Background Implementation of clinical medication reviews in daily practice is scarcely evaluated. The Opti-Med interven- tion applied a structured approach with external expert teams (pharmacist and physician) to conduct medication reviews. The intervention was effective with respect to resolving drug related problems, but did not improve quality of life. Objective The objective of this process evaluation was to gain more insight into the implementation fidelity of the intervention. Setting Process evaluation alongside a cluster randomized trial in 22 general practices and 518 patients of 65 years and over. Method A mixed methods design using quantitative and qualitative data and the conceptual framework for implementation fidelity was used. Implementation fidelity is defined as the degree to which the various components of an intervention are delivered as intended. Main outcome measure Implementation fidelity for key components of the Opti-Med intervention. Results Patient selection and preparation of the medication analyses were carried out as planned, although mostly by the Opti-Med researchers instead of practice nurses. Medication analyses by expert teams were performed as planned, as well as patient consultations and patient involvement. 48% of the proposed changes in the medication regime were implemented. Cooperation between expert teams members and the use of an online decision-support medication evaluation facilitated implementation. Barriers for implementation were time constraints in daily practice, software difficulties with patient selection and incom - pleteness of medical files. The degree of embedding of the intervention was found to influence implementation fidelity. The total time investment for healthcare professionals was 94 min per patient. Conclusion Overall, the implementation fidelity was moderate to high for all key components of the Opti-Med intervention. The absence of its effectiveness with respect to quality of life could not be explained by insufficient implementation fidelity. Keywords Drug-related problems · Implementation barriers · Implementation fidelity · Medication review · Process evaluation Impact on practice Electronic supplementary material The online version of this article (https ://doi.org/10.1007/s1109 6-018-0615-y) contains supplementary material, which is available to authorized users. Performing medication analyses for clinical medication * F. Willeboordse reviews by external expert teams is feasible. email@example.com Cooperation between fixed expert teams, consisting of a Department of General Practice & Elderly Care Medicine, physician and a pharmacist and the use of an online deci- Amsterdam Public Health Research Institute, VU University sion-support medication evaluation facilitates the imple- Medical Center, Amsterdam, The Netherlands mentation of clinical medication reviews. NIVEL, Netherlands Institute for Health Services Research, Time, cost reimbursement, training and a dedicated prac- Utrecht, The Netherlands tice nurse or coordinator in the GP practice seem to be Department of Information and Computing Sciences, Utrecht necessary for successfully implementing clinical medica- University, Utrecht, The Netherlands tion reviews. In addition, software programs for patient Department of Public Health and Primary Care, Leiden selection, exchange of medical and medication files and University Medical Center, Leiden, The Netherlands outcomes of medication evaluation are needed. Department of Clinical Pharmacology & Pharmacy, VU University Medical Center, Amsterdam, The Netherlands Vol:.(1234567890) 1 3 International Journal of Clinical Pharmacy (2018) 40:550–565 551 The Opti-Med study included three innovative CMR ele- Introduction ments. First, medication analyses were carried out by trained external expert teams consisting of a pharmacist and a physi- Implementation fidelity is defined as the degree to which cian, not being the patient’s own GP and pharmacist. the various components of an intervention are delivered as The second innovative element was a new target group. intended . Convenience of use and degree of implemen- We included patients of 65 years and over who chronically tation exert considerable influence on the applicability of a used ≥ 1 prescribed drug and had one or more geriatric prob- complex healthcare intervention in daily practice. Imple- lems, also called geriatric giants (immobility, instability, mentation fidelity gives researchers and practitioners a better incontinence and impaired cognition) instead of polyphar- understanding of how and why an intervention is effective or macy patients, which is the usual target group. Inappropri- ineffective, and the extent to which health outcomes can be ate medication use may be associated with a higher risk on improved. Implementation fidelity reflects the adherence to the occurrence and persistence of these geriatric problems. content, frequency, duration and coverage of the interven- The nature of this association is complex, as the causes of tion. In addition, there may be moderating factors that influ- these problems are multifactorial; however these geriatric ence the degree of implementation fidelity [1, 2]. As long as problems are among the most common adverse drug reac- the evaluation of the implementation fidelity has not been tions [14–19]. performed, it remains unclear whether ineffectiveness is due The third innovative element was the method of patient to a poor implementation of the intervention or inadequacies involvement. Patients gave input for the medication analyses inherent to the intervention itself. by means of completing a questionnaire and discussed the In this study, the complex intervention of a clinical medi- results of the analyses during a consultation with their GP. cation review (CMR) has been evaluated. A CMR is a struc- We hypothesized that these three elements would facili- tured, critical examination of the patient’s medicines with tate the implementation of CMRs in daily practice and the objective of reaching an agreement with the patient about thereby increase their effectiveness. The results of our treatment, optimising the impact of medicines, minimising effectiveness study showed that the Opti-Med CMRs indeed the number of drug related problems (DRPs) and reducing improved appropriate prescribing, i.e. more DRPs were waste . CMRs can