Article Text

Effectiveness of providing financial incentives to healthcare professionals for smoking cessation activities: systematic review
1. F L Hamilton,
2. F Greaves,
3. A Majeed,
4. C Millett
1. Department of Primary Care and Public Health, School of Public Health, Imperial College London, London, UK
1. Correspondence to Dr F L Hamilton, Department of Primary Care and Public Health, School of Public Health, Imperial College London, London W6 8RP, UK; f.hamilton{at}imperial.ac.uk

## Abstract

Objective Financial incentives are seen as one approach to encourage more systematic use of smoking cessation interventions by healthcare professionals. A systematic review was conducted to examine the evidence for this.

Methods Medline, Embase, PsychINFO, Cochrane Library, ISI Web of Science and sources of grey literature were used as data sources. Studies were included if they reported the effects of any financial incentive provided to healthcare professionals to undertake smoking cessation-related activities. Data extraction and quality assessment for each study were conducted by one reviewer and checked by a second. A total of 18 studies were identified, consisting of 3 randomised controlled trials and 15 observational studies. All scored in the mid range for quality. In all, 8 studies examined smoking cessation activities alone and 10 studied the UK's Quality and Outcomes Framework targeting quality measures for chronic disease management including smoking recording or cessation activities. Five non-Quality and Outcomes Framework studies examined the effects of financial incentives on individual doctors and three examined effects on groups of healthcare professionals based in clinics and general practices. Most studies showed improvements in recording smoking status and smoking cessation advice. Five studies examined the impact of financial incentives on quit rates and longer-term abstinence and these showed mixed results.

Conclusions Financial incentives appear to improve recording of smoking status, and increase the provision of cessation advice and referrals to stop smoking services. Currently there is not sufficient evidence to show that financial incentives lead to reductions in smoking rates.

• Financial incentives Pay for performance prevention smoking cessation healthcare
• cessation
• prevention
• primary healthcare
• surveillance and monitoring
• health Services
View Full Text

## Introduction

Preventing smoking-related diseases is a key international public health objective.1 ,2 Article 14 of the WHO Framework Convention of Tobacco Control (FCTC) recommends that countries implement ‘demand reduction measures concerning tobacco dependence and cessation’ alongside population level strategies such as the introduction of smoke-free workplaces, restricting advertising and raising taxation on tobacco. Individual-level smoking cessation interventions by healthcare professionals are highly cost effective when compared to other medical interventions and, if delivered equitably, can reduce health inequalities related to tobacco use.3–5

Despite their cost effectiveness, smoking cessation interventions have historically been underprovided in healthcare.6 Reasons for this may include practitioners' negative attitudes towards discussing smoking cessation with their patients, doubts about the effectiveness of providing smoking cessation advice, concern about compromising the doctor-patient relationship, perceived lack of expertise, lack of time in the consultation and competing priorities.7 Healthcare workers who smoke are more likely to have a negative view of their personal effectiveness in giving smoking cessation advice to patients8 ,9 and a belief that such advice is ineffective.8 Despite these views, practitioners generally consider disease prevention activities to be important and worth spending time on during the consultation.10

Pay for performance schemes are becoming more widespread in healthcare, particularly in the USA11 ,12 and the UK.13–15 They aim to improve the quality of healthcare by providing financial incentives to practitioners for achieving performance targets. Financial incentives are seen as one approach to encourage more systematic use of smoking cessation interventions and have been incorporated into many financial incentive programmes. For example, the Quality and Outcomes Framework (QOF), introduced in the UK in 2004, rewards smoking cessation activities mainly as a part of secondary prevention management of long term conditions such as coronary heart disease, diabetes and chronic obstructive pulmonary disease.16

We conducted a systematic review to examine evidence for the effectiveness of providing financial incentives to healthcare professionals on the provision and impact of smoking cessation interventions.

## Methods

### Search strategy for identification and selection of studies

We identified studies by searching the electronic databases Medline, Embase, PsychINFO, the Cochrane Database of Systematic Reviews, ISI Web of Science and sources of grey literature. The search terms are described in full in online only appendix A on the Tobacco Control website. We included papers in languages other than English in the search. The titles and abstracts of those studies identified by the initial searches were scanned, and for those that appeared relevant the full paper was obtained and reviewed. Further papers were identified by looking at the citations and reference lists of review papers and papers identified in the initial search.

### Methods of review

Inclusion criteria for studies to review:

1. Types of studies: randomised controlled trials (RCTs), controlled trials and observational studies with a before and after design reporting quantitative results.

