Effectiveness of medicines review with web-based pharmaceutical treatment algorithms in reducing potentially inappropriate prescribing in older people in primary care: a cluster randomized trial (OPTI-SCRIPT study protocol)

Older people are among the biggest consumers of healthcare services, particularly drug therapy [1]. They tend to have multiple conditions, requiring multiple drug treatments [2]. They also experience age-related changes in physiology and body composition that influence the body’s ability to process medications efficiently, in terms of both pharmacokinetics (the body’s ability to absorb, distribute, metabolize and excrete a drug) and pharmacodynamics (the drug’s physiological effects) [3, 4]. Thus, prescribing for older people is a complex and challenging task, with the potential for adverse outcomes including drug-drug interactions, adverse drug reactions and potentially inappropriate prescribing (PIP) [5].

The term ‘potentially inappropriate prescribing’ covers a number of suboptimal prescribing practices, particularly the use of medicines that introduce a greater risk of adverse drug-related events where a safer and equally effective alternative is available to treat the same condition [6]. Inappropriate prescribing in older people can result in increased morbidity, adverse drug events, hospitalizations and mortality [7‐9]. Potentially inappropriate prescribing may be measured with explicit (criterion-based) or implicit (judgment-based) tools [9]. A recently developed explicit process measure, the Screening Tool of Older People’s Prescriptions (STOPP), has been published for use in European settings [10].

When these criteria were applied to an Irish pharmacy claims database (containing the prescription records of 97% of those aged ≥70 nationally), it was found that 36% of those aged ≥70 years received at least one potentially inappropriate medication. Total PIP expenditure was estimated at over €45 million (or 9% of expenditure on pharmaceuticals in that age group) [11]. The clinical and economic burden of PIP is an important public health concern and it is important to minimize PIP where possible, to increase patient safety and encourage cost-effective prescribing.

Changing professional practice is a complex and difficult task. While the usefulness of behaviour change theory in intervention and implementation research has been questioned, [12, 13] there is general consensus that interventions are more likely to have an impact when they target all stages of behaviour change [14]. Studies to date have yielded mixed results in terms of which intervention types are most effective in improving prescribing and there is no one interventional strategy that has proved to be most effective [15]. Strategies shown to be effective for improving prescribing outcomes include educational outreach visits (academic detailing) [16] and interventions involving a pharmacist [17]. Pharmacist services, such as conducting medicines reviews or providing advice to general practitioners (GPs) may lead to improvements in prescribing outcomes, including more appropriate prescribing in older people [9, 17‐21]. Patient-mediated interventions do not consistently show effects in improving prescribing [22]; however, evidence suggests that providing patients with information is important. Patient information leaflets may be helpful in improving patient outcomes, and older people appreciate receiving brief, clearly written information leaflets [23, 24]. A number of commentators have argued that a multifaceted intervention, an approach that combines a number of techniques within a single intervention [20], may be more likely to work than any one single intervention [9, 25, 26].

The aim of this study is to determine the effectiveness and acceptability of a complex, multifaceted intervention in reducing the level of PIP in primary care. The intervention combines academic detailing, medicines review with web-based pharmaceutical treatment algorithms that provide recommended alternative treatment options, and tailored patient information leaflets. The intervention development was informed by the Medical Research Council (MRC) guidelines for the development and evaluation of randomized controlled trials (RCTs) [27, 28]. It was piloted with a group of five GPs and found to be feasible and acceptable within this group.

Secondary objectives are to evaluate the effect of the intervention on patient outcomes in terms of the number of GP visits, the number of hospitalizations, patient well-being, beliefs about medicines and health status. The views and experiences of participants concerning the possible reasons for the intervention to be effective or ineffective will be explored using qualitative methodology. In addition, an economic evaluation will be carried out.

