Background
Heavy drinking affects one in four adult primary care patients in the United Kingdom (UK), [
1] contributes to 61 disease conditions, [
2] and results in a £3.5 billion annual cost to the National Health Service (NHS). [
3] Yet it is one of the most important modifiable causes of premature morbidity and mortality. [
4] There is strong evidence of effectiveness supporting screening and brief alcohol interventions at reducing consumption, [
5] but historically there have been low rates of delivery. [
6] Voluntary financial incentives were employed between 1
st April 2008 and 31
st March 2015 to encourage English general practitioners (GPs) to screen their patients for heavy drinking using a validated self-report questionnaire [
7], and deliver brief behavioural interventions to those in need of support. [
8]
Under the national alcohol-related risk reduction directed enhanced service (DES) scheme, participating practices were paid £2.38 for each newly registered adult patient recorded as being screened for heavy drinking. [
9] Whilst the scheme did not directly remunerate intervention delivery, contractual and audit guidance made clear that those patients identified as drinking at hazardous or harmful levels should be offered a brief or extended intervention. Locally-negotiated enhanced service (LES) schemes for alcohol were also introduced in certain areas. LES schemes varied in their scope and reimbursement rates, although generally they were more generous than their national counterparts and involved a more opportunistic approach to screening. For example, in one area of Northern England, practices received £8.00 for each registered patient aged 16+ (excluding newly registered patients covered via the DES) who screened positive for risky drinking and received brief advice. Both national and local schemes were voluntary, and separate to the Quality and Outcomes Framework (QOF), the principal UK incentive scheme, which links up to 25% [
10] of GPs’ income to performance against a series of clinical and organisational priority areas [
11]. Currently, the QOF does not include specific payments for alcohol interventions, although GPs are expected to record alcohol consumption in some disease management areas such as hypertension, coronary heart disease and mental health.
Financial incentives have been used to stimulate improvements in healthcare quality in the UK and other parts of the world. [
12] Yet the extent to which this approach has been effective remains contested, [
13] with concerns voiced over the perverse and unintended consequences of structuring a health system around targets and process. [
14‐
16] In theory, routine data provide a timely, cost-effective and comprehensive information source to support evaluation of incentive schemes. [
17] However, evidence highlights their low sensitivity in capturing the management and treatment of complex and chronic conditions. [
18] Moreover, even where such data suggest improvements in health-related outcomes, a common criticism of pay-for-performance programmes is that they merely promote better recording of care rather than better care itself. [
19] At the extreme end of the scale, there have also been accusations of GPs ‘gaming’ the system, by manipulating recording of care to boost financial reward. [
20]
There has been limited research into the impact of financial incentives on alcohol prevention work in England. [
21] The primary aim of our study was to assess the impact of two specific national and local pay-for-performance schemes to encourage screening and brief alcohol intervention delivery using routinely recorded Read Code data. Given the use of routine data as a proxy measure of care under such incentive schemes, our secondary aim was to examine the value of such data from the GP perspective.
Discussion
The introduction of financial incentives in England appears to have had some success in encouraging primary care providers to identify and support patients to reduce heavy alcohol consumption. However, whilst rates of screening and brief alcohol intervention delivery were higher in incentivised practices, this trend needs to be set against very low levels of recorded activity overall. Since the national DES incentivised the screening of newly registered patients only, a relatively small group within the wider patient population, such low rates may be expected. Our estimated rates of short screening test administration amongst newly registered patients indicate a more sizeable reach for the national enhanced service scheme. However it must be emphasised that these rates are estimates only: as the 100% + rates achieved by two of our practices indicate, alcohol screening tests were conducted with both new as well as existing registered patients.
Previous research by Purshouse et al suggests that if such a screening programme is implemented effectively, it would achieve around 40% coverage of the registered adult population over a ten year period, leading to reduced health service costs and increased health benefits [
36]. However, this scenario is based on an assumption that a brief intervention for alcohol is delivered immediately after a patient screens positive for risky alcohol consumption. In many of our practices, there seemed to be a mismatch between screening patients for heavy drinking, and the delivery of subsequent advice or counselling about alcohol. Available prevalence data would suggest that approximately one in four patients [
37] are likely to be drinking excessively, meaning that one might expect that at least a quarter of all patients screened would have received an alcohol intervention. In reality, for some practices, intervention delivery rates appeared much lower than expected. Yet, in others, the rates were much higher when compared with the proportion actually screened. We could not determine if this was due to an absence of coding or a lack of follow through to advice or counselling. We were not told of any deliberately erroneous or ‘over-coding’ by GPs incentivised for enhanced performance. However our interview findings suggest that additional factors beyond monetary incentives shaped routine practice. In particular, despite consistent evidence to the contrary, [
5] many physicians remain unconvinced of the effectiveness of brief alcohol interventions: a barrier to delivery raised in previous implementation literature in this field [
38].
The strength of our study was the mixed methods design, which helped provide a rich and nuanced understanding of the impact of financial incentives on GPs’ delivery and recording of screening and brief alcohol interventions. By using routinely recorded data as a measure of activity, we also provide a valuable insight into the rates of alcohol prevention work being delivered in real world as opposed to research trial settings.
However, there are several limitations to our data. Practices were not randomly sampled and thus there was a potential for self-selection bias. [
39] In the absence of corroborating observational data (such as consultation recording), it was only possible to assess activity rates via formal coding. It is possible that these data were not an accurate record of all care provided, [
40] particularly given our searches were limited to Read Coded rather free text data. Moreover, as we extracted aggregated counts of each variable of interest at practice level, there is potential for double-counting: that is an individual patient may have been screened twice with both shorter and longer tools; or received both brief advice as well as an extended intervention during the period of interest. Changes in the Read Codes used to record alcohol interventions since 2008 make it challenging to compare trends pre and post the introduction of financial incentives. Thus whilst our study suggests there are significant differences in delivery rates between incentivised and non-incentivised practices, these trends cannot be interpreted as causal.
The majority of interviewees were drawn from practices signed up to at least one enhanced service for alcohol. Interview accounts may have differed if we had recruited more ‘non-incentivised’ practitioners, but these were less inclined to be recruited. Further, whilst overall, recorded levels of alcohol prevention activity were significantly higher in incentivised practices, rates at individual practice level were nevertheless low even for some incentivised practices (P06, P07 and P16), suggesting that our sample captured a range of perspectives. In addition, despite the central role they appear to play in delivering alcohol screening tests, we did not interview nurses in this study. This study sampled 16 practices and interviewed 14 GPs across two NHS areas in one region in England, representing approximately eight per cent of the total practice population. In comparison to other parts of the UK, Northern England is relatively deprived [
41] and has higher average alcohol consumption [
42], potentially limiting the generalisability of our data. Whilst the low delivery rates reported here are unlikely to be higher elsewhere, a larger study would be required to validate our findings.
To some extent, our findings support previous research, suggesting that financial incentives can have positive impacts on screening and brief alcohol intervention delivery in primary care. [
21,
43,
44] Interestingly, our results do not indicate that there was any cumulative effect for practices that were signed-up to both the local and national incentive schemes. Given that the GP interviews suggested that routine FAST or AUDIT-C delivery was more likely when embedded within nurse-delivered practice and associated recording systems (such as registering a new patient, as incentivised by the DES), limited instances of opportunistic screening (as incentivised by the LES) might be expected.
However, the recorded activity rates reported here were comparatively low for all outcomes of interest, irrespective of financial incentive status. Alongside the design of the scheme, this may be also due to the low level of remuneration associated with the national DES in particular. For example, practices participating in the QOF+ scheme assessed by Hamilton et al, were paid up to £5,607 for screening and delivering a brief intervention to all eligible patients. [
45] DES practices in this study would have received just £2.38 for each newly registered patient screened, meaning they would need to screen around 2,300 patients per annum to achieve similar financial returns. Our qualitative findings suggest that differing incentive schemes are not seen as equal by GPs. Specifically, the QOF has the largest effect on practice income, so the delivery and recording of enhanced service activities are often accorded lower priority. [
46]
At the same time, a strong theme from our interviews was the challenge that GPs experience when recording complex and potentially stigmatising conditions such as harmful alcohol consumption. In particular, their concern to preserve ‘patient-centred’ consultations sometimes clashed with the need to record simple diagnoses and outcomes, especially when uncertainty or sensitivity was at play. [
47] This tension has been described as a
“rational-reality gap”, [
48] requiring clinicians to maintain a
“dual orientation” towards coding. [
49] Whilst other research suggests that financial incentives such as QOF serve to promote an increasingly biomedical agenda in terms of the management of chronic and complex health issues in primary care [
50], our results suggest more nuanced behaviour on the part of GPs. Importantly, we found numerous examples of GPs adapting the system to allow their behaviour to more closely align with their preferred ‘patient-centred’ approach, with the ‘tick-box’ elements of alcohol-related care devolved to nurses.
Conclusions
This study provides further evidence that policy initiatives that focus solely on the extrinsic motivations of GPs, such as financial incentives, are unlikely to have the desired level of impact without acknowledging the values, attitudes and beliefs that also shape care. The incongruity we observed between recorded rates of screening versus actual alcohol intervention delivery also highlight the potentially distorting effects of pay-for-performance on healthcare recording, and in particular, of incentivising process as opposed to outcomes [
51]. Our findings also suggest that some incentive schemes are more impactful than others, with QOF unarguably most influential as far as English primary care is concerned. Given the radical reduction to the number of clinical indicators covered by the current QOF, [
52] the addition of screening and brief alcohol intervention in the future seems unlikely. This is despite that the fact that over one in four patients continue to drink above recommended levels, [
37] whilst three quarters of the English population do not have any of the diseases listed in the QOF. [
53] Since April 2015, the national enhanced service for alcohol has been withdrawn, although local level incentive schemes remain in place for some areas [
54], and there is now a contractual requirement for practices to identify newly registered adult patients drinking above recommended levels [
55]. Based on the findings from this study, which highlight the substantial challenges experienced by GPs seeking to prioritise non-incentivised care over their routine management of QOF conditions, this seems a risky strategy. If NHS England is to deliver on its promise of “hard-hitting” action on tackling risky lifestyle behaviours in the future, [
56] a fresh look at the financing and organisation of preventative care is urgently required.
Acknowledgements
The authors would like to thank the GPs and other practice staff that participated in the study.