Senior Researcher, Department of Drug Management and Informatics, Stockholm County Council and Karolinska Institutet, Centre for Pharmacoepidemiology and Department of Laboratory Medicine, Division of Clinical Pharmacology, Karolinska University Hospital - Huddinge
A[˚]ke Pehrsson BA
Maria Juhasz-Haverinen, MSc Pharm
Pharmacist, Division of Finance and Healthcare Planning, Stockholm County Council, Sweden
Aniko Veg PhD
Senior Researcher, Department of Public Health and Caring Sciences, Uppsala University, Sweden
Maria Edlert BA
Administrator, Department of Drug Management and Informatics, Stockholm County Council
Gunilla To¨rnwall-Bergendahl BA
Henrik Almkvist MD
Head of Department
Division of Finance and Healthcare Planning, Stockholm County Council, Sweden
Brian Godman BSc
Researcher, Institute for Pharmacological Research ‘Mario Negri’, Milan, Italy
Fredrik Granath PhD
Statistician, Centre for Pharmacoepidemiology and Clinical Epidemiology Unit, Department of Medicine, Karolinska Institutet and Karolinska University Hospital, Stockholm, Sweden
Ulf Bergman MD PhD
Senior Medical Officer, Department of Drug Management and Informatics, Stockholm County Council and Karolinska Institutet, Centre for Pharmacoepidemiology and Department of Laboratory Medicine, Division of Clinical Pharmacology, Karolinska University Hospital - Huddinge
Received date: 7 January 2009 Accepted date: 13 April 2009
Background Financial incentives have been suggested to be effective in increasing the quality and efficiency of drug prescribing. Concern has been raised in relation to potential negative consequences on the quality of care. Aims To describe and analyse the impact of an incentives model linking payment with adherence to drug and therapeutics committee (DTC) guidelines and self-reflection of prescribing pattern in a ‘prescribing quality report’. Methods The study was performed in the county of Stockholm, Sweden, with 139 (out of 154) primary healthcare centres (PHCs) participating in the project and 15 PHCs not participating. The study consisted of two parts: a quantitative observational study of prescribing patterns and a qualitative analysis of the submitted prescribing quality reports. All prescriptions issued from PHCs and dispensed at pharmacies during October to December 2005 andOctober to December 2006 were analysed, using adherence to the regional DTC guidelines as the main outcome measure. Adherence was assessed using the drug utilisation 90% methodology, i.e. focusing on drugs constituting 90% of the prescribed volume and the proportion of drugs included in the guidelines. The qualitative analysis focused on reports on the quality of drug prescribing submitted by each PHC in early 2007. Results The 139 PHCs participating in the programme accounted for 85% of all prescriptions issued in primary care during October to December 2006. Mean adherence to guidelines increased among participating practices by 3.3 percentage units (95% confidence interval (CI) 2.9–3.7%) to 83% (82.6–83.7%) during the year. The adherence among practices not participating increased by 3.1 percentage units (95% CI 1.7–4.4%) to 78.8% (95% CI 76.7–80.9%). The higher adherence achieved during the year corresponded to savings estimated at five times greater than the cost of running the programme including the financial incentives. In addition, many areas for improving prescribing were identified, such as limiting the prescribing of drugs with uncertain safety profiles and documentation as well as reporting adverse drug reactions. Conclusion Although no causal effect can be attributed without a control group, we have shown the feasibility of a model linking payment to DTC adherence. This approach with its own quality assessment and goal setting offers an example to other regions and countries of how to increase the quality and efficiency of drug prescribing within limited resources.
general practice, incentives, prescrib-ing, primary health care, quality indicators.
Most countries are facing the challenge of growing healthcare demand with limited available resources. The cost of pharmaceuticals is particularly in focus since the growth in expenditure has been greater than for other healthcare components,[1-6] and will accelerate with increased prevalence of chronic diseases combined with the continued launch of new expensive medicines. Consequently, various strategies have been applied in pharmaceutical policymaking to influence the quality and efficiency of prescribing, including positive and negative financial incentives, educational interven-tions, prescribing targets and regulatory changes.[6,8-15]
Financial incentives can be used to reduce the use of healthcare resources, improve compliance with practice guidelines or achieve general health targets. Allocating drug budgets to doctors has been suggested as an effective method to influence physicians and increase the cost-effectiveness of prescribing.[6,14,6-19] Conse-quently, incentive schemes and drug budgets are applied today in many countries in Europe and also in the United States.3,6,12,18,20,21
In Sweden, initiatives to improve the quality of drug prescribing have been organised by drug and therapeutics committees (DTCs).[6,22,23] These activi-ties include decision-support systems for prescribing, educational programmes, feedback on prescribing pat-terns, and evidence-based guidelines for drug treatment. In recent years, financial incentives have also been introduced. A range of models have been used, from financial incentives linked to certain targets to capi-tation-based drug budgets.[6,24] Some regions apply population-based models, while others allocate a specific budget to each practice based on historic prescribing patterns. The former model is more common in rural areas and the latter more common in major cities.
