Smeets HM, de Wit NJ, Zuithoff NPA, van Dijk PCM, van der Lee APM, Hoes AW. A Health Insurance Company–Initiated Multifaceted Intervention for Optimizing Acid-Suppressing Drug Prescriptions in Primary CareA Randomized Controlled Trial. Arch Intern Med. 2010;170(14):1264-1268. doi:10.1001/archinternmed.2010.224
Copyright 2010 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2010
To evaluate the effectiveness of a health insurance company–initiated intervention strategy aimed at optimizing acid-suppressing drug (ASD) prescriptions in primary care.
In a cluster randomized controlled trial design, 112 primary care physician (PCP) peer review groups (993 PCPs) in the central region of the Netherlands were randomized. The PCPs in the intervention group received an ASD prescription optimization protocol, a list of their patients taking ASDs frequently on a long-term basis, and financial compensation for additional consultations with these patients. The PCPs in the control group did not receive any of these interventions. Prescription data on 23 433 patients were extracted from the database of the regional health insurance company. The main outcome measures were the proportion of patients who reduced ASD consumption by more than 50% and changes in annual volume and costs of ASD prescriptions. Differences in ASD reduction and in volume were analyzed applying multilevel regression analyses.
At baseline, 2.4% of the patients (n = 967 506) of the participating practices used ASDs frequently on a long-term basis (>180 daily defined doses [DDDs] annually). During the 6-month intervention, 14.1% of the patients in the intervention group reduced ASD consumption compared with 13.7% in the control group (adjusted relative risk, 1.04; 95% confidence interval [CI], 0.97-1.11). Changes in intervention and control group in mean volume of ASD prescription per patient were similar (β = 0.33 for DDD; 95% CI −3.00 to 3.60).
A health insurance company–initiated multifaceted intervention, including practical tools and financial incentives, did not alter ASD prescription practice in primary care. More tailored interventions, including patient-targeted initiatives, are required to optimize ASD prescription.
Acid-suppressing drugs (ASDs) are among the most frequently prescribed drugs; up to 10% of the population uses H2 antagonists or proton pump inhibitors (PPIs) on a regular basis.1- 3 In most European countries, the use of ASDs, especially that of PPIs, is growing annually.4,5 Generally, ASDs offer a prompt and effective reduction of acid-related symptoms.
More than one-third of all patients taking ASDs use more than 180 daily defined doses (DDDs) annually, suggesting frequent and long-term drug use, yet there are only few indications for continuous ASD prescribing. Moreover, reports have shown an increased risk of both pulmonary and gastrointestinal infections as well as osteoporosis in patients using PPIs for long periods, but these reports did not change PPI prescription habits.6- 9 Most current guidelines recommend continuous acid suppression only for a small subgroup of patients: in cases of esophagitis grade C and D or in cases where gastric protection is required for patients using nonsteroidal anti-inflammatory drugs (NSAIDs).10,11 Most patients with symptoms of gastroesophageal reflux disease (GERD) or dyspepsia will experience symptom control on a so-called on-demand or intermittent ASD regimen of 1 to 2 weeks. In clinical practice, it is difficult to limit patients in their ASD consumption. Drug dependency is maintained by both the (perceived) effect on acid-related symptoms and by the prompt acid rebound effect that follows cessation of drug use.12- 17
In most Western countries, about 10% of the national pharmaceutical budget is spent on ASDs, and 30% to 40% of the patients use ASDs on a daily basis.1 This largely exceeds the number of patients for whom long-term ASD treatment is actually indicated, according to clinical guidelines. Cost-effectiveness of ASD prescription could be substantially improved by limiting long-term prescription of ASDs in clinical practice and changing continuous to on-demand regimens. Primary care should be the main focus of efforts to change prescribing behavior because most ASDs are prescribed by primary care physicians (PCPs). To date, compliance with clinical practice guidelines for dyspepsia proves difficult to achieve without supporting interventions.18,19
Many studies have demonstrated that an active, multifaceted strategy may be effective in implementing clinical guidelines. To execute such a strategy, tailored interventions must be applied to each practice setting to break down the barriers that PCPs experience during implementation.20,21 To improve compliance with guidelines, the strategy should be integrated into routine practice; complex changes should be avoided; and practical support should be offered.22- 26 Providing financial incentives is also suggested to improve compliance with clinical guidelines, but to our knowledge, their effectiveness in optimizing ASD prescription has not been assessed in prospective research.27- 29
Agis, a major health insurance company in the Netherlands, initiated a multifaceted intervention program aimed at optimizing ASD prescription in primary care.30- 32 We report the effectiveness of this managed care program, which included financial incentives, on volume and costs of ASD prescription.