improve the appropriateness of drug identified and solved after 6 months of follow-up compared prescribing and medication use and are increasingly used to usual GP care, but there was no effect on patients’ quality and recommended in primary care [4–7]. However, in daily of life . A process evaluation of the Opti-Med interven- practice the implementation of CMRs is difficult and time tion could clarify whether the limited impact of the Opti- consuming. [8, 9] A recent review highlights the need for Med intervention was due to a poor implementation or due research on intervention development and process evalua- to inadequacies inherent to the intervention itself. tions to improve the understanding of how effective interven- tions to prevent potentially inappropriate prescribing can be sustained and ultimately be translated into improvements in patient outcomes . Therefore, the Opti-Med randomised Aim of the study controlled trial (RCT) was recently carried out in a primary care population to test the effectiveness of CMRs on the The aim of this process evaluation study is to gain more quality of life and DRPs. insight into the implementation fidelity of the Opti-Med The Opti-Med study design and its results have been pub- CMR intervention in daily practice. lished separately [11, 12]. In short, The Opti-Med study was designed as a cluster RCT in 22 general practices (Fig. 1) . We studied the effects of CMRs on quality of life Method and DRPs in 518 older patients (≥ 65 year). Patients were selected and invited when they chronically used one or more Study design prescribed drugs and newly presented themselves to the gen- eral practitioner (GP) with one or more geriatric problems This process evaluation was conducted alongside the (immobility, instability, incontinence and impaired cogni- Opti-Med RCT. Within the present study, the implemen- tion). Patient selection was facilitated by software specifically tation fidelity of the Opti-Med intervention was evaluated. developed for the Opti-Med study based on electronic medi- Quantitative data was collected from the start of the study cal records (EMRs). CMRs were conducted by the expert and qualitative data was collected at the end of the study. teams according to a structured program using the STRIPA For the evaluation we distinguished five key intervention tool . Patients in control practices received usual GP care components: with no specific attention to their medication use. 1 3 552 International Journal of Clinical Pharmacy (2018) 40:550–565 Fig. 1 Overview of the Opti-Med intervention and important ele- right treatment, STOPP screening tool of older person’s prescriptions, ments for the process evaluation. DRPs drug related problems, EMR STRIP systematic tool to reduce inappropriate prescribing, STRIPA electronic medical record, GP general practitioner, PTP pharmaco- systematic tool to reduce inappropriate prescribing assistant. Ques- therapeutic treatment plan, START screening tool to alert doctors to tionnaire by Willeboordse et al.  A. Patient selection and invitation by GPs and practice E. GP consultation with the patient and implementation of nurses to participate using EMRs through a newly devel- the PTP. oped software; B. Patient involvement through a patient questionnaire ; C. Preparation of the medication analysis by practice nurses Conceptual framework for implementation and Opti-Med researchers; fidelity D. Medication analysis and drafting of a Pharmacothera- peutic Treatment Plan (PTP) by an expert team. The The adapted Conceptual Framework for Implementation expert teams followed accredited online courses for Fidelity was used (Fig. 2) [1, 2]. The framework allows to CMRs and two face-to-face CMR workshops. An elec- evaluate both adherence to the intervention and to assess tronic medication evaluation tool, the Systematic Tool moderating factors for adherence to the intervention. to Reduce Inappropriate Prescribing Assistant (STRIPA)  was used for the medication analysis; 1 3 International Journal of Clinical Pharmacy (2018) 40:550–565 553 Fig. 2 Adapted conceptual framework for implementa- tion fidelity for the Opti-Med process evaluation. The measurement of implementa- tion fidelity is the measurement of adherence of the categories content, frequency, duration and coverage Adherence to the intervention includes the dimensions Study administration content, frequency, duration and coverage. Moderating factors for adherence to the intervention Data on selection, inclusion and drop-out of participants, include the dimensions participant responsiveness, strategies time planning, performing medication analyses by the expert to facilitate implementation, quality of delivery and context. teams, and consultations with the GP were recorded by the Specific research questions and outcomes per key inter - researchers alongside the RCT. vention component (A–E) for each dimension of the concep- tual framework are presented in Tables 1 and 2. A subjective Focus group with experts rating was used to evaluate the implementation fidelity and the researchers assigned the ratings for each dimension of A focus group was held with seven members (one GP, two the framework using four categories: very low, low, moder- elderly care specialists and four pharmacists) of the four ate, high. ‘Very low’ means that almost none of the interven- expert teams to collect data on their experiences with con- tion elements were carried out as planned, ‘low’ means that ducting the medication analyses. The meeting lasted 70 min some elements have been carried out as planned, ‘moderate’ and was audio recorded. To facilitate the discussion a topic means that the majority of the elements have been carried list was developed beforehand (online resource 1). out as planned and ‘high’ means that almost all elements have been carried out as planned. Interviews with the patients’ GPs From each intervention practice that performed more than ten consultations, a GP was invited for an semi-structured Data sources interview; all participated. The interviews were held by the researchers, lasted 15–30 min and were audio-recorded. The The following data sources were used to address the specific objective of the semi-structured interviews was to discuss research questions. the experiences of the GPs with this method of conducting 1 3 554 International Journal of Clinical Pharmacy (2018) 