The provision and financing of health services in Sweden is a public sector responsibility, primarily resting with 21 county councils. Primary health care is the basis of the Swedish healthcare system but it has no gatekeeper function and therefore many healthcare providers are involved in patient management and drug prescribing. In the county of Stockholm, the local primary healthcare centres (PHCs) only account for one-third of consultations and dispensed prescriptions to the population in the surrounding area.[25,26] Conse-quently, it is not feasible to link drug budgets to specific patient populations, and a model with incen-tives linked to prescribing behaviour and adherence to guidelines was introduced in Stockholm in 2006.
The new model included extra payments linked to the level of adherence to the DTC guidelines (meas-ured as the proportion of the drugs prescribed included in the guidelines) and the submission of a ‘prescribing quality report’. Templates were issued including ques-tions about the doctors’ opinion of their adherence to DTC guidelines, goals for improvement including their prescribing of new medicines, documentation and reporting of adverse drug reactions (ADRs), con-tacts with the pharmaceutical industry, participation in clinical trials and continuing professional develop-ment. The intention of the model was to use financial incentives to increase doctors’ cost-consciousness, while at the same time stimulating them to assess the quality of their drug prescribing and finding potential ways to improve it. The aim of this study is to describe the model and PHCs’ experiences with it.
The study was performed in the county of Stockholm, Sweden, which has 1.9 million inhabitants and 169 PHCs; 154 PHCs were invited to participate, with one municipality excluded due to a separate healthcare organisation. This was an observational study without a formal control group and was based on quantitative data on prescribing patterns from 2005 to 2006 and qualitative data summarised in the ‘prescribing quality reports’ for 2006. The quantitative analyses provided a description of the adherence to the guidelines before and after the incentives were introduced, and the qual-itative analysis contributed to a deeper understanding of which factors the PHCs considered important to improving the quality of drug prescribing.
The quantitative analysis was performed with routinely collected data on dispensed prescriptions in ambulat-ory care patients from all PHCs participating in the programme. Data were collected from the Swedish National Prescription Register administered by the National Corporation of Swedish Pharmacies.
The time periods for analyses were October to December 2005 and October to December 2006. These periods were chosen to reflect prescribing before and after the schemes were introduced. Using dispensing data from the last quarter of each year would minimise the problem with older repeat prescriptions issued before the project started (a prescription is valid for one year in Sweden).
Data were classified according to the ATC (Anatomic Therapeutic Chemical) classification. Drug utilis-ation was expressed as defined daily doses (DDDs), prescription items and expenditures in Euros. Adher-ence to DTC guidelines was assessed using the DU90% (drug utilisation 90%) method which assesses the num-ber of drugs constituting 90% of the prescribed volume expressed in DDDs and adherence to guidelines within this segment (see Figure 1).[28-30]
Figure 1: Example of a drug utilisation 90% (DU90%) prescribing profile for a PHC centre based on drugs dispensed at all pharmacies in the country October to December 2006. DU90% = drug utilisation 90% – the number of drugs constituting 90% of the volume expressed in DDDs. Adherence is calculated as the percentage of DDDs for drugs in the regional DTC guideline compared with the total number of DDDs within the 90% segment. Rx = number of prescription items, cost is presented in Swedish Crowns (SEK), 100 SEK = 10.5 Euro (March 2009).
The guidelines for comparison were the list of drugs recommended by the DTC in Stockholm in 2006, the so-called ‘Wise Drug List’.[6,22,23,31] These guidelines are produced by over 20 expert groups, which include general practitioners (GPs), hospital specialists, phar-macists and clinical pharmacologists. They consist of diagnosis-specific evidence-based recommendations with some 200 to 240 pharmaceutical products sug-gested as first-line choices for outpatient treatment of common diseases.[6,22,23] In this study, adherence was calculated by substance regardless of which pharma-ceutical product (brand or generic) was prescribed and dispensed to the patient.