2. Participants: participants aged 15 and over, with and without chronic disease, registered with any healthcare provider.

3. Types of financial incentives: studies that examined the effects of financial incentives (pay for performance) for individual and groups of healthcare providers (doctors, nurses or other members of healthcare teams) to provide smoking cessation advice, referral and/or prescription of medication to help with smoking cessation.

Exclusion criteria:

1. Studies that examined financial or other rewards to patients, patient competitions, or provision of reduced cost or free medication to help with smoking cessation (unless this was associated with a provider financial incentive).

2. Papers that reported results as a composite quality score including other measures of chronic disease management if it was not possible to isolate impacts on smoking-related activities.

### Assessment of methodological quality, data extraction and data synthesis

Two researchers (FLH and FG) scored each paper for methodological quality using a modified version of the Downs and Black guidelines17 for assessing the quality of randomised and non-randomised studies of healthcare interventions. The scoring system ranged from 1 (poor) to 4 (excellent). Any differences of opinion were resolved by discussion with CM. FLH and FG extracted the numerical results from the identified papers for the results section of the review. FLH used the data to estimate the effect size (OR) for each outcome for each study if not reported.

## Results

### Search results

The flow chart of our search strategy for included studies is summarised in figure 1. We identified 18 studies for inclusion in the review.

Figure 1

Flow diagram of literature search and study identification.

### Included studies

There were two sets of duplicate papers, reporting results using the same data. For the first set, Twardella and Brenner 200718 and Salize et al 2009,19 we included the later paper as it also reported cost data. For the second set, Coleman et al 200720 and Coleman et al 2010,21 we included the earlier paper as the numerator and denominator variables were reported in full, whereas the later paper only showed graphs. The final set of studies included 10 observational before and after studies examining the effect of QOF in the UK and 8 studies looking at specific financial incentives for smoking cessation. Tables containing full details of each included study including setting, design, intervention and outcomes are available on the Tobacco Control website (see online only tables 1 and 2).16 ,20 ,22–36

### Intervention, setting and study design

A total of 10 papers looked at the effects of QOF in the UK. The financial rewards available through QOF for smoking cessation activities are paid to general practices rather than individual clinicians and depend on the achievement of points for quality targets in chronic disease management rather than improvements from baseline. There are a maximum of 1000 points available per practice for achieving quality targets, of which 79 are for smoking indicators, payable on a sliding scale (see online only appendix B on the Tobacco Control website). Practices are paid on average £130.50 per point in 2011/12, representing a potential maximum income for smoking cessation work of £10 309.50 (US$16 325 at the exchange rate on 15 September 2011). The financial rewards are paid to general practices rather than individual clinicians. We report these papers separately as several looked at smoking cessation in patients with particular chronic diseases such as coronary heart disease, asthma, diabetes and stroke while others looked at smoking cessation for all registered patients, and we wanted to see if there would be any difference between these two approaches. In addition, the papers used data from different regions of the UK, apart from two that used data from representative general practices from all over the UK. Only two of the QOF studies were designed to take account of secular changes in smoking prevalence, resulting from new guidelines for smoking cessation, recent fiscal policy or legislation, either by comparing actual outcomes against predicted outcomes,24 or by using interrupted time series analysis.25 Other studies we identified examined the effect of fee for service or bonus-type payments on smoking cessation activities alone in the UK, Germany, Taiwan and the USA. They included two randomised controlled trials,19 ,30 one cluster randomised controlled trial,36 two serial cross-sectional studies comparing health maintenance organisations (HMO)s with and without financial incentives for smoking cessation activities,33 ,34 and three before and after designs.31 ,32 ,35 The designs of the before and after studies did not take account of secular changes in smoking prevalence. Four studies examined the effect of financial incentives on individual doctors.19 ,30 ,31 ,35 In one study it was unclear whether the incentive payment was to individual doctors or to groups of doctors in the organisation.34 The remainder looked at the effects on providing incentives to groups of healthcare professionals such as clinics or general practices. Incentives included large, bonus-type payments such as for An et al 30 in which US$5000 was provided to participating clinics for achieving 50 referrals to a stop smoking telephone advice service, then US$25 per patient after the first 50. For Roski et al 36 the incentive was US$5000–US$10 000 if ≥75% of patients at participating clinics had their smoking status recorded and if ≥65% of smokers had been given smoking cessation advice. For the studies by Chang et al 31 ,35 the bonus was paid per smoker advised and was of the order of US$24. In two of the papers the payment was paid per smoker who stopped smoking, varying from US$2432 to US$152.19 For the cross-sectional studies of HMO funding33 ,34 the amount of the incentive was not reported and direct communication from the authors confirmed they had not collected this information for the studies.