The OPTI-SCRIPT study is a pragmatic two-arm cluster RCT, incorporating qualitative analysis. Qualitative approaches can contribute in several ways to the development and evaluation of health interventions; they have been used during intervention development and piloting and will be used to undertake a process evaluation of the trial [29]. The study design was developed in line with the Consolidated Standards of Reporting Trials (CONSORT) statement extension to cluster RCTs [30]. A cluster design was chosen to avoid the possibility of contamination across arms. A GP who is treating both intervention and control patients might find it difficult to behave differently towards each group [31]. The trial will be conducted in primary care with GP practices as the unit of randomization. A practice can be a single-handed GP or can comprise two or more GPs (group practices); however, not all GPs in group practices may wish to participate. Participation may be decided by one GP on behalf of the group practice, in which case that GP will be the main point of contact for the study. We will keep a record of participating GPs. Participation in the intervention arm will be defined as attendance at the academic detailing visit and undertaking medicines reviews, while participation in the control group as providing patient prescription data. Practices that do not meet these criteria will be considered lost to follow-up (see Figure 1). The intervention will be delivered at the GP level while the unit of analysis will be at the individual patient level, in terms of medicines prescribed, adjusting for clustering.

Practices are eligible to participate if they have approximately 80 or more older patients (aged ≥70 years) on their patient panel (based on the need to recruit 10 patients per practice, and allowing for refusals) and they are based in the greater Dublin area (to facilitate the academic detailing process).

Practices are excluded where they have been involved in the development and piloting of the intervention or other concurrent medication quality related studies.

Patients will be considered eligible if they are aged 70 years or over and they are being prescribed one or more selected potentially inappropriate prescriptions on a repeat basis (only patients with an existing PIP will be included, as the intervention is specifically targeting the management of PIP).

Patients will be excluded if they have significant mental or physical illness that is likely to impair their ability to participate in the study, or they are unable to attend the GP surgery for consultation (for example, they are nursing home residents) or they are participating in another medication quality related study.

Potentially eligible practices (based in the greater Dublin area) will be identified from the Health Research Board (HRB) Primary Care Research Centre research network. All eligible practices will be invited to participate by an email (or letter where email address is unavailable), which will include a study information leaflet outlining steps of the intervention and availability of continuing medical education (CME) points for participation. When a practice agrees to participate, a member of staff (for example, a GP, practice manager or nurse) will be selected by the practice. The research team will instruct the designated person on how to identify a random sample of 50 patients aged 70 years and over from the patients of participating GPs within the practice. They will pseudo-anonymize the records by assigning the patients a study ID and send a copy of the pseudo-anonymized prescription records to the research team, where a research pharmacist will generate a list of potentially eligible patients, that is, patients with PIP. A maximum of ten patients per practice is required; if more than ten patients are identified, ten patients will be selected at random from the list of eligible patients. Eligible patients will be sent a letter of invitation, a patient information leaflet and a consent form, asking them to participate and answer a questionnaire. Prior to practice allocation, baseline patient data (including prescription data, process-of-care measures and patient-reported outcomes – see sections on outcomes and data collection) will be collected. Practices will then be assigned to intervention or control using minimization. This approach offers the advantage of ensuring balance between the groups [32] in terms of prognostic factors: in this case, practice size (number of whole-time-equivalent GPs) and practice location (urban or rural, where an urban area is defined as a relatively small centre of population, with 5000 or more residents [33]). A chart of the flow of participants through the study is presented in Figure 1. Because of the nature of the intervention, it is not possible to blind GPs or participants to the intervention.

The intervention consists of academic detailing, medicines review with web-based pharmaceutical treatment algorithms and tailored patient information leaflets. The academic detailing will involve a research pharmacist visiting intervention GPs in their own practices. The visit will include a short educational presentation about PIP as a concept, the criteria used to measure it and a summary of studies conducted in Irish primary care on the topic, as knowledge of PIP may be a barrier to appropriate prescribing in older patients [34]. Subsequent to this, practices will be asked to complete ten medicines reviews within a 6 to 8 week period, with a reminder issued if they are not completed within that period. An extended period may be negotiated, should a practice require it. During the medicines review, the GPs will use web-based treatment algorithms specifically designed for this study and accessible using a link and designated password. The algorithm will guide the process from the GP perspective, and does not incorporate patient involvement. It has a page-by-page structure, which will be completed when the GP fills in a review outcome form, detailing decisions made by the GP and patient together, including the reasons for maintaining a PIP, which is a key element of this study. Once the review outcome form has been filled in, the medicines review is complete. The medicines review will take place in the GP practice and will be scheduled at a date and time that is convenient to both the GP and the patient. In group practices where more than one GP is participating, the reviews may be divided between them in a manner that is most suitable to the practice workload. The pilot study indicated that the preparation for the review might be more time consuming for the GP than usual but there was no indication that the consultation itself would be significantly longer. These components are presented in Table 1.