Descriptive statistical values (mean, median, 95% confidence interval (CI) and range) were calculated. Crude comparisons of adherence to the DTC in 2005 with respect to activities reported in the prescribing quality reports were performed by t tests. The same comparisons were also performed using a multiple linear regression model. Crude and mutually adjusted differences are presented together with 95% CIs. The association between reported activities and change in adherence between 2005 and 2006 was assessed using analysis of covariance, where the 2005 value was included as a covariate in order to allow for ceiling effects. A corresponding analysis of covariance was performed to assess the change in adherence between units with and without incentives. Differences were considered statistically significant for P < 0.05.
Figure 2: Baseline adherence to the DTC recommendations using DU90% in October to December 2005 and change between October to December 2005 and October to December 2006. Primary healthcare centres in the county of Stockholm were invited to participate in the schemes (n = 154). Dark diamonds represent the 139 practices participating in the schemes, white circles show the 15 non-participating PHCs serving as ‘controls’.
Correlation between adherence and cost/DDD (see Figure 3) was calculated using Pearson’s correlation coefficient (r). A value for P < 0.05 was considered significant. Potential savings relating to increased adherence to the guidelines were calculated by multiplying the proportional decrease in cost/DDD for each percentage increase in adherence with the total number of DDDs prescribed.]
Each PHC received a questionnaire by email. The questionnaire was developed through a consensus procedure by the regional division of finance and healthcare planning at the beginning of 2007 (see Box 1). Most questions were closed (yes/no), to facil-itate analyses. The reports were submitted by the head physician in each PHC to enhance the robustness of the answers, and embodied reflections on one year’s prescribing patterns at the local PHC. Two open-ended questions (Q2 and Q8, Box 1) were analysed qualitatively, since they were strongly related to the aim of the reports and the answers were sufficiently long to build a short text.
A thematic analysis of the contents of the submitted quality reports was made by two of the authors (AV and ME). The first step of the analysis was a thorough reading of each short text. The second step involved formulating the initial categories in order to start collating the replies. Additional categories were de-rived with ongoing analysis of the text. The derived final list of categories constituted both factors pro-moting and factors explaining difficulties in reaching good adherence to recommendations (see Table 3). These categories were subsequently discussed with the other members of the research team before embarking on the analysis to enhance the robustness.
A total of 139 out of 154 invited PHCs (90%) agreed to participate in the study. The main reasons for not participating were that PHCs missed the deadline for inclusion, or concerns about additional workload. The 15 non-participating PHCs subsequently served as the ‘controls’. The first quality reports were sub-mitted in early 2007 based on 2006 data. In 2006, a total of e2 million was spent on incentives to the 139 participating PHC centres, with payments per practice varying depending on their performance. As an ex-ample, a PHC with seven GPs and an adherence to the DTC guidelines of 87% received e18 000; average adherence was 82% in October to December 2006.
During October to December 2006, after the incen-tives were introduced, 4.4 million prescription items were dispensed to the inhabitants of Stockholm County. This represents an average of 2.4 prescription items per inhabitant. Forty-three percent of all pre-scriptions had been issued in primary health care. The 139 participating PHCs accounted for 85% of all pre-scriptions issued by all 169 PHC centres in the county. The total expenditure for prescribing in primary health care was e35 million, constituting 26% of the total ambulatory care prescribing in the county. The num-ber of prescriptions increased by 10% and the expen-diture by 4% compared to October to December 2005.
Adherence to the DTC recommendations was on average 83.1% (95% CI 82.6-83.7%) in October to December 2006, and varied between 62% and 90% among participating practices. The practices not in-cluded were smaller (8000 versus 11 500 dispensed prescription items/practice/quarter). They also had a significantly lower adherence to guidelines than those participating in the scheme (78.8% (95% CI 76.7- 80.9%)).
Adherence to DTC guidelines increased by 3.3 percentage units (95% CI 2.9-3.7%) among practices participating compared to 3.1 percentage units (CI 1.7-4.4%) in those not participating. A significantly lower increase among non-participating centres was observed after adjustment for the ceiling effect, i.e. that the participating practices had a significantly higher adherence prior to introduction of the schemes (see Figure 2).