40:550–565 1 3 Table 1 Research questions for the evaluation of adherence, data sources and outcomes for the implementation fidelity of the Opti-Med intervention Key intervention components Data Specific research questions Outcomes Rating* source 1. Evaluation of adherence: content 1a. Patient selection I To what extent was the patient selection implemented as Patient selection was carried out as planned according to the inclusion cri- Moder- planned? teria. However, in practice it was not fully carried out by practice nurses ate but researchers provided extensive support or carried it out completely 1b. Patient involvement IV To what extent did the patient questionnaire information Patient questionnaire information was often used to tailor the PTP. Face- High influence and tailor the PTP? to-face patient contact might have resulted in more useful information according to the expert teams, e.g. compliance problems 1c. Preparation of medica- I To what extent was the preparation of the medication analyses The preparation of the medication analyses was carried out by the Moder- tion analysis implemented as planned? researchers, therefore not fully implemented as planned. The gathering ate of information (medical EMR data and medication data from pharmacy) was planned to be carried out by the practice nurses. Medication analysis preparation was deemed sufficient by the expert teams 1d. Medication analysis II To what extent was the medication analysis implemented as Medication analysis by the expert team was carried out in a structured High planned? (structure, cooperation, STRIPA, knowledge and manner due to the use of the IT application STRIPA. Cooperation was drafting the PTP) good and complementary knowledge helpful. All expert teams formed fixed couples which improved cooperation and efficiency. Frequency, often once per month, also improved cooperation, efficiency and knowl- edge. All expert teams used primary care guidelines and applied STOPP and START criteria. The drafting of the PTP was deemed easy due to the structured STRIPA format but the lay-out and overview could be improved 1e. GP consultation II, III To what extent were patient consultations delivered and GPs differently performed the consultation: most GPs planned double con- High prepared as planned? sultation time and used a few minutes to prepare the consultations using the PTP form. In one practice, consultations were thoroughly prepared and discussed by phone, in another practice over half of the patients were visited at home. In two practices, the practice nurse did the consultation with the patient and only discussed major changes with the GP. As the result there was more attention for patient knowledge, compliance and preferences According to GPs the use of external expert teams brought advantages such as efficiency and feasibility (as needed when conducting CMRs for larger numbers of patients), objectivity, expertise, and extra convincing power towards the patient. However, they also considered a final evalua- tion by the GP always necessary but requiring a certain time-investment. The expert teams mentioned advantages like objectivity. Not knowing the patient may also circumvent preconceptions by the patients’ GP 2. Evaluation of adherence: frequency 2a. Patient selection I How many times a patient selection was performed? Patients were selected approximately every 2–3 months and a list with High eligible patients was composed. Out of 112 possible lists, 105 (94%) lists were successfully processed. In total 3 lists could not be produced due to software problems and 4 lists were produced but not processed by the GP due to time constraints International Journal of Clinical Pharmacy (2018) 40:550–565 555 1 3 Table 1 (continued) Key intervention components Data Specific research questions Outcomes Rating* source 2b. Patient involvement IV, VI How many patient questionnaires were completed and com- All questionnaires were filled in by the participants. 17% of the patients High pleted by the patient themselves? did not fill out the questionnaire independently but were assisted by fam- What was the influence of the patient input on the identified ily or other informal carers or visited at home by the researchers DRPs? 19% of all DRPs were identified on the basis of patient questionnaire spe- cific data on actual medication use, DRPs, geriatric problems and pain 2c. Preparation of medica- I How many CMRs were prepared? All 518 CMRs were prepared as planned. For 11 patients the medication High tion analysis list from the community pharmacy was not received (in time), and the medication list provided by the patient and/or the GP was used A medication analysis was performed for 274 of 275 participants in the High 2d. Medication analysis I, IV, V How many medication analyses were performed? intervention group (one drop-out before expert team started) and for all How many proposed interventions and DRPs were formu- 243 control patients lated? See Fig. 3 for the frequency, nature and implementation rate of the pro- To what extent were the proposed interventions implemented posed interventions and drug related problems, including reasons for not as planned? follow-up the interventions. For 275 intervention patients, 1282 interven- Were there differences in implementation rate for different type of proposed interventions? tions were proposed by the external expert teams and documented on the pharmacotherapeutic treatment plans. Retrospectively, the researchers identified 1212 drug related problems with the DOCUMENT tool, out of these proposed interventions. In total, there were 8 patients without any DRPs The implementation rate was higher for non-pharmacological interventions than pharmacological interventions, 69.2% compared to 42.6% (t test p < 0.001) The implementation rate for addition of drug was higher than for cessation of drug, 46.7% compared to 34.7% (t test, p = 0.002) 2e. GP consultation I How many GP consultations were performed? 