The control-group GPs will continue to provide usual care but will also be provided with simple feedback. Data for patients in the control group will be reviewed during recruitment and a personalized patient list for the ten recruited patients will be given to the GP. The list will summarize the medication class to which the individual patient’s potentially inappropriate medication belongs, not the specific PIP and will not provide actionable recommendations for change. Participants will not receive an academic detailing visit, will not be prompted to carry out a medicines review with the individual patients, and will not have access to the pharmaceutical treatment algorithms with alternative therapy options. GPs will continue to provide usual care. In terms of repeat prescribing for public general medical services (GMS) patients, this means that a GP can give a prescription on a monthly basis or for a maximum of three months. At present, the Health Service Executive Primary Care Reimbursement Services (HSE-PCRS) has an on-line GP Application Suite where GPs can review administrative information on their GMS patient panel. They can also access prescribing analysis reports, which contain detailed financial and cost information related to their prescribing. Specific prescribing alerts and recommendations for older patients are not provided.

The control arm will receive simple feedback about their patients based on baseline data collection. Feedback has been found to promote slight improvements in professional practice but is most effective when it is provided intensively [35, 36]. By participating in this research, the control group might also alter their behaviour, resulting in changes in prescribing patterns (that is, the possible Hawthorne effect). To address this, we will also analyze national prescription patterns for all GMS doctors via the HSE-PCRS prescription database after the trial. This is a national prescribing database based on GP and pharmacy claims in a number of community schemes, including the GMS scheme. Data from the PCRS GMS database can be used to compare practices participating in the study with nonparticipating practices, acting as a contemporary national control. Some 330,000 people aged 70 and over were eligible for the GMS scheme in 2009 [37]. Comparisons with previous PIP patterns nationally [11] will determine whether there have been changes over time at the population level.

The study focuses on a select number of PIP criteria identified in the literature that have been determined to be of clinical relevance by academic GPs and pharmacists, and are prevalent in Irish primary care (see Table 2).

The primary outcomes to be determined are the proportion of participant patients with PIP (as a composite measure, that is, any number of PIP criteria as listed in Table 2, to address the issue of multiple PIP in individual patients) and the mean number of potentially inappropriate prescriptions per patient.

Secondary outcomes (summarized in Table 3) will include individual measures of the composite measure, that is, drug-specific outcomes, process evaluations, process-of-care measures and patient-reported outcomes (Health status (EQ-5D) [38], Patients’ Beliefs about Medicine Questionnaire (BMQ) [39], Well-being Questionnaire (WBQ-12) [40]).

Prescription data, process-of-care measures and patient-reported outcomes will be collected at baseline and on intervention completion, that is, the point at which all ten reviews have been completed (this must be within a 6 to 8 week period). Follow-up data will also be collected 12 months after the intervention completion. Data will be collected from the following sources:

The minimization process will ensure balance between treatment groups in terms of certain prognostic factors. Descriptive statistics will be used to evaluate differences in other baseline characteristics between participating physicians and patients in the two arms of the trial. The primary analysis will be carried out using multilevel modeling (such as mixed linear effects modeling or generalized estimating equations [41]) to control for the effects of clustering and baseline differences. All analysis will be conducted under the intention-to-treat principle.

Subgroup analyses will be performed for the primary outcome to assess whether the intervention varies by practice size or GP characteristics, such as sex or number of PIP drugs. As this is a pragmatic trial, a secondary, per-protocol analysis will also be conducted. This form of analysis includes only those participants who completed the treatment protocol originally allocated, providing results on the efficacy of the trial [42].