A cluster randomized controlled trial was conducted in the central region of the Netherlands. In this area, Agis Health Insurance Company insures more than 1.5 million patients. All PCPs in the region who treat more than 250 patients insured by Agis (N = 993) were selected to participate in the program and were randomly allocated to either the intervention or the control group. Owing to the presumed interaction between cooperating PCPs, who regularly discuss guidelines and prescription policy in pharmacotherapeutic consultation groups, we chose to randomize at the PCP peer group level.
In the intervention group, PCPs received an ASD prescription optimization protocol by mail. This package included (1) instructions for rational ASD prescription according to current professional guidelines; (2) suggested strategies for (gradual) withdrawal of ASD treatment; (3) an updated list, per practice, of the patients currently using ASDs on a long-term basis; and (4) patient information letters about rational ASD use in different languages. Further implementation of the ASD prescription optimization protocol was left to the PCP. As compensation for their efforts to change ASD prescribing behavior, PCPs were granted 3 extra consultations (for a total of €75.00) for each patient included in the protocol. Every 3 months during the intervention period, PCPs received an updated list of the patients using ASDs on a long-term basis and a list of those who had reduced ASD use.
In the control group, care was provided as usual. The PCPs did not receive the ASD prescription optimization protocol, nor were they informed about the long-term ASD users in their practice. They did not receive financial reimbursement for efforts to reduce long-term ASD use by their patients.
The Agis Health Database contains computerized prescription data of all enlisted patients. It is updated every month and includes names and demographic characteristics of the patients, Anatomical Therapeutic Chemical codes, doses, amounts and prices of drugs, prescribing PCPs, and delivering pharmacists. Long-term ASD use was defined as use of more than 180 DDDs of ASDs (H2 receptor antagonists or PPIs) annually. Patients who also took more than 30 DDDs of NSAIDs were excluded because of the possible need for gastrointestinal protection.10
The volume of ASDs was defined as the mean number of DDDs per patient. Subsequently, the mean ASD cost per patient was calculated. Costs were fixed on the drug price level at the start of the program. Baseline ASD prescription during the 6 months before the intervention was categorized according to the average number of DDDs per prescription per PCP: less than 90, 90 to 180, or more than 180 DDDs. Patients were categorized as Moroccan, Turkish, or “other” ethnic background. Patients were further categorized as employed, work-disabled, unemployed, or receiving social insurance. Urbanization of the practice population was categorized as urban (high), suburban (median), and rural (low).
The primary outcome parameter was the proportion of patients who reduced ASD use by at least 50% during the first 6 months of the intervention compared with the average number of DDDs that the patients used in the 6 months before the intervention. We pragmatically chose 50% reduction as the threshold because such a change was considered clinically relevant. In addition, ASDs are generally prescribed for 90 days, and reduction or cessation of ASD use during the 6-month intervention period would likely have at least a 50% effect on prescription volume. Secondary outcome parameters were the change in mean number of DDDs per patient and mean cost of ASD medication per patient during the first 6 months of the intervention.
We used χ2 tests to compare baseline characteristics of PCPs and patients in the intervention and control groups. To assess the effects of the protocol on ASD reduction, a multilevel Poisson regression model was used, reporting the relative risk (RR) for the percentage of responders. We measured the effect of the intervention on DDDs and costs in a multilevel linear regression model, reporting the regression coefficients of the differences in changes between the 2 groups. In both multilevel models, data were analyzed at PCP practice and at PCP peer group level. The full multilevel models of both analyses were performed with DDDs per prescription at baseline as a random effects variable together with fixed effects variables such as sex, age, ethnicity, insurance group, and urbanization level of the patients, as well as sex, age, and the size of the practice population of the PCPs. The RRs and regression coefficients (βs) are reported with 95% confidence intervals (CIs).
