value. As patients do not pay for their health care there is no direct measurement of how much they value it. This means that there is no simple measure of input. The value of health output is perhaps even more complex. In most health care systems there is routine collection of some health indicators, including life expectancy and infant mortality, but improving health involves more than just reducing mortality. Health status measures such as EQ-5D 4 and SF-36 5 are not yet used routinely in health care systems to measure population health over time, despite their widespread use in clinical trials. With complexities on both sides of the productivity equation in health care, perhaps it is unsurprising that this concept remains relatively neglected. Donabedian 6 distinguished between the structure or inputs; process; and outcomes of production of goods and services. In health care, process measures, particularly measures of activity, are often used as a proxy for health outcomes. Although highly imperfect, use of activity data may be a first step in the process of planning, appraisal and regulation of the performance of medical practitioners and other health care staff. Activity data may enable managers to determine the overall volume of services being provided, the casemix structure within overall volume, variations in activity levels, changes in procedures (e.g. the adoption and abandonment of technologies) and the distribution of health activity across social and other groups within the population. The UK government is beginning to measure activity and productivity over time in the public services, including the health service. In 2004 the Office for National Statistics estimated a productivity index of NHS ‘outputs’, (NHS treatment activities weighted by their relative costs), divided by ‘inputs’ (labour, procurement of goods and services, and capital). This showed NHS productivity to have fallen between 1995 and 2004 by an average of between 0.6 and 1.3% per year 7. The Atkinson Review 8 recommended a number of improvements in the measurement of productivity, including better measurement of output in primary care, and better measurement of dimensions of quality of health care. Research has been carried out on including ‘quality’ improvements in estimates of NHS productivity 9, with indicators including survival rates, waiting times and patient experience. Including these quality improvements in ONS estimates means that NHS productivity has changed by between −0.5 and 0.2% per year between 1999 and 2004. Adding an adjustment for increasing ‘value’ of NHS output (proxied by real earnings in the economy) shows productivity to have increased by between 0.9 and 1.6% per year over the same time period 10. Health care has seen considerable technological change, including better diagnostic tools, pharmaceutical therapies and developments in surgery. Despite changes in technology and frequent reorganisation of structures, there has apparently been a failure substantially to improve workforce productivity in the UK NHS. If technology and organisational change is not contributing sufficiently to increasing activity, it may be contributing to the quality of care, but this is a difficult hypothesis to test. Assumptions by workforce planners seem to be that the overall level of health care activity is entirely determined by the size of the workforce 11. Reward and contract mechanisms for physicians, and the inherent financial and non-financial incentives within them, are matters of substantial policy importance, and opportunities for realigning incentives, for instance in the renegotiation of the UK physician contracts, were largely missed 12. In health care, contract structures need to balance reward and regulation. Financial incentives such as fee for service payment mechanisms may stimulate activity, in which case regulatory mechanisms need to prevent over-treatment. Alternatively, salary mechanisms provide financial incentives to avoid over-treatment, but in this case regulation may be needed to ensure adequate levels of activity. In the words of Canadian health economist Morris Barer, we can ‘pay em or flay em’ 13. The design of a payment system based on real performance, aligning the incentives of health care payers and providers, requires measurement of health outcomes, with adequate risk adjustment to avoid incentives for ‘cream skimming’. While health care systems routinely collect activity information in varying forms and with differing degrees of accuracy, often these data remain unanalysed. For decades, the UK NHS has invested millions of pounds in collecting annually hospital activity data (Hospital Episode Statistics). However, until relatively recently these data have been little used by both researchers and policy makers. Researchers are now exploiting the data more fully, for example to identify the effects of different types of specialist remuneration in UK hospitals 14 and general practice fund holding in England in the 1990s 15. NHS policy makers are increasingly encouraging hospitals to ‘benchmark’ their own performance against other hospitals, using this routinely collected data source. An example of this is a recently distributed information package on variations in consultant clinical activity rates 16. This increasing use of administrative NHS data in the UK focuses on activity and begs the question of whether diagnostic and therapeutic interventions improve the health status of patients. It is complemented with process measures that are used to imply ‘quality’, but as positive outcomes remain unmeasured, these may be statements of hope rather than substance. In UK primary care an optimistically named ‘quality and outcomes framework’ (QOF) has been introduced into general practices with strong fee for service incentives to ensure compliance. These QOF targets are to some extent based on evidence 17 and have been asserted to be a ‘great success’ with delivery rates above 90%. However this assertion has to be conditioned by the absence of baseline data with which to measure change (the British may be paying GPs for what they were already doing) and the problem that achievement is based largely on self-reporting. The NHS QOF in principle incentivises the delivery of improved care for chronically ill patients such as those suffering from diabetes, hypertension, asthma and chronic obstructive pulmonary disease. Other elements of the GP contract incentivise the provision of childhood immunisation and vaccination and the screening for cervical cancer. The overlap between these activities and elements of ‘quality assurance’ in other countries is considerable. Thus in China and Kyrgyzstan there is concern about the failure of their health care systems to deliver demonstrably cost effective interventions to control chronic diseases such as hypertension and diabetes. In the United States the Rand Corporation has concluded that Americans get only 55% of the care they need as again their expensive health care system fails to care for the chronically ill 18. One element of ‘quality’ management in the USA involves surveying the benefit packages of insurers to determine the extent to which they offer, let alone deliver the very same elements that concern policy makers in the UK and elsewhere internationally. Again the focus of attention is process rather than outcome. The intriguing policy issue is why, when generic quality of life outcome measures have existed for decades, they are not used routinely in clinical practice and in the study and management of productivity. In neither the public systems of Europe nor the insurer system of the USA is there interest in applying such measures systematically, both as a means of measuring and managing productivity, and to ensure efficient consumer protection from inadequate clinical practice. The only real alternative to measuring the output of health systems in terms of QALYs is to measure in terms of willingness to pay, but in health care, almost always funded by third parties (insurers or government), this will always be hypothetical. The 19th century British nursing pioneer, Florence Nightingale, advocated the measurement of outcomes in terms of whether patients were dead, relieved or unrelieved 19. One hundred and fifty years later, health care systems still approach the measurement and management of productivity with a fixation on activity. Surely it is time to test carefully the routine use of generic quality of life measures in health care 20? Such measures of success are an essential ingredient into systems of clinical governance and in informing patient choices, as well as measuring and managing the real ‘productivity’ of health care.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.