Financial pressures

The evidence suggests that periods of significant growth in income for publicly funded hospitals are associated with reduced crude productivity. Conversely, financial pressure on hospitals has consistently been found to be associated with greater crude productivity – as measured by activity rising faster than real spending on inputs (Appleby, Reference Appleby1999). However, it is not clear at what level a financial squeeze on hospitals might inevitably impact adversely on quality of patient care.

Recent analysis of the experience of the NHS in England over the period 2000–2010 suggests quality-adjusted productivity fell during a time of increased budgets (National Audit Office, 2011). Subsequent analysis, though, has emphasised the limitations of this study, particularly in respect to taking into account changes in quality, which have tended to be overlooked following its original publication (Black, Reference Black2012).

Elsewhere, evidence from cross-country comparisons appears to show that relative abundance is associated with relative inefficiency. Appleby (Reference Appleby2008) compared productivity in England and Scotland. Historically Scotland has spent more on health per capita than England (16% higher in 2005/6, falling to 9% higher in 2009/10 – HM Treasury, 2011). However, crude ratios of outputs to inputs suggest the Scottish NHS is less productive than England: operations per consultant are around 12% less than in England; admissions to hospital 18% less and inpatient and day case episodes per bed around 40% less. Connolly et al. (Reference Connolly, Bevan and Mays2010) extended this comparative analytical work within the United Kingdom in 2010 and confirmed these findings, which are also consistent with evidence from other sources.

Evidence from the United States suggests that efficiency improvements can be achieved by raising the volume of activity, rather than cutting costs. For example, Bazzoli et al. (2004–200Reference Bazzoli, Lindrooth, Hasnain-Wynia and Needleman5) looked at the consequences of the tariff squeeze for US hospital operations following the implementation of the Balanced Budget Act in 1997, which reduced payments to hospitals under Medicare mainly by restricting inflation updating of diagnostic-related group (DRG) payment rates over a number of years. Although the authors did not investigate hospital efficiency as such, they found that hospitals under the greatest financial pressure tended to increase both inpatient and outpatient Medicare volumes, and to maintain staffing and bed levels. In turn, this implies that within a case-based payment regime health care spending may rise following a tariff squeeze, if the volume of activity is allowed to rise irrespective of the underlying levels of population health need. Thus, if cost containment is the overriding concern, it may be necessary to discourage additional volume. Rosko (Reference Rosko2004) found similar efficiency gains in the United States following the Balanced Budget Act.

Payment methods

Payment methods and their reforms can improve hospital efficiency. For example, one method consistently associated with lower length of stays is national, fixed-price, case-based payment systems using DRGs or health care resource groups (HRGs) (Busse et al., Reference Busse, Geissier, Quentin and Wiley2011). Such payment systems have been introduced in a number countries in recent decades (Kimberly et al., Reference Kimberly, de Pouvourville and D’Aunno2008). Although their impact on activity is varied, Hurst and Williams (Reference Hurst and Williams2012) argue that fixed-price, case-based payments can encourage efficiency by rewarding hospitals for activity and cost-reduction, by informing purchasers about relative prices, and by encouraging competition on quality rather than price. Recent experience with ‘Payment by Results’ in England suggests that such systems can be managed centrally to encourage hospitals to lower unit costs, to encourage changes in case mix, or to incentivise quality improvements (Farrar et al., Reference Farrar, Scott, Ikenwilo and Chalkley2007). However, there may also be less desirable side-effects such as: increasing unneeded as well as needed activity; increasing transaction costs, increasing risk selection and by encouraging upward drift in DRG/HRG coding by hospitals (Farrar et al., Reference Farrar, Scott, Ikenwilo and Chalkley2007). In some circumstances, case-based payments have been accompanied by technical efficiency but lower cost efficiency (full cost-efficiency being the least-cost mix of inputs to produce any one output or outcome). For example, in Norway, Biørn et al. (Reference Biørn, Hagen, Iversen and Magnussen2002) found that the introduction of case-based funding for hospitals led to an increase in needed activity along with a rise in relative remuneration rates for staff, lowering cost-efficiency. This suggests that the effect of introducing case-based payments on efficiency will depend partly upon whether such changes are accompanied by expanding or contracting budgets and activity, and ideally would be assessed against overall levels of population health need.