### Outcome measures examined

Most of the studies examined process measures such as recording of smoking status, smoking cessation advice and/or referral to smoking cessation services. For these measures, most of the studies showed an improvement after financial incentives were introduced. Three studies examined the impact of financial incentives on quit rates. There was too great a degree of statistical heterogeneity for the studies to be combinable for meta-analysis even when undertaken separately for QOF and non-QOF studies (I2 >90%, p<0.001 using RevMan37).

### Smoking status

For QOF studies the improvements in recording smoking status ranged from 19%22 (72.4% in 2003, 91.4% in 2005, OR 3.12, 95% CI 2.80 to 3.48), to 52%27 (44% in 2004 to 96% in 2005, OR 24.19, 95% CI 22.42 to 26.11). The improvement in the RCT by Roski et al 36 was 7.3% (54% for incentives clinics, 46.7% for control clinics, OR 1.34, 95% CI 1.17 to 1.54) and 1.7% for incentive plus registry clinics (48.4% for incentive plus registry clinics, OR 1.07, 95% CI 0.92 to 1.25).

### Smoking advice or referral

For QOF studies recorded smoking advice increased by between 12.2% (from 83.8% in 2004 to 96% in 2006, OR 4.64, 95% CI 5.23 to 5.34)27 to 16.4% (80.6% in 2003, 97.0% in 2005, OR 7.87, 95% CI 5.68 to 10.90).24 The other studies' findings on smoking advice were less consistent. Roski et al 36 and Coleman et al 32 found no difference between control and financial intervention groups. Otherwise the improvement ranged from an increase of 2.25% in Chang et al's patient database study (0.50% in 2004, 2.75% in 2005, OR 5.05, 95% CI 4.98 to 5.12)35 to 5.7% in the study by Chang et al using the Taiwan Tobacco survey (21.1% in 2004, 26.8% in 2005, adjusted OR 1.26, 95% CI 1.11 to 1.42).31

### Prescriptions for nicotine replacement therapy (NRT)/buproprion

Two studies found financial incentives were associated with an increase in the proportion of smokers receiving prescriptions. For Coleman et al 32 comparing pre/post-QOF, the OR was 6.32, (95% CI 5.85 to 6.83). For McMenamin et al 33 comparing HMOs with financial incentives to those without, the OR was 2.75 (95% CI 1.33 to 5.65).

### Quit rates

Three non-QOF studies examined quit rates and longer-term abstinence but produced mixed findings. Chang et al (2008)35 found no improvement in quit rates over the previous 6 months when funding for smoking cessation activities in Taiwan increased between 2004 in 2005 (25.2% in 2004 vs 21.3% in 2005, OR 0.96, 95% CI 0.87 to 1.06). Roski et al 36 found no difference in 7-day quit rates in their RCT (22.4% vs 19.2%, OR 1.21, 95% CI 0.98 to 1.49 for incentive clinics compared with control clinics; 21.7% vs 19.2%, OR 1.16, 95% CI 0.91 to 1.48 for registry clinics compared with control clinics). Salize et al 19 also found no difference in effect in their cluster RCT, for the financial incentive group (TI) compared with the usual care group (TAU). However, they did find an improved quit rate in the group with general practitioner (GP) training plus patient-reimbursed medications (TM) of 12.1% compared with 2.7% for TAU, OR 4.98 (95% CI 1.22 to 22.16) and also in the group with GP training, patient reimbursement plus GP incentives (TI/TM), of 14.6%, OR 6.16 (95% CI 1.44 to 26.37), the large CIs reflecting the relatively small sample size of the study. However, the TI/TM group did not significantly outperform the TM group, suggesting that the cost-free medication may have accounted for the effect observed.

### Changes in smoking prevalence

Three QOF studies looked at smoking prevalence. They were not able to examine quit rates as such as these are not specifically recorded on GP electronic medical records. Cupples et al 28 found no difference between recorded smoking prevalence between patients with coronary heart disease (CHD) in the Ireland where there was no financial incentive scheme compared with Northern Ireland (QOF). Millett et al 16 found a reduction in smoking prevalence in patients with diabetes following the introduction of QOF (from 20.0% to 16.2%, OR 0.73, 95% CI 0.69 to 0.86). Simpson et al 29 found a reduction in smoking prevalence in UK from 28.4% in 2001/2 to 22.4% in 2006/7, OR 0.73 (95% CI 0.72 to 0.73). However, it was not clear whether this reduction was due to smokers quitting through GP management or whether it can be explained by secular trends in the UK.