After the incentives were introduced, a clear corre-lation was observed between high adherence and low cost/DDD (see Figure 3). An increased adherence of 1% corresponded to e0.47 lower cost/prescription item. For a PHC of average size (six GPs), this corresponded to an approximately e21 000 lower annual drug ex-penditure. Consequently, with a total of 6.4 million prescription items dispensed in 2006 at participating PHC centres, increasing adherence by 3 percentage units resulted in estimated annual savings of more than e10 million.
More than half of all PHCs (58%) claimed that they had participated in previous projects with the aim of improving prescribing quality and/or increasing the cost-effectiveness of prescribing. A majority of partici-pating and non-participating practices, 84% and 83% respectively, received support from information doc-tors and/or pharmacists (medical doctors or pharma-cists with special training employed or financed by the Drug and Therapeutics Committee to disseminate guidelines and educate healthcare professionals in rational pharmacotherapy) or had guideline drugs highlighted in the electronic prescribing support sys-tem to enhance the quality and efficiency of prescrib-ing. One-quarter of the PHCs (26%) had participated in clinical trials and 22% had doctors who were mem-bers of the DTC or one of the expert groups.
The analysis of the prescribing quality reports em-phasised the need for improving documentation and reporting of ADRs. Fifty-two percent of the PHCs had local processes for documenting ADRs, and 85% claimed that they regularly discussed cases of ADRs at their internal meetings. However, many PHCs also suspected a substantial under-reporting of ADRs. In the prescribing quality reports, these PHCs stated that they had submitted a total of 300 ADRs to the regional ADR monitoring unit in 2006. This corresponded to half of all submitted ADR reports (n = 585) from all PHCs in the region.
There were certain differences in adherence to the DTC recommendations between participating PHCs in October to December 2005, before the programme started, with a significantly higher adherence at base-line for PHCs previously participating in projects or with doctors that were members of the DTC or one of the expert groups (see Table 1).
However, the change in adherence during the year showed the opposite pattern, with greater increases in adherence observed among PHCs having the lowest baseline adherence rates initially (see Table 2). These differences disappeared when adjusted for the ceiling effect.
A total of 137 prescribing quality reports were sub-mitted. The result of the analysis of question Q2, ‘Describe three observations acquired when analysing your prescribing patterns’, showed most PHCs were satisfied with their own improvements in drug pre-scribing, and they considered themselves to have good adherence to the DTC guidelines. A common conclusion was that the most frequently prescribed drugs were recommended in the guidelines. However, there were also observations of high prescribing of certain drugs that were not recommended, which is now being addressed, such as ‘reducing unnecessary prescribing of antibiotics’. Some explanatory factors behind high or low adherence to the guidelines are presented in Table 3.
The PHCs identified therapeutic areas or single drugs where a substantial improvement in adherence would be possible, for example: ‘reducing the pre-scribing of angiotensin receptor blockers in favour of ACE [angiotensin-converting enzyme]-inhibitors’ or ‘increasing the prescribing of start-packages when possible’. The suggested strategies were in accordance with the guidelines; these included increasing pre-scribing of recommended drugs when initiating drug therapy with new patients, or reserving certain drugs as second-line choice for more restricted indications. Furthermore, many PHCs wanted to increase their knowledge of pharmacotherapy through educational activities, perform regular reviews of their prescribing patterns, and in general: ‘increase knowledge about drugs and their adverse effects’.
This study has demonstrated the feasibility of using financial incentives to stimulate the quality of pre-scribing through increased adherence to evidence-based drug therapies recommended by the expert groups in the DTCs and through self-assessment of ways to improve future prescribing. In the first year of the programme, adherence to DTC guidelines increased on average by 3 percentage units from 80% to 83%, a substantially higher increase than the 0-2% achieved in previous years. Although 3% does not seem to be a high figure, it is more than generally achieved through educational interventions.[11,13] In addition, the study started from a high average adherence rate of 80%. It is likely that the programme supported this increase, although this cannot be concluded with this observa-tional study since it does not correct for other factors influencing prescribing patterns, such as pharmaceutical company marketing activities for new and existing drugs, new indications for existing drugs and changes in regulatory policies.[32-34] It is interesting that adher-ence rates increased by a similar extent among PHCs not participating in the programme. However, these PHCs were not completely comparable. As discussed, they were smaller and had a lower adherence to the guidelines initially with, consequently, greater room for improvement. Furthermore, they may have been contaminated by the intervention since they participated in the same professional networks and educational activities.
It is a challenge to change professional behaviour. Simple diffusion or dissemination of printed material and mailed feedback on prescribing patterns may influence professionals’ awareness and knowledge, but they seldom change behaviour.[11,35-37] More intensive strategies such as ‘academic detailing’ as well as guideline development coupled with comprehensive dissemination strategies do change behaviour, although few studies have reported if the benefits/savings achieved using these strategies outweigh the costs of performing them.[11,15,35,38,39] Although it cannot be definitely con-cluded that the increased adherence was solely due to the incentives programme, the calculated savings were five times higher than the cost of running the pro-gramme. Alongside this, physicians identified a num-ber of activities to help improve future prescribing quality. Consequently, our model suggests a potential way to increase the quality and efficiency of drug prescribing with relatively limited resources. The calcu-lated savings were based on the observed correlation between adherence and cost/DDD. This negative corre-lation between adherence and cost has arisen in recent years and is explained by ongoing reforms in Sweden to achieve low prices for generic medicines coupled with programmes to encourage generic prescribing as first line.[6,23] In most cases generic drugs are equally effective and less expensive than the corresponding patented drugs.[6,40-43] Recently, the National Audit Office in the UK also documented that considerable savings (£227million) could be achieved by increasing prescribing of generic simvastatin, omeprazole and ACE inhibitors compared with branded drugs in the same or related classes.[6,34]
Overall, financial incentives have been shown in previous studies to be effective in improving the quality and efficiency of prescribing.[18,19,21,44,45] However, the effects can be short lived,[16,17] and the long-term effects on the quality of care are less well studied. In a systematic review of studies published between 1993 and 1999, it was suggested that financial incentives for drug pre-scribing could decrease the quality of care by limiting continuity, reducing the preventive services offered and increasing inappropriate use of emergency services. Concern was also raised about the potential change in the doctor-patient relationship, and the potential negative consequences in the long term of reducing time for teaching and research. Consequently, such initiat-ives must be carefully planned and monitored. We consider the risk of negative consequences to be low with our approach, since the incentives were not linked to specific medical decisions but used to stimulate overall adherence to guidelines that had been developed with robust technologies. In addition, the DU90% method used to monitor adherence offers advantages since it focuses on those medicines accounting for 90% of the volume, thereby leaving some latitude to deviate from guidance if needed. In addition, we gave physicians the opportunity to reflect on ways to im-prove their own prescribing. We believe that measuring adherence with routinely collected data is preferable since it is known that self-reported data produce an overestimate of adherence to guidelines.[39,47] Further-more, too strong a focus on auditing and payment for performance may also pose a threat to the validity of the data recorded if physicians take the opportunity to manipulate the data to increase their incomes.[39,48,49]
As previously stated, the quality reports were an addition to the prescribing incentive scheme, enhanc-ing the opportunity for physicians to increase both the quality and efficiency of their prescribing. They also offered the potential for local ownership and learning, which contrasts with the top-down approaches in most incentive schemes. The importance of local ownership is in line with the recent experience from the UK and the Netherlands.[13,21,45,50] Perhaps not surprisingly, future recommendations included reducing prescribing of drugs that were not recommended, as well as generally increasing knowledge about the effectiveness and safety of the drugs prescribed to improve future decision making. Future analysis will reveal whether PHC recom-mendations have been implemented. This will also reveal whether PHCs subsequently increase ADR documen-tation and reporting, compared to the under-reporting that was identified.
Finally, it is important to emphasise that adherence to prescribing guidelines and quality reports were only two aspects of quality of care. Quality assessment should include all aspects of care, including structure, process and outcome. Nevertheless, drug prescribing is im-portant and our findings should stimulate debate in other countries on future methods to enhance the quality and efficiency of prescribing. The model could also easily be adapted to include other aspects of patient management and outcome.
Not commissioned; externally peer reviewed.
HenrikAlmkvist,MariaJuhasz-Haverinen,AkePehrsson and Gunilla To¨rnwall-Bergendahl were employed by the Department of Finance and Healthcare Planning in Stockholm county responsible for running the pro-gramme. Ulf Bergman and Bjo¨rn Wettermark were both members of the regional Drug and Therapeutics Committee, responsible for development of the DTC guideline.