Competition

US studies suggest that competition can be a driver of efficiency (Rosko, Reference Rosko2001; Jiang et al., Reference Jiang, Bernard and Begun2006a, Reference Jiang, Bernard and Begun2006b; Huerta et al., Reference Huerta, Ford, Peterson and Brigham2008; Valdmanis et al., Reference Valdmanis, Rosko and Mutter2008). A key point in this literature is the effect of competition between hospitals on quality. Here it has been widely shown that competition with fixed prices improves quality (Cooper et al., Reference Cooper, Gibbons, Jones and McGuire2011; Gaynor et al., Reference Gaynor, Moreno-Serra and Propper2011). However, evidence about the effects of competition on quality when prices are market determined are more mixed. A recent study by Propper et al. (Reference Propper, Burgess and Gossage2008a), suggest that although price competition in the NHS internal market of the 1990s had a favourable effect on waiting times among acute hospitals in England (between 1991 and 1999), it had an adverse effect on acute myocardial infarction (AMI) mortality. In other words, an improvement in observable quality (waiting times) was achieved at the expense of unobservable quality (mortality). More widely, debates continue about the reliability of the evidence on competition either way. Bevan and Skellern (Reference Bevan and Skellern2011), for example, question whether measures of AMI morality are the right metric to explore the effects of competition on elective care arguing that the effect of patient choice for elective surgery on the quality of elective surgery can only be tested by measures specific to elective surgery. Moreover, it might also be questioned whether econometric ‘difference in difference’ studies (such as Propper et al., Reference Propper, Burgess and Gossage2008a; Cooper et al., Reference Cooper, Gibbons, Jones and McGuire2011; Gaynor et al., Reference Gaynor, Moreno-Serra and Propper2011), even when carried out rigorously (as they are in these studies), can account for the effects of competition on quality, given inherent difficulties in accurately measuring competition and change in competition over time, in estimating the overall likely health gain achievable by the interventions, and in establishing cause as well as association.

Performance monitoring and management

External performance measurement and management are often considered to be major tools for improving efficiency, with health care systems paying significant policy and management attention to the use of targets, benchmarking, public reporting and rewarding performance of health care providers. However, the evidence is mixed on external reporting and its effect on the performance of health care organisations. Fung et al. (Reference Fung, Lim, Mattke, Damberg and Shekelle2008) undertook a major review of the evidence on whether publishing patient-care performance data in the United States improves the quality of health care (selection of providers, quality improvement activity and outcomes). The authors found inconsistent results with some studies showing stimulation of quality-improvement activities and others not. Later criticism by Hibbard (Reference Hibbard2008) suggests that at least some of this variation may depend how information on performance is published. In England, Bevan and Hood (Reference Bevan and Hood2006) reviewed the effect of the publication of performance ratings together with targeting some selected indicators of quality for NHS trusts in England between 2001 and 2005. They concluded that the then ‘star rating’ system had improved performance on key targets such as waiting times. Similarly studies by Propper et al. (Reference Propper, Sutton, Whitnall and Windmeijer2008b, Reference Propper, Sutton, Whitnall and Windmeijer2010; cf. Connolly et al., 2010), comparing evidence from performance in England and Scotland appear to show that the policy of setting targets for elective care waiting times in the English NHS shortened waiting times and increased admissions for elective care. However, such policies were enacted at a time of rising budgets. There is an open question, therefore, about whether public reporting can help maintain quality when budgets are under pressure.

Pay for performance

Pay for overall performance at the organisational level is another apparently attractive policy tool for those seeking to improve productive efficiency. Mehrotra et al. (Reference Mehrotra, Damberg, Sorbero and Teleki2009) identified evaluations of the Centre for Medicare and Medicaid Services’ Premier Hospital Quality Incentive Demonstration Project (HQID) comparing HQID hospitals with control hospitals. The HQID offers financial rewards for good quality care as assessed by 33 process and outcome indicators of quality for five clinical conditions. The evaluation suggested that the scheme generated a 2–4% improvement in process quality in the intervention hospitals (fuller discussion in Tanenbaum, Reference Tanenbaum2009). However, in a study by Ryan (Reference Ryan2009) of over 11 million individual Medicare patient records in 3570 acute hospitals from 2000 to 2006, no evidence was found that HQID had a significant effect on risk-adjusted mortality for AMI, heart failure, pneumonia or coronary artery bypass graft patients. A pay-for-performance scheme similar to the HQID was introduced into 24 acute hospital trusts in north-west England in 2008. Trust performance was assessed by a score based on mortality and re-admission rates (variously) in five clinical domains. A three-year evaluation of this scheme using individual patient data suggested that the introduction of pay for performance was associated with a reduction in mortality of 1.3 percentage points in the combined mortality for the three conditions studied. It was not clear, though, whether these favourable results were entirely due to the payment of bonuses, and evaluators argued that the details of the implementation of incentive programmes and the context in which they are introduced – in particular investment in organisational development and support – may have had an important bearing on their outcome (Sutton et al., Reference Sutton, Nikolova, Boaden, Lester, McDonald and Roland2012).

Leadership and management practice

Leadership, management and organisational culture are widely believed to influence productivity. Bloom and Van Reenen (Reference Bloom and Van Reenen2006) collected information on 18 management practices from 732 medium-sized firms in the United States, France, Germany and the United Kingdom and found that they were strongly associated with firm productivity, profitability and firm-survival rates. Similar results have been published recently for the NHS. Bloom et al. (Reference Bloom, Propper, Seiler and Van Reenen2010) interviewed managers and clinicians in orthopaedics and cardiology in acute hospitals in England using an evaluation tool of 18 management practices falling into four groups: patient pathways; targets; performance measurement and management of staff; staff rewards and sanctions. They found that their measure of management quality was favourably correlated with indicators of hospital performance such as mortality rates, waiting times, financial performance, staff satisfaction and the overall rating from the Health care Commission, although authors warn that this may not imply causal relationships.

Vina et al. (Reference Vina, Rhew, Weingarten, Weingarten and Chang2009) investigated the association between various organisational factors and performance among 92 hospitals which had implemented ‘pay-for performance’. They compared management between hospitals in the top two and bottom two deciles of performance. More top-performing hospitals reported having: adequate human resources; support of nursing staff; and an organisational culture that supported coordination of care, pace of change, willingness to try new projects, and a focus on identifying system errors rather than blaming individuals. They also reported more use of clinical pathways; organisation into multi-disciplinary teams; and use of computerised physician-order entry into the hospital.

Manager–physician co-operation

There is a significant literature that associates hospital productivity with the degree of cooperation and engagement between general managers and physicians. In a recent review of the US literature on hospital/physician collaboration, Burns and Muller (Reference Burns and Muller2008) reported on a previous study that suggested that a key distinguishing feature between high and low performing hospitals was ‘the level of both hospital executives’ and physicians’ behavioural skills’. These skills included: physicians’ trust in hospital executives; mutual respect and support; frequent and candid communication; physicians’ involvement in all clinically related decision making; transparency of hospital finances to physicians; consistent physician and hospital executive leadership over time (i.e. low turnover); physician leadership development; and physician-led efforts to promote a sense of shared economic risk (Burns and Muller, Reference Burns and Muller2008).

Technology adoption

The availability and diffusion of new, cost effective technologies might be thought of as a critical determinant of gains in hospital efficiency. However, it is often difficult to reap net expenditure savings from them. Many advances – such as magnetic resonance imaging – represent an addition to, rather than substitution for, the range of diagnostic options in a hospital. New therapies offering care for hitherto untreatable, or partially treatable, patient constituencies, also put upward pressure on demand and expenditure, with such technologies under patent frequently introduced at premium prices. In a Dutch study of the impact of 63 broad innovations on expenditure in 66 general hospitals between 1995 and 2002, only one group of technologies was found to have significant impact in terms of reducing costs, this being the adoption of ICT for various hospital processes (Blank and Van Hulst, Reference Blank and Van Hulst2009). Similar issues might be raised with respect to bodies such as the National Institute for Health and Care Excellence (NICE). In England, NICE is tasked with appraising the cost-effectiveness of all technologies falling within its remit in order to ensure that purchasers within the NHS receive value-for-money. However, recent work by Claxton et al. (Reference Claxton, Martin, Soares, Rice, Spackman, Hinde, Devlin, Smith and Sculpher2013), has suggested that NICE’s purported threshold for cost-effectiveness [widely considered to be around £20,000 – £30,000 per Quality-Adjusted Life-Year (QALY) gained] may be higher than that prevalent in the wider NHS (£12,936 per QALY gained), meaning any NICE-evaluated technologies approved for purchase could displace existent technologies with better cost-effectiveness.

There is some evidence that the rate of diffusion of new, cost effective technologies varies significantly across hospitals with profound consequences for outcome. This was vividly demonstrated by Beech and Morgan (Reference Beech and Morgan1992) when tracing the first favourable trial of day surgery (for hernia repair) in the United Kingdom back to 1955. They pointed out that the (then) Ministry of Health began advocating day surgery in 1967, yet there was still a need to recommend use of day surgery to late adopters nearly 55 years after this first trial (NHS Institute for Innovation and Improvement, 2009).

An important question arising from the situation facing health systems currently is whether increased financial pressure will tend to slow the rate of adoption of new, cost-effective technologies inhibiting valuable technological advances or potential improvements in productivity. All other things being equal, the answer to this question is likely to be ‘yes’. Tightening the health budget will be likely to raise the cost-effectiveness threshold for accepting any new technology, potentially squeezing out some new technologies.

Benchmarking operational processes

Operational processes lie at the heart of the question of productive efficiency in hospitals. Improving hospital efficiency is a matter of changing productive processes to increase output (including quality) for any given level of inputs, or to reduce one or more input for any given output. In their empirical work, Hurst and Williams (Reference Hurst and Williams2012) found that tactics for improving productive efficiency within the processes of English hospitals included those set out in box 2 below. Evidence from research suggests that as well as efforts to improve average productivity across all hospitals, there is almost always variation across hospitals in the relevant parameters, and that benchmarking can help to point towards ‘the efficiency frontier’.Footnote ³ Opportunities to improve efficiency in any one hospital at the operational level will also depend heavily on the initial conditions in that hospital and in the short run some inputs will be fixed.

Reducing length of stay is a clear target for improving efficiency, and there is evidence from the American literature that shorter length of stay has played a part in improving apparent efficiency in US hospitals. A Norwegian study (Martinussen and Midttun, Reference Martinussen and Midttun2004), used frontier techniques to measure variations in hospital efficiency across 51 hospitals, suggesting that increasing day surgery was positively and significantly linked to hospital efficiency over the period 1999–2001. However, in order for efficiency gains to be realised, reduced lengths of stay must result in bed and staff reductions and concentration of services (Clarke, Reference Clarke2002).

Similarly, a substantial body of research into evidence of variations in hospital admission rates (e.g. Wennberg, Reference Wennberg2011), suggests there may be potential to reduce admissions, together with their attendant costs, without any lose in health of those not treated.