The intervention group consisted of 61 peer groups with 559 PCPs, the control group of 51 peer groups with 434 PCPs. Mean ages of the PCPs and the prescription rate at baseline did not differ between the 2 groups (Table 1).
Of the 967 506 enlisted patients in the participating practices, 2.4% used ASDs on a long-term basis. Long-term ASD users in the intervention group (n = 12 841) did not differ in sex (59% female), age, or income status from those in the control group (n = 10 592) (Table 2). The mean number of long-term ASD users per 1000 patients was 24.2 in both groups, who were prescribed, on average, 206 DDDs of ASD in the 6 months prior to the start of the trial. In both groups, 8% of the patients were lost to follow-up after implementation, mainly owing to their switching to another insurance company.
In the first 6 months after the implementation of the ASD prescription optimization protocol, 1812 of the long-term ASD patients in the intervention group (14.1%) reduced their ASD use by at least 50%, compared with 1456 in the control group (13.7%) (RR, 1.02; 95% CI, 0.95-1.09) (Table 3). After adjustment for PCP and peer group clustering and case-mix variables in the full model, the RR was 1.04 (95% CI, 0.97-1.11) (Table 3).
In the intervention group, the mean prescription volume per patient rose from 206.1 DDDs in the 6 months before to 210.2 DDDs in the 6 months after the introduction of the intervention. In the control groups, the mean volume rose from 206.8 to 210.9 DDDs per patient. There was no difference between the changes in the 2 groups in DDDs per patient (Table 3). This finding did not change after correction for prescription rates at baseline or for PCP and peer group clustering in the full model (β = 0.33; 95% CI, −3.00 to 3.60) (Table 3). The mean cost of ASD prescription per patient in the intervention group rose from €164.10 in the 6 months before the intervention to €165.80 in the 6 months after the intervention. In the control group, the mean cost for ASD per patient rose from €161.10 to €163.20. The difference in the changes in cost between the 2 groups was €0.40 (95% CI −€3.41 to €4.21) (Table 3). Multilevel analyses showed similar findings: β = 1.63 (95% CI, −1.3 to 4.5) (Table 3). Including the costs of the financial incentive (€12.00 per patient), the additional costs of the intervention were estimated at €15.00 per patient per year.
The present health insurance company–initiated multifaceted intervention for optimizing ASD prescriptions in primary care did not reduce ASD prescription rates: by the end of the study, both the number of long-term ASD patients and the volume and cost of ASDs were similar in the intervention and control groups. These results confirm earlier reports that simply introducing an ASD prescription optimization protocol and giving feedback information to physicians is ineffective in influencing routine prescription practice.33
Hurenkamp et al13 and Krol et al34 demonstrated that after intensive individual support of both patients and PCPs, 10% to 15% of the patients stopped using ASDs. In an earlier systematic review, our research group35 found that most studies on the effectiveness of ASD reduction programs did not demonstrate an effect of the interventions on ASD consumption. In contrast, some studies reported an increase in ASD use after implementation in PCP practice by experts.33,35 In the design of the present trial, we assumed that simply distributing evidence-based prescription guidelines to PCPs would be an ineffective implementation strategy,20 but by combining guideline distribution with practical aids and financial incentives, we expected improved PCP compliance with prescription guidelines and a greater reduction in ASD prescriptions.28
A number of factors might explain the lack of effect observed in our trial. The duration of the intervention might have been too short. The time required to switch from a continuous to an on-demand or intermittent ASD prescription regimen has not been studied adequately, but a prolonged follow-up might result in a more positive effect.30,33 In addition, some PCPs might have considered the initiative as mainly an economically driven program of the health insurance company to alter their prescribing behavior. Also, our sending the PCPS the protocol by mail might have been ineffective motivation for them to participate fully and could have resulted, at least in some cases, in a rather passive contribution. Finally, the focus of the intervention was mainly on the PCPs, and we might have underestimated the role of the patient in the program. Given the drug dependency of patients taking ASDs, PCPs might have had problems convincing the patients to change their drug regimen.