Chang et al (2010)31 also noted a reduction in smoking prevalence between 2004 and 2007 (23.9%, 22.2%, 21.4% and 21.1% for each year respectively) and an increase in the proportion of ex-smokers in Taiwan (6.4%, 5.7%, 7.4% and 7.4% respectively) associated with the increase in funding for smoking cessation activities, but the authors acknowledge they could not distinguish whether this was due to the funding change or more widespread information about smoking through media campaigns and hospital-based smoker identification programmes.

## Discussion

### Principal findings

We identified 18 studies examining the effects of financial incentives for healthcare providers on smoking cessation activities and outcomes. The studies scored in the mid range for quality with a validated scoring guideline, so there were no methodologically poor studies, but there were also no excellent studies. We expected the RCTs to score higher than observational studies but this was not the case. This may be because those observational studies examining system-wide pay for performance schemes scored highly for having large sample sizes and very representative samples.

Most of the studies examined process measures such as recording of smoking status, recording that smoking cessation advice had been given and/or that patients were referred to smoking cessation services. For these measures, almost all studies showed improvements when financial incentives were introduced. In the study by Coleman et al (2001)32 where the financial incentive did not result in more patients recalling receiving smoking cessation advice, the authors considered that the smoking cessation training given to practice staff before the control period may have diluted the effect of the financial incentive. For the study by Roski et al,36 the authors suggested that management directives to improve productivity and reduce costs may have affected outcomes negatively.

The two studies examining the effect of financial incentives on rates of prescribing nicotine replacement therapy (NRT)/bupropion found an increase in prescribing rates,32 ,33 but in the study by Coleman et al 32 the increase in the proportion of smokers receiving prescriptions (from 1% in 2000 to 6% in 2004) was far less than the increase in the proportion of smokers given advice (from 7% to 37%) and the authors suggest that this might be because prescribing was not incentivised. Another explanation might be that patients were prompted to attend community stop smoking clinics where these medications are available or to buy NRT over the counter.

Studies that examined quit rates had mixed results. Those examining system-level incentive schemes found a reduction in smoking prevalence, but limitations in study design meant it was not possible to determine whether this was due to smokers quitting through doctor management, or an effect of merely being asked about smoking by a clinician, or due to secular changes in smoking behaviour, or a combination of all three. QOF papers were not able to look at quitting smoking as an outcome as this is not recorded consistently by GPs, possibly because currently practices are not incentivised to do so. Also, a large proportion of smoking cessation activity takes place outside primary care, in community pharmacies and stop-smoking clinics, and information about quitters is often not relayed to GPs.

The results from Salize et al's cluster RCT19 suggested that financial incentives might influence quitting behaviour if combined with no cost NRT and/or buproprion prescriptions in addition to GP training, but a similar level of impact was seen with just the free medication group. This result is pertinent to the USA and the UK. In the USA, the Affordable Care Act38 mandates Medicaid coverage of tobacco-dependence treatments for pregnant women from October 2010. Coverage for all Medicaid enrolees will be increased by January 2014, whereby tobacco-dependence cessation drugs will no longer be excluded from covered benefits. In the UK, people can access nicotine replacement treatment, bupropion or varenicline without charge from UK National Health Service (NHS) smoking cessation services or from GPs for the price of a prescription (currently £7.20, equivalent to US US\$11.83), or without cost if exempt from prescription charges.

### Strengths and weaknesses of the review

We employed a comprehensive search strategy to identify relevant papers and included those with observational designs as well as RCTs. This meant that a larger number of papers were included in our study compared to previous systematic reviews. Petersen et al,39 looking at the effects of pay for performance schemes on improving healthcare quality, identified nine studies that looked at prevention activities, of which two were for smoking cessation. This review was conducted in 2005 before pay for performance schemes became more common, which might explain the paucity of relevant studies. In a recent systematic review of strategies to increase the delivery of smoking cessation activities in primary care settings, Papadakis et al 6 identified only three papers examining the effect of financial incentives. The authors excluded trials that were not indexed within Medline as randomised controlled trials, controlled clinical trials, or evaluation studies, whereas we did not limit our search in this way. Observational studies are inherently less robust methodologically compared with cohort or randomised controlled trials and we acknowledge this. However, as interventions for smoking cessation are necessarily complex, we felt it was important to include these studies to attempt a comprehensive view of the literature currently available.