90% (247) of the PTPs were discussed with the patient by the GP High 3. Evaluation of adherence: duration 3a. Patient selection I What was the estimated duration to select a patient? About 1 min per patient NA 3b. Patient involvement – NA NA 3c. Preparation of medica- I What was the estimated duration to prepare a medication 15 min per patient (including gathering of information, enter data and NA tion analysis analysis? process the PTP) How many days were there between inclusion and GP consul- Median (IQR) number of days between inclusion and the consultation was tation date? 33.0 (15–51) days 42% of the patients had their consultation within the planned 1 month after inclusion 3d. Medication analysis VII What was the mean duration of a medication analysis by the Mean [sd] 22  minutes per expert team member per patient NA expert team? 556 International Journal of Clinical Pharmacy (2018) 40:550–565 1 3 Table 1 (continued) Key intervention components Data Specific research questions Outcomes Rating* source 3e. GP consultation VII What was the mean duration of a GP consultation? Mean [sd] 34  minutes per patient NA 4. Evaluation of adherence: coverage 4a. General I, VIII What proportion of the selected patients was invited to 2401 patients were initially selected on the basis of their GP EMR, 2037 High participate? (85%) patients were invited to participate. 364 (15%) patients were What proportion of the invited patients participated and how excluded after selection by the GP because they were terminally ill or was the drop-out and follow-up? due to a specific reason why it was not desirable to invite the patient Were there differences between GP practices? (range 4–33% between GP practices) Were there differences in patient characteristics between the 25% were included (range 12–33% between GP practices) responders and non-responders? 15% was considered not eligible (range 0–22% between GP practices) 41% did not respond at all (range 26–64% between GP practices) 19% declined to participate (range 8–31% between GP practices) See figure for the patient flow in the Opti-Med study in “Online resource 1” Patients who declined to participate did not differ in age as compared to participants, but among them there were significantly less women (χ , p = 0.02) and they used less medication (t test, p < 0.001) Patients who did not respond did not differ in age and gender as compared to participants CMR clinical medication review, DRP drug related problem, EMR electronic medical record, GP general practitioner, IT information technology, NA not applicable, PTP pharmacotherapeutic treatment plan, START screening tool to alert doctors to right treatment, STOPP screening tool of older person’s prescriptions, STRIPA systematic tool to reduce inappropriate prescribing assis- tant I. Study administration II. Focus group with expert teams III. Semi-structured interviews with GPs IV. PTPs and evaluation forms V. Assessment of DRPs and STOPP and START criteria VI. Inclusion patient questionnaire VII. Time registration by expert teams and GPs VIII. GP EMR data IX. Patient survey after 3 months X. Short survey among GPs of control practices *Rating of implementation fidelity (very low, low, moderate, high) International Journal of Clinical Pharmacy (2018) 40:550–565 557 CMRs. To facilitate the interview a topic list (Electronic Patient survey Supplementary Material 3) was developed. The intervention patients completed a survey 3 months Evaluation of the implementation of the results after baseline. The survey assessed the preparation and of the medication analyses usefulness of the CMR and satisfaction about the consulta- tion with the GP. An evaluation form was used by the GPs to record the fol- low-up of the changes in the medication regime as proposed Survey among GPs in control practices by the expert team, including the reason(s) why (part of) these proposals were not implemented. The expert team also GPs from the control practices received a short survey indicated for each proposal whether this was influenced by to assess whether CMRs were conducted unintentionally the input of the patient via the questionnaire. during the study period for patients of the control group. Classification and assessment of DRPs Analyses The changes in the medication regime as proposed by the expert teams were classified by the researchers (FW, JH) into Descriptive statistics were used for quantitative data using DRPs using the DOCUMENT DRP classification system SPSS Statistics 23, using t tests for continuous variables . and χ statistics for categorized variables. For a random sample of 21 (8%) of all patients a medica- For qualitative analyses, audio files were transcribed tion analysis was performed by two different expert teams verbatim. Transcripts of the focus group and interviews to assess reproducibility. were coded by two independent researchers (respectively Subsequently, the STOPP and START criteria were FW and MD, and FW and SY) top-down with a pre-defined applied to these DRPs to establish their external validity. code-list which was formulated based on the topic lists and STOPP (Screening Tool of Older Person’s Prescriptions) is a knowledge of the intervention. Differences in coding were list of medications that are potentially inappropriate for older discussed until consensus was reached, a few codes were people. START (Screening Tool to Alert doctors to Right added retrospectively. Citations and coded transcripts were Treatment) is a list of medications that should be prescribed arranged to broader themes using Atlas.ti software . for older people for a number of conditions. The assessment was carried out by one researcher (HvD) by means of an iterative process. Eventual difficulties were discussed with a second researcher (FW) until consensus was reached. A Results random sample of 10% of the patients was independently assessed by a second researcher (FW). Outcomes per key intervention component for each dimen- sion of the framework are shown in detail in Table 1 and 2. Patient questionnaire At inclusion, patients completed a questionnaire about their Adherence to the intervention actual medication use and experienced problems with their medication. The patients indicated whether they filled out Patient selection was carried out according to the inclu- the questionnaire independently or whether they received sion criteria. However, for this topic, we deviated from help. the study protocol, most practice nurses did not carry out patient selection and invitation themselves due to difficul- Time registration ties in using the newly-developed software application and due to time restraints. The Opti-Med researchers provided The time investment of the expert teams and the GPs in the extensive support or carried out the patient completion intervention practices for completing the respective elements themselves instead. of the intervention was calculated by the researchers. Also, the Opti-Med researchers collected most informa- tion (GP EMR data, medication overview from pharmacy Electronic medical records and patient questionnaire) for the medication analyses instead of the practice nurses, due to time restraints. Data on gender and age from the GPs’ EMRs was used for the non-responder analysis. 1 3 558 International Journal of Clinical Pharmacy (2018) 40:550–565 1 3 Table 2 Research questions for the evaluation of moderating factors, data sources and outcomes for the implementation fidelity of the Opti-Med intervention Key intervention componentsData source Specific research questions Outcomes 1. Moderating factors: participant responsiveness 1a. General I, III, IX How were patients informed about, and engaged in the interven- Patients received an information letter including a customized tion? leaflet to prepare for the GP consultation How was the patient recall of consultation? GPs found most patients were well informed and pleased with the How did patients prepare for the consultation? extra attention for their medication How did patients perceive the intervention? 231 intervention patients filled out the questionnaire at 3 months: 14% of the patients that had a consultation with their GP did not recall the consultation 48% did not prepare particularly for the consultation, 23% brought or studied his/her medication overview, 24% thought of or noted down questions beforehand and 4% brought someone to the consultation 88% of the patients perceived the consultation as pleasant or very pleasant 72% thought the consultation was useful or very useful 2. Moderating factors: strategies to facilitate implementation 2a. Patient selection I What strategies were used to support patient selection? Patient selection was carried out using a specially designed ICT application that searched GP EMR records on the basis of the study inclusion criteria. Due to difficulties in applying the applica- tion and time restraints only a few practice nurses were able to carry out the patient selection independently. The majority needed help from the researchers 2b. Patient involvement I What were strategies to support implementation of the interven- The patient questionnaire and the customized leaflet to prepare tion and patient involvement? patients for the consultation were strategies to involve patients in their own CMR and tailor it to their needs Patients could ask for assistance in filling out the patient question- naire on actual medication use and DRPs as needed to prepare the medication analysis. Only a very limited number of patients used this option 2c. Preparation of medication analyses I What strategies were used to support the preparation of the medi- Although time-consuming and prone to error, convenient use was cation analysis? made of the STRIPA. Collecting information from the GP EMR and pharmacy was also convenient but time-consuming due to limitations of the GP IT systems International Journal of Clinical Pharmacy (2018) 40:550–565 559 1 3 Table 2 (continued) Key intervention componentsData source Specific research questions Outcomes 2d. Medication analysis I, II What strategies were used to support expert teams in implement- Expert teams followed an online course, 5 h professional training ing the medication analyses? to prepare for the medication analyses and a 2 h feedback meeting How were these strategies perceived by the expert teams? after 2 months into the intervention. During the first sessions all expert teams were assisted by the researchers to help with the software package and available for questions. STRIPA was used to support the medication analyses The training was deemed useful, especially to get acquainted with STRIPA and with the fellow expert team member. Most skills and knowledge were acquired during the course of the study. The expert teams all indicated that the STRIPA was a big support for the structured medication analysis. A barrier was that STRIPA was not supporting the drafting of PTPs 2e. GP consultation I, III What were strategies to support implementation of the interven- Intervention GPs were informed by the researchers during a kick- tion by the GPs and practice nurses? off meeting and received printed materials on the intervention. How were these strategies perceived by the GPs? Practice nurses received a workbook with practical steps and the researchers assisted the practice nurses when needed and were available for questions via e-mail or phone. We tried to adapt the PTP forms to a format usable in the GP EMR but integration proved impossible to integrate the PTP How the kick-off meeting printed materials and communication was perceived is unknown. GPs indicated that once they got used to the PTP evaluation forms they were easy and structured but some considered the non-compatibility with the GP IT system a barrier 3. Moderating factors: quality of delivery 3a. Patient selection II How was the quality of the patient selection and how was this Quality of the patient selection is not relevant and not addressed. evaluated by the GPs? 3b. Patient involvement IV How was the quality of the patient involvement? Implementation rate of DRPs modified on basis of patient input was significantly higher as compared to DRPs not modified on basis of patient input (respectively 60 and 46%, t test p < 0.001) 3c. Preparation of medication analyses II, III How was the quality of the preparation of the CMRs and how was The quality of the preparation was good but occasionally medical this evaluated by the expert team and GPs? or medication files were incomplete. The quality of the medical files differed between GP practices. As a consequence in these cases recommendations were less useful and it required more effort of the GP to conduct the patient consultation. However, GPs reported that in most cases incorrect data could be easily corrected and incorrectly proposed interventions were ignored or adjusted 560 International Journal of Clinical Pharmacy (2018) 40:550–565 1 3 Table 2 (continued) Key intervention componentsData source Specific research questions Outcomes 3d. Medication analysis III, IV, V How was the reproducibility of the medication analysis? PTP reproducibility between different expert teams was moderate. To what extent were DRPs and proposed interventions related to A mean [sd] of 1.5 [1.2] in the number of DRPs and 2.4 [1.4] the STOPP and START criteria in the intervention patients? deviations in type of DRPs was found per patient between two How was the quality of the medication analyses evaluated by the different expert teams GPs? In total 33.1% of the DRPs was related to a STOPP criterion and 19% to a START criterion. For details see Table 3 GPs considered PTPs of good quality and more elaborate than they were used to from other polypharmacy projects or community pharmacist initiatives 3e. GP consultation IX How was the quality of the GP consultation according to the 82% indicated to understand everything or almost everything during patient, in terms of understanding and asking questions? the consultation 75% indicated that they could ask all questions or almost all ques- tions during the consultation 4. Moderating factors: context 4a. General I, III, X How did the organization of GP practices affect the implementa- There were differences between GP practices in how easy the inter - tion? vention was embedded into daily practice. Implementation went How did attention for polypharmacy in primary care and in soci- much smoother in GP practices in which a practice nurse was ety affect the implementation? assigned to organize this type of interventions. Personnel changes To what extent was the usual care in the control group imple- during the course of the study were barriers for continuation of mented as planned? (possible contamination)? the intervention and good implementation How were expert teams and GPs reimbursed? GPs in the intervention group found specific attention for poly - pharmacy, medication reviews and the primary care guideline encouraging and considered them important GP care topics The 11 control GP practices confirmed that no structured CMRs were conducted, except in a small student project. However, all practices were involved in an elderly care project with extra attention for pharmacotherapy but only a small number of their patients was involved No reimbursements were offered to patients. Expert teams were paid an hourly rate for their work in the medication analyses. GP practices were paid per patient included CMR clinical medication review, DRP drug related problem, EMR electronic medical record, GP general practitioner, IT information technology, PTP pharmacotherapeutic treatment plan, START screening tool to alert doctors to right treatment, STOPP screening tool of older person’s prescriptions, STRIPA systematic tool to reduce inappropriate prescribing assistant I. Study administration II. Focus group with expert teams III. Semi-structured interviews with GPs IV. PTPs and evaluation forms V. Assessment of DRPs and STOPP and START criteria VI. Inclusion patient questionnaire VII. Time registration by expert teams and GPs VIII. GP EMR data IX. Patient survey after 3 months X. Short survey among GPs of control practices International Journal of Clinical Pharmacy (2018) 40:550–565 561 Nineteen percent of all DRPs identified were based on someone to the consultation. Fourteen percent of the patients the input from the patient questionnaire. The majority who had a consultation with the GP did not recall it. Of the of these DRPs were related to medication knowledge or patients who did recall the consultation, the majority con- adherence to medication. sidered it useful. The expert teams carried out medication analyses for all but one of the 275 participants of the intervention group and for all 243 control patients. According to the Strategies to facilitate implementation expert team members, medication analyses were con- ducted in a highly structured manner, mainly due to use Patient selection was facilitated by software specifically of the STRIPA tool. They also mentioned that the method developed for the Opti-Med study. However, most practice and high number of medication analyses by fixed couples nurses considered it difficult to use and time consuming. improved efficiency and collaboration. The expert team Collecting information from the GPs’ EMRs and pharmacy members and the GPs mentioned the ‘external’ nature of records in preparation of the medication analyses was use- the team as an additional value, because of the fresh per- ful but time-consuming. The quality of the preparation for spective of such a team allowing an independent ‘objec- the medication analysis was deemed sufficient by the expert tive’ assessment. teams. In 90% (247/275) of the patients, GPs discussed the pro- Training in performing CMRs was deemed useful by the posed changes in medication with their patients. 42% of the expert team members. However, they indicated that most patients had their consultation within the planned first month knowledge and skills were acquired when performing the after inclusion. The method of consultation was deliberately medication analyses. The use of the STRIPA tool was found not specified by the researchers. Most GPs planned double to greatly support and to highly structure the medication consultation time and used a few minutes to prepare the con- analysis. Some GPs indicated that the form with the PTP sultations using the PTP. was not very user-friendly; however, after a few consulta- Figure 3 gives an overview of the frequency, nature of tions, most GPs became familiar with it. Seventeen percent DRPs and proposed changes in medication as well as their of the patients reported to have been assisted in completing implementation rate, and reasons for not implementing as the patient questionnaire. proposed. Nearly 50% of all proposed medication changes were (partially) implemented (consented implementation). Quality of delivery ‘Addition of a drug’ was significantly more often imple- mented than ‘cessation of drug’ (46.7 vs. 34.7% (t test, The GPs considered the PTPs drafted by the expert teams p = 0.002). The implementation rate of non-pharmacological of very good quality. recommendations (e.g. laboratory tests) was significantly The mean difference between the number of DRPs per higher than proposed changes in medication (69.2 vs. 42.6% patient identified by two expert teams was 1.5 (standard (t test p < 0.001). The most frequent reasons for non-imple- deviation (SD) 1.2) and the mean number of differences in mentation were: ‘proposed change is based on incomplete type of DRPs was 2.4 (SD 1.4). medical or medication files’, ‘prescription originates from a In total 33.1% of the DRPs identified were related to a medical specialist in secondary care’ or ‘the change in medi- STOPP criterion and 19% to a START criterion (Table 3), cation has been tried before by patient and/or prescriber’. but a considerable part of the identified DRPs could not be The total time spent by all healthcare providers for one related to a STOPP or START criterion (e.g. practical medi- patient was estimated at 94 min. This includes 1 min for cation problems, changes in dosage or evaluation of drug patient selection, 15 min for preparation, 22 min per expert effect). team member for medication analysis and 34 min for GP The majority of the patients indicated that they could ask consultation. (almost) all questions and understood (almost) everything during the consultation with the GP. The implementation rate of proposed medication changes influenced by patient input was significantly higher as Moderating factors compared to the implementation rate of proposed changes not influenced by patient input (respectively 60 and 46%, Participant responsiveness p < 0.001). Over half of the patients reported to have prepared them- selves for the consultation with the GP by bringing or study- ing their own medication, preparing questions, or bringing 1 3 562 International Journal of Clinical Pharmacy (2018) 40:550–565 the course of the study was a barrier for the continuity and Contextual factors implementation of the intervention. GPs considered the increased attention for polypharmacy, medication reviews, and the recently published Dutch mul- Discussion tidisciplinary guideline on polypharmacy  encouraging and important for GP care. CMRs were not performed for For all key intervention components the implementation patients in the control practices, therefore contamination was minimal. fidelity was moderate to high. Almost all key intervention components were generally carried out as planned. How- The embedding of the Opti-Med intervention varied between GP practices. GPs and practice nurses reported ever, for the elements patient selection and preparation of the CMR analyses the researchers were more involved than less complaints and questions from patients when a practice nurse was specifically assigned to the organization of the intended. Almost half of the proposed changes in medication were implemented, starting new medications seemed easier intervention. GPs mentioned that personnel changes during Fig. 3 Frequency and nature of proposed changes and drug related teams. Retrospectively, the researchers identified 1212 drug related problems. For 275 intervention patients, 1282 pharmaceutical and problems with the DOCUMENT tool , out of these proposals non-pharmaceutical changes were proposed by the external expert 1 3 International Journal of Clinical Pharmacy (2018) 40:550–565 563 Table 3 Prevalence of STOPP-START among intervention patients The 94 min time spent is acceptable compared to other per DOCUMENT DRP type studies and estimations in guidelines [7, 27]. Almost a quarter of the time is spent by the practice nurse instead DOCUMENT DRP type Total STOPP START of the GP and/or pharmacist, which is less costly. How- N (%) N (%) N (%) ever the time investment is still considerable, but may Drug selection 471 (38.9) 372 (30.7) 17 (1.4) reduce over time. A previous study with Opti-Med data Over or underdose prescribed 99 (8.2) 7 (0.6) 3 (0.2) shows that the expert teams can improve the efficiency Compliance 45 (3.7) 3 (0.2) 1 (0.2) over time [28.] Un(der)treated indications 343 (28.3) 1 (0.1) 212 (17.5) The moderate reproducibility of the medication analyses Monitoring 145 (12.0) 0 0 between the expert teams could be partly explained by vari- Education or Information 38 (3.1) 0 0 ations among experts. In a recent Dutch qualitative study Not classifiable 17 (1.4) 0 0 on case vignettes with polypharmacy and multimorbidity, it Toxicity or ADR 54 (4.5) 18 (1.5) 0 was concluded that GPs varied in medication management Total 1212 (100) 401 (33.1) 233 (19.2) strategies which resulted in differences in proposed medica- tion changes . ADR adverse drug reaction, DRP drug related problem, START screening tool to alert doctors to right treatment, STOPP screening tool of older person’s prescriptions Lessons learned for CMRs in a non‑RCT setting DRPs were identified by the expert team at baseline and classified by the researchers according to the validated DOCUMENT  classifi- This process evaluation provides a better insight into the cation system to categorize DRPs into 8 categories. Retrospectively, implementation fidelity of an innovative method for CMRs. STOPP and START criteria were assigned to the DRPs Implementation fidelity was studied alongside a pragmatic cluster RCT, which does not resemble daily practice. E.g., than stopping medications. Patient involvement may also be the efforts and time investment of the researchers are appli- considered accomplished as planned, one-fifth of the pro- cable in daily practice. posed medication changes was influenced by patient input. As the selection of patients and preparation of the CMRs Training of the expert teams, the use of the STRIPA tool in this study was mainly performed by researchers there and the structured PTP forms facilitated implementation are still some barriers to overcome before these key inter- of the intervention. Difficulties with patient selection due vention components can be successfully implemented in to non user-friendly software and incomplete medical and daily practice. Time, training and dedication of a practice medication files used for the medication analyses appeared assistant or practice nurse in the GP practice for CMRs factors promoting non-adherence to the intervention. The are necessary. reproducibility of the medication analyses between the The medication analyses being performed by external expert teams was moderate. There were differences in the expert teams seems feasible, however reimbursement and embedding of the intervention between GP practices. A des- organization of expert teams outside the scope of a research ignated and motivated practice nurse was an important con- project will be necessary. Currently in The Netherlands GPs textual facilitating factor for adherence to the intervention. and pharmacists are reimbursed for conducting CMRs. A To our knowledge, this is one of the first comprehensive dedicated coordinator may be needed to organise the work process evaluations of a CMR intervention study. Other of expert teams within e.g. an existing regional collaboration studies on CMRs did not included or only a limited process structure between GPs and/or pharmacists. evaluation or a different method of CMR [23, 24]. A com- Reimbursements for the GPs and reminders by the parison with previous studies is therefore difficult, however, researchers for GPs and patients may have increased the some results can be compared. implementation rate of the GP consultations. Of the invited The implementation rate of proposed medication changes patients, almost 60% did not reply or indicated that they did of almost 50% is within the range found in other studies [25, not want to participate. It might be that in daily practice, a 26], higher implementation rates may be found when the part of this group may need a different approach with pos- patient’s own pharmacist and GP are involved in the medi- sibly more face-to-face contact to identify the actual medica- cation analysis and less non-relevant recommendations may tion intake, DRPs and preferences. be formulated. However, GPs did not experience the irrel- Identified barriers for implementation in daily practice, evant recommendations as inefficient and time consuming such as time restrains and incompleteness of medical files and reported that this disadvantage often was outweighed by are commonly known from other pharmaceutical care stud- the advantage of the efficiency, objectivity and expertise of ies or evaluation projects [8, 24, 30]. the external expert team. 1 3 564 International Journal of Clinical Pharmacy (2018) 40:550–565 interviews transcripts. Hanna van Daal (HvD), thank you for your work Limitations with the assessment of all STOPP and START criteria and data entry. Several limitations may have influenced the evaluation of Funding This study was funded by a research grant by the Dutch the adherence to the intervention and moderating factors Organization for Health Research and Development (ZonMw). determining the implementation fidelity of the intervention. Conflicts of interest All authors declare that they have no conflict of First, the researchers who carried out the Opti-Med inter- interest. vention were also involved in the process evaluation. We used a subjective rating to measure implementation fidel- Ethics approval This study was approved by the Medical Ethics ity, an objective rating is impossible in this type of process Committee of the VU University Medical Center (approval reference 2011/408) Informed consent was obtained from all individual partici- evaluations. pants included in the study. Second, as compared to the framework of Hasson the moderating factors ‘comprehensiveness of the policy Open Access This article is distributed under the terms of the Crea- description’ and ‘recruitment’ have not been included in tive Commons Attribution 4.0 International License (http://creat iveco the present evaluation. Comprehensiveness of the policy mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- description was not assessed since the number of key compo- tion, and reproduction in any medium, provided you give appropriate nents in the intervention is limited and it was not feasible to credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. obtain an external assessment of the policy description with respect to the complex intervention. Recruitment is covered under the adherence dimension ‘coverage’. Furthermore, not all dimensions of adherence and of the moderating fac- References tors have been assessed extensively. The assessment of the quality of delivery of the intervention for GP consultations 1. Carroll C, Patterson M, Wood S, Booth A, Rick J, Balain S. A conceptual framework for implementation fidelity. Implement Sci. and patient involvement was very limited. Video recordings 2007;2:40. of consultations might have provided more insight into the 2. Hasson H. Systematic evaluation of implementation fidelity of quality of delivery. The duration and topic list of the GP complex interventions in health and social care. Implement Sci. interview was limited. Finally, results from a patient survey 2010;5:67. 3. Clyne W, Blenkinsopp A, Seal R. A guide to medication review. gave us only limited insight into the patients’ responsiveness UK: NHS National Prescribing Centre; 2008. and quality of delivery of the patient involvement, compared 4. Cooper JA, Cadogan CA, Patterson SM, Kerse N, Bradley MC, to e.g. qualitative patient interview data. Ryan C, et al. Interventions to improve the appropriate use of polypharmacy in older people: a cochrane systematic review. BMJ Open. 2015;5(12):e009235. 5. NHS Cumbria Medicines Management Team. Clinical medication review: a practice guide. UK, 2013. Conclusion 6. Pharmaceutical Society of Australia. Guidelines for pharmacists providing home medicines review (HMR) services. Australia, Overall, the implementation fidelity was moderate to high 7. Nederlands Huisarts Genootschap (NHG). Multidisciplinaire for all key intervention components of the CMR interven- richtlijn Polyfarmacie bij ouderen. (Multidisciplinary guideline tion. This means that almost all intervention key compo- Polypharmacy for elderly). The Netherlands, 2012. nents were delivered as intended. The absence of its effec - 8. de Bruijne MC, Kemper PF, Bakker L, Heeren MJ, Meijler AP, Dolwel GO, et al. Farmacotherapeutische zorg voor voor kwets- tiveness with respect to enhancing quality of life cannot be bare ouderen met polyfarmacie (Pharmaceutical care for frail explained by insufficient implementation fidelity. Neverthe- elderly with poypharmacy). Amsterdam, 2014. less, this process evaluation provides insight into how this 9. Rosenthal M, Holmes E, Banahan B, III. Making MTM imple- method of conducting CMRs can be implemented in daily mentable and sustainable in community pharmacy: Is it time for a different game plan? Res Soc Adm Pharm. 2015. practice. Barriers on organizational level must be overcome; 10. Clyne B, Fitzgerald C, Quinlan A, Hardy C, Galvin R, Fahey T, the availability of user-friendly software, easy exchange of et al. Interventions to address potentially inappropriate prescribing medