The Quality and Outcomes Framework (QOF), the pay-for-performance protocol in general practice, has been the subject of significant debate since it was rolled out in 2004. At the core of the debate is whether the quality of general practice can be quantified for the purposes of performance management. There are concerns that the QOF has facilitated a “box ticking culture” which undermines the doctor-patient relationship and incentivises the overuse of pharmacological therapies (see Gillam, 2010). Recently, the Scottish NHS has announced plans to lessen the use of economic incentives in general practice and move to a system of peer-to-peer quality improvement; the data collection component of the QOF is to remain in place but only as an aid for local decision-making. In the English NHS, by contrast, the QOF is to be retained but reduced as a proportion of practice income.
In this piece, I evaluate the QOF as a performance management tool through a case study of type 2 diabetes. The QOF appears to have had a positive effect on the ‘process dimension’ of diabetes care, but the complexity of type 2 diabetes is such that performance management is pushed to the limits of what it can achieve, in part due to the difficulties of applying evidence to diverse individuals. The surest route to quality in primary care is support for genuinely patient-centred care. The QOF may play a bit part in this through the incorporation of patient-centred quality measures, but ultimately a shift beyond performance management is required.
Type 2 diabetes and the QOF
Diabetes is one of the most significant public health challenges of today, affecting four million people at a cost to the NHS of an estimated £10 billion a year. It is obviously vital that the health service responds to this challenge in an appropriate way. Current trends to develop primary care and put GPs at the forefront of decision-making seem warranted, given that diabetes is a chronic condition and treatment is mostly preventative. Yet exactly how to ensure the quality of diabetes services in primary care is unclear and there is some debate over the appropriateness of the QOF, as in other conditions.
Some commentators view the QOF as having improved the processes followed in the treatment of diabetes care (see Calvert, 2009). Initial versions of the QOF in relation to diabetes contained various process indicators that are important for effectively managing the condition and detecting early-stage complications, including annual checks for blood glucose (HbA1c) and eye and foot scans. The incentives attached to the QOF, along with the information it provides on the relative performance of GP practices across the indicators contained within it, have created the impetus for improvement in diabetes care.
Still, process measures provide only a partial indication of quality and there are a number of caveats to consider. For a start, not every diabetes case will require the activities and checks contained within the QOF. High-performing general practices are likely to have a better hold on which of their patients need what and when, but they are incentivised to provide checks on a universal basis – a potential misallocation of resources. But some general practices struggle to achieve basic standards and it is among these poorly performing practices that the process improvements are most apparent. Some commentators link the QOF to gains in equity because such improvement has occurred to the greatest extent in disadvantaged areas (see Dixon et al, 2010).
That said, process improvements do not necessarily translate into improved health outcomes: it is one thing to check for poor glycaemic control or early-stage complications, another thing altogether to develop and implement an effective treatment plan. Treatment plans require physicians to work with patients, set realistic treatment goals and decide upon appropriate interventions, likely a combination of lifestyle changes and pharmacological therapies. The process is complicated by the fact that treatment regimens are intrusive such that patient nonadherence can be an issue: part of the challenge for physicians is therefore to elicit the motivation of patients to engage with the treatment plan when it might not always be a priority for them.
Later versions of the QOF sought to incentivise physicians to engage in this more complex activity of developing and implementing treatment plans. Various risk factors for chronic conditions were incorporated into the protocol to serve as intermediate outcome measures, thereby moving beyond the focus on processes. In diabetes care, the most significant of these is HbA1c: a measure of blood glucose.
On the surface, the inclusion of HbA1c levels within the QOF seems warranted because it appears to fulfil all the requirements of an effective performance indicator: evidence links glycaemic control to improved health outcomes and HbA1c is also easily measurable, already playing a fundamental role in the day-to-day management of diabetes. Moreover, because HbA1c levels are affected by health interventions, it is possible to use them to evaluate the actions of health professionals. The QOF therefore became seen as a means to improve outcomes in diabetes care by incentivising GPs to reduce the HbA1c levels of their patients.
Yet the use of targets for glycaemic control in this way has proved controversial. For instance, an initial QOF target of a HbA1c measure of below 7 mmol was raised to 7.5 due to concerns that it may have put some patients at risk of complications. Recent versions of the QOF have revised the target even further, incorporating three HbA1c measures – 7.5, 8 and 9 mmol – to allow for greater differentiation, but they still incentivise improvements in blood glucose control across the diverse population of diabetes patients.
Issues with the evidence base
Controversy over targets for glycaemic control touch upon significant issues regarding the application of evidence in medical practice. As is well known, Evidence-Based Medicine prioritises Randomised-Control Trials (RCTs) and systematic reviews of RCTs because it is thought that these provide the strongest protections against error and bias, permitting confidence in research findings: the so-called “hierarchy of evidence” (see Burns et al, 2012). Yet the populations in clinical trials and medical practice can differ markedly, creating problems when evidence is used to inform clinical guidelines and performance protocols like the QOF.
In type 2 diabetes, a widely-held consensus exists regarding the health benefits of glycaemic control. But the optimal intensity of that control is disputed, especially when it comes to older patients.
A number of studies published during 1980s and 1990s demonstrated the benefits of glycaemic control, establishing the principle “the lower, the better” when it comes to HbA1c (see Ceriello, 2010). Of these, the UK Prospective Diabetes Study (UKPDS) is the most significant, involving 5,102 patients and lasting from 1977 to 1997. The UKPDS demonstrated a correlation between glycaemic control and reductions in diabetes-related complications, if more strongly for microvascular complications (eye and kidney disease) than macrovascular complications (cardiovascular events and strokes) (see King et al, 1999). However, in the late 2000s, a clinical trial dubbed the ACCORD study was halted halfway through when a statistically significant increase in mortality was recorded in a group undergoing intensive blood glucose control.
What explains this apparent contradiction in the evidence-base? The UKPDS involved younger and relatively healthy patients, newly diagnosed with Type 2 diabetes; the ACCORD study involved older patients who are less healthy, indeed characteristics more typical of people receiving treatment in primary care. It is the ACCORD study – and other studies published around the same time – which prompted the revision of QOF targets.
Additional weaknesses in the evidence-base further complicate the situation. There are examples of pharmaceutical companies withholding critical trial data from regulators, while diabetes drugs can be evaluated, approved and marketed on the basis of their effects on HbA1c alone (see Naci et al, 2015). Rosiglitazone, for example, was approved by the European Medicines Agency (EMA) in 2000 for its potential to improve glycaemic control, only to be withdrawn in 2010 after studies emerged linking the drug to heart failure.
The Limits of the QOF: Diabetes and Shared Decision-Making
To summarise, performance management appears to have had both positive and negative effects in diabetes care. It is an open question whether ending the QOF – or even reducing its significance as a proportion of practice income – will undo the gains that have occurred in the process dimension of diabetes care. Yet the QOF appears less suited to more complex decision-making: universal targets for glycaemic control, while on the surface fulfilling the requirements of an effective performance indicator, may have exposed some patients to unnecessary harm. It may also have contributed to the overuse of drugs which reduce HbA1c but have a questionable impact on health.
It therefore appears that there are limits to what the QOF can achieve. In management parlance, it is useful for quality assurance – where the emphasis is on maintaining a certain level of quality and avoiding adverse events – but less so for quality improvement. Here, the emphasis is on creating an organisational environment which fosters continuous improvement.
Where now for diabetes care? There is increasing recognition of the importance of patient-centred care in diabetes. Exciting developments in the treatment of type 2 diabetes highlight the potential of intensive lifestyle interventions to possibly even fully reverse the condition, opening up new possibilities for patients motivated to tackle diabetes in this way. Yet a patient-centred approach is perhaps most important when it comes to the treatment of elderly patients, who are unlikely to live long enough to experience the full benefits of glycaemic control but can suffer significant treatment burdens. The harms caused by the side-effects of drugs and other inconveniences like using finger-pricking devices and restricting one’s diet can outweigh the benefits of the medical treatment of diabetes (see Barnhart, 2016; Vijan et al, 2014).
Involving patients through Shared Decision-Making (SDM) enables patients’ values, motivation and their knowledge of treatment burdens to be incorporated into treatment plans, thereby improving patient adherence and ensuring that the benefits of treatment really do outweigh the harms. Decision-aids have been developed in diabetes and other chronic conditions to facilitate discussions between GPs and patients, providing both with accessible information on different treatment options and the risks and benefits attached to them.
Where does this leave the QOF? One option is to shift the focus of the QOF away from risk factor management to patient-centred quality measures. The QOF could contribute to the dissemination of SDM, perhaps by including an indicator of the percentage of patients who report having had the opportunity to make informed decisions.
At the same time, however, SDM conflicts with performance management in important respects. SDM is an altogether different task than the delivery of pre-specified activities, demanding a revision to how evidence is applied in medical practice. Indeed, good practice is less about correctly applying evidence than adapting it to each individual case, requiring both professional judgement and the input of patients. Furthermore, the complexity of the task requires a different kind of motivational state on behalf of health professionals. Here, it is interesting to note the distinction in psychology between intrinsic and extrinsic motivation: economic incentives are appropriate for simple, standardised tasks but less appropriate for complex tasks because they create pressure to demonstrate one’s ability to external actors rather than engage in problem-solving for its own sake.
The danger is that relying on economic incentives to encourage uptake of SDM will reduce it to a tick box exercise. Hence, while the QOF may have some role to play in the future to improve patient-centred dimensions of quality, policymakers would be well-advised to ensure that GPs have enough time and support to implement patient-centred care effectively. Ultimately this will require greater investment in primary care and action to address the complaints of many GPs of excessive bureaucracy, unfunded increases in workloads and restrictions on the time spent with patients.