Most studies focused on process measures (recording smoking status and advice given) rather than quit rates as outcomes, so improvements may be due to improved recording rather than increased delivery of smoking cessation interventions. Of the non-QOF studies, follow-up times for quit rates were reasonable at between 6 and 12 months, with the exception of the 7 day quit rates reported by Roski et al 36 As previously mentioned, most of the observational studies we identified did not take account of secular changes in smoking during the intervention period.

Out of the 18 included studies, 10 examined the impact of the UK's QOF. Their findings may not be generalisable to other countries as the size of the incentive is large and is supported by prompts from the electronic medical records (EMRs)40 ,41 and a comprehensive specialist smoking cessation service. The non-QOF studies identified in our review examined financial incentives that were mainly aimed at doctors. Therefore, the generalisability of most of these studies to clinicians other than doctors may be limited. However, those examining QOF would include work performed by practice nurses.

The financial incentives tested differed in amount and in this review, we were not able to identify an optimal amount. If health practitioners were offered very large amounts for each smoker who stopped then the intervention would likely to be successful but would not be financially practical. In addition, the results from Salize et al 19 suggest that subsidised smoking cessation medication may have more of an effect that a financial incentive. Only one study examined the cost effectiveness of financial incentives19 and found that GP training and remuneration per abstinent patient was not effective compared to usual treatment. However, GP training plus cost-free smoking cessation medication and a combination of GP training, free medication and remuneration were both more cost effective interventions compared with usual treatment.

Our findings fit with those from other systematic reviews42–44 that have found multicomponent interventions more effective than single-component interventions in helping primary care doctors to deliver prevention services, including smoking cessation.45 Therefore, financial incentives may have more impact when combined with other interventions, such as clinician education and subsidised smoking cessation prescriptions.

Financial incentive schemes can have unintended outcomes, such as gaming,46 adverse patient selection,47 poor performance for unrewarded activities48 and taking away doctors' internal motivation49 and the studies did not examine these. Also, unless quit rates are rewarded, such schemes may become tickbox exercises in recording process measures. For example, recording that smoking cessation advice has been given is no indication of the quality of advice given, and having someone stop smoking is obviously more valuable than simply recording smoking status.

### Suggestions for further research

Most of the studies we identified used a before and after design, only two of which took into account secular changes in smoking prevalence. Further RCTs or cohort studies would be required to determine the effectiveness of financial incentives. Further cost effectiveness studies, including calculation of price elasticity and possible optimum incentive levels, are also necessary.

## Conclusions

Financial incentives can be effective in improving the recording of smoking status, recording of smoking cessation advice and referral to smoking cessation services. Given the effectiveness of doctor advice to smokers in reducing smoking rates4 any intervention to increase such advice is important. However, only five studies evaluated smoking rates and for these there were mixed results. Overall, these results are encouraging but the area does require more comparative studies. As several areas of the UK are currently developing local versions of the QOF50 in which prevention activities such as smoking cessation are more strongly incentivised, this gives a further opportunity to examine the effectiveness of financial incentives for smoking cessation by comparing the effects in these areas with other similar areas where these activities are not additionally incentivised.

### What this paper adds

• Previous systematic reviews for prevention activities identified only a small number of studies addressing smoking cessation activities in healthcare settings. Financial incentive schemes have become more widespread recently, and this systematic review identified 18 relevant papers, including observational studies as well as randomised controlled trials.

• This systematic review found that financial incentives can improve the recording of process measures for smoking such as smoking status and the giving of smoking cessation advice by clinicians. However the effect on quit rates is more variable and further studies examining this important outcome are needed.

## Acknowledgments

The Department of Primary Care & Public Health at Imperial College is grateful for support from the NIHR Collaboration for Leadership in Applied Health Research & Care (CLAHRC) Scheme, the NIHR Biomedical Research Centre scheme, and the Imperial Centre for Patient Safety and Service Quality.

View Abstract

• ## Supplementary Data

This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Files in this Data Supplement:

## Footnotes

• Competing interests None.

• Provenance and peer review Not commissioned; externally peer reviewed.

• Data sharing statement This is a systematic review and all data is available online from the individual included papers.

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.