Following their empirical work, Hurst and Williams (Reference Hurst and Williams2012) noted that there can sometimes be a ‘lack of synergy’ between system level objectives to create efficiencies in health care, and objectives at the operational level on the part of hospitals to pursue increased efficiency. This suggests external policy drivers, internal management processes and priorities of hospitals need to be better aligned. Indeed, the persisting variation in performance across hospitals (within and across health systems) in relation to benchmarks of processes such as those set out in Box 2 points to the importance of policy levers which can move providers to improve the productive efficiency of their basic operational activity.

Use of staff and skill mix

Optimising the use of staff is critical for improving hospital efficiency, particularly given the high percentage of hospital expenditure devoted to personnel. Research evidence suggests that staff productivity can be raised by reducing the use of temporary staff, cutting overtime, improving occupational health services and taking measures to reduce sickness absence (Hassan et al., Reference Hassan, Austin, Celia, Disley, Hunt, Marjanovic, Shehabi, Villalba-Van-Dijk and Van Stolk2009). Benchmarking of workforce data is also highlighted as a way of reducing spare capacity among staff (Fogel, Reference Fogel2004). The importance of such benchmarking by hospitals is underlined by work from the NHS Institute (2006), which revealed that patient admissions and finished consultant episodes per consultant vary by over 100% between NHS trusts in England, the implication being that if ways could be found to bring all trusts up to the average for consultant productivity, consultant staffing could be reduced by 7% nationally.

In the case of nursing, there is literature which suggests positive associations between various indicators of patient quality and a number of factors, including aspects of the nurse working environment; richer nursing skill mix; and higher nurse staffing levels. Lower nurse staffing levels are associated with lower nurse job satisfaction, higher nurse burnout and higher nurse turnover. This literature has been reviewed by the US Association for Health Care Research and Quality (Kane et al., Reference Kane, Shamliyan, Mueller, Duval and Wilt2007) concluding that higher registered nurse staffing is associated with less hospital-related mortality, cardiac arrest, hospital-acquired pneumonia and other adverse events.

Limited evidence also suggests that a richer nurse skill mix is associated with lower mortality. It is possible, however, that these associations are indirect – both nurse staffing characteristics and patient outcomes might be attributable to variation in the management and quality cultures across hospitals. An English study on variation in nurse staffing levels (Rafferty et al., Reference Rafferty, Clarke, Coles, Ball, James, McKee and Aiken2007) contained very similar findings to those gathered in the United States. If the associations between nurse staffing and outcomes were causal, findings from the United States would suggest that there is a clear business case for many hospitals to improve nursing skill mix but not necessarily to increase overall nursing numbers (Needleman et al., Reference Needleman, Buerhaus, Stewart, Zelevinsky and Mattke2006).

Positive associations can also be found between levels of physician staffing and patient outcomes. For example, Pronovost et al. (Reference Pronovost, Angus, Dorman, Robinson, Dremsizov and Young2002) conducted a literature review of studies of physician staffing variations across intensive care units (ICUs). They concluded that high intensity medical staffing was associated with lower ICU and hospital mortality and with reduced ICU and hospital length of stay. Given that most, if not all, hospital care depends on teamwork, this leads to wider questions about hospital staff skill mix generally. There is literature which suggests that nurses can substitute successfully for physicians in some circumstances, but most of this literature appears to relate to primary care and there is little or no evidence available on the cost effectiveness of these arrangements (Buchan and Calman, Reference Buchan and Calman2005).

Management of back office functions

In the NHS in England, independent analysis of back office functions (e.g. finance, human resources, estates management, information management and technology, governance and risk) has suggested that at least £600 million (of a total of £2.8 billion) could be saved each year if NHS bodies were able to standardise good practice (Fairhurst and Reilly, 2010). This drew on prior work by the UK Treasury that identified the potential to save 20–30% of back office expenditure across the public sector (HM Treasury, 2009). Benefits of such standardisation have been cited as improved efficiency, spending on buildings and equipment, better information and data, improved customer service, and a basis for comparisons of performance and processes across organisations.