As all the patients in this study will be selected on the basis of already having one or more potentially inappropriate prescriptions, the sample size calculations are based on a 100% prevalence rate. Separate sample size calculations were performed for the two primary outcomes:

A trial steering committee will be established. The aim of the trial is to identify older patients with existing PIP. These patients will be known to the research team by study ID number only. One member of the research team (BC) will have access to patient contact details for follow-up data collection purposes. The GP remains responsible for all treatment decisions made. Informed consent will be sought from all study participants. All data collected will be stored on a secure, password-protected server. All interviews will be audio recorded and transcribed, the digital recording overwritten and the transcripts pseudo-anonymized and stored on a secure, password-protected server.

The academic detailing will demonstrate the process of the medicines review with the intervention practices but the research team will not monitor how the GP implements the study protocol after this, other than to remind the practices to complete the process within the allotted period. This study is pragmatic in nature, measuring the intervention’s effectiveness in real clinical practice.

Ethical approval was granted by the Research Ethics Committee of the Irish College of General Practitioners (ICGP). At the request of the Ethics Committee, some changes were made to the study protocol. The patient information letter and questionnaire were rewritten to be clearer and simpler for an older audience. Initially, it was proposed that a member of the research team (BC) would become a research agent of the practice, in order to minimize the effort required by the practice staff to recruit patients [44]. However, the ethics committee requested that the practices should be responsible for the patient consent process. The committee also requested that any prescribing pattern of concern identified by the research team should be referred to an external academic GP with no involvement in the trial to assess the case and to determine the necessary next steps; this arrangement has been put in place.

A qualitative process evaluation will be conducted to explore participant attitudes towards the intervention and the experience of the intervention delivery. Specifically, semistructured interviews will be conducted with a sample of participants, both GPs and patients, from the intervention arm. The interviews will be structured using a topic guide, which will be developed with the stages of the interview in mind (introducing the research, beginning the interview, and so on) [45]. These interviews will address such research questions as:

A random sample of participants in the intervention arm will be invited to participate. The number of interviews required to reach data saturation (where no new themes emerge) will be considered, alongside feasibility issues (resources and timing), but a sample of 10 to 15 patients and 10 GPs is proposed [46]. For individual patients, the interviews will take place within one month of the medicines review. For GPs, the interviews will be conducted within one month of them completing their assigned reviews, that is, after all ten reviews for single-handed practices and after completion of assigned reviews in group practices. The interviews will be conducted either in person (in a setting of the participant’s preference) or via telephone. Telephone interviewing is generally used in qualitative research where time or cost is an issue and there is evidence that there is little difference in the answers obtained this way [47]. In this case, scheduling face-to-face interview time with GPs and older patients may be difficult, so telephone interviewing is an option. In addition, as part of the process evaluation, we will conduct brief telephone interviews with the control-group GPs to ascertain any potential impact of the feedback they receive based on the baseline data. Control-group participants will not be required to document any of the decisions made with regard to the control data, so this approach may be more prone to recall bias. However, we consider it a suitable option, given the limited time available in busy GP practices. All interviews will be audio recorded (on loudspeaker for telephone interviews) and transcribed. The data will be collated and a thematic analysis will be conducted. There are four main steps to conducting a thematic analysis:

NVivo 9 will be used to assist with organizing the data for analysis.

A health economic analysis will be conducted following the RCT to explore the direct costs of the intervention and link these to its potential effectiveness. We will compare the direct costs of delivering the OPTI-SCRIPT intervention as an alternative to usual care. Economic analysis is particularly important in relation to quality of prescribing, owing to the considerable costs invested by healthcare systems in medicines and their prescribing. There are potential cost savings from reducing doses and quantities of inappropriately prescribed medications, and from reducing potential adverse events associated with suboptimal prescribing. There is also potential for cost increases if the appropriate alternative medicine recommended is more expensive. Therefore, it will be important to determine the cost/benefit ratios for any changes made in medicines prescribed as a result of the intervention.

Cost-effectiveness analysis will be undertaken, in which effectiveness will be measured in terms of the reduction in the proportion of potentially inappropriate prescriptions. Direct-cost data will be calculated for all the health resources consumed. All contacts with the health service will be recorded and valued, including GP visits, hospital attendances, hospital admissions and drug prescriptions.

At the time of submission of this article, 22 GP practices had been recruited. Patient identification and recruitment was just commencing.