The observed reductions in ASD volume in both experimental and control groups were not accompanied by a reduction in costs, likely because volume reduction was based on either lower doses or shorter prescription periods and not on fewer prescriptions. Because pharmacist delivery fees form a considerable and integral part of the drug costs, the number of prescriptions must be reduced if costs are to be lowered.
To our knowledge, this is the first study in which a health insurance company–initiated intervention program was evaluated using a randomized controlled trial design.35 Given the validity of the design, the conclusions are robust, and the generalizability of the findings seems warranted because all PCPs in a large region in the Netherlands participated in the implementation of the program. In addition, owing to the economic function of the Agis Health Database and its systematic collection process, its prescription data are likely to be complete and accurate.
The lack of effect observed herein may have resulted from an unexpectedly large number of patients in the control group who reduced their use of ASDs. At the time of our study, a public discussion about the status of ASDs as a lifestyle drug was ongoing and led to government-initiated cuts in ASD prices and changes in ASD reimbursement policy. As a result, the PCPs in the control group might have been more aware of the societal impact of ASD prescription, resulting in a drop in ASD prescriptions for some patients. However, because overall ASD prescription rates in the control group increased during the study period, this effect is likely to be small.
In addition part of the changes we observed may have been due to naturally occurring developments in prescription and drug consumption patterns in clinical practice such as temporary cessation, noncompliance, or nonrefill. However, these “regression to the mean” phenomena would have occurred equally in both groups.
The decision of which patients to include in the study was made by the PCPs. At the start of the program all long-term ASD users in the participating practices (with the exception of NSAID users) were considered eligible for the intervention. We anticipated that PCPs would exclude more than 50% of the long-term ASD users because they considered it too difficult for the patients to change drug regimens. The low number of responders may be an indication that PCPs selected only the small proportion of patients whom they presumed to be able to reduce ASD use without much additional pressure. Instead of leaving the selection process to the PCPs, a direct approach of eligible patients by the research team might have resulted in a beneficial effect of the intervention strategy.
In a pragmatic study like this one, in which the intervention is closely related to daily practice, there is no detailed insight gained into the motivation of the PCPs or into the responses of the patients. Though all PCPs were adequately informed beforehand, the protocol was sent to them without asking for their cooperation individually. Some PCPs may have disregarded participation immediately, while others actively participated. This varied participation rate obviously affected the number of ASD patients included in the study and the observed effect of the intervention.
This health insurance company–initiated managed care program to optimize ASD prescription did not result in a reduction of ASD prescription in primary care. Although this finding might be attributable to the fact that PCPs do not want health insurance companies to interfere with their professional prescription policies, we believe that this managed care program did not adequately address the problems that PCPs face in changing prescribing patterns of an effective drug for patients who are satisfied with their drug regimen. More successful might be a multifaceted intervention that includes not only practical attributes and financial incentives but also interventions tailored to the patients as well as the problem under study.21- 23 In the case of ASD prescription, it is likely that both the prescribing physician and the patient need to be convinced of the benefits of reduction of antacid drug use. More research is required to find out how this can be achieved.
Correspondence: Hugo M. Smeets, PhD, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, PO Box 85500, Utrecht, 3508 GA, the Netherlands (firstname.lastname@example.org).
Accepted for Publication: January 7, 2010.
Author Contributions:Study concept and design: Smeets, de Wit, van Dijk, van der Lee, and Hoes. Acquisition of data: Smeets, de Wit, and van der Lee. Analysis and interpretation of data: Smeets, Zuithoff, and Hoes. Drafting of the manuscript: Smeets, de Wit, and van der Lee. Critical revision of the manuscript for important intellectual content: de Wit, Zuithoff, van Dijk, van der Lee, and Hoes. Statistical analysis: Smeets, Zuithoff, van der Lee, and Hoes. Obtained funding: de Wit, van Dijk, van der Lee, and Hoes. Administrative, technical, and material support: van der Lee. Study supervision: de Wit and Hoes.
Financial Disclosure: During the implementation and evaluation of the program, Drs Smeets and van Dijk and Mr van der Lee were employed by Agis Health Insurance Company. Program costs were part of regular costs of Agis Health Insurance Company.
Additional Contributions: Ignace Konig, MSc, selected the eligible patients from the Agis database and prepared the feedback information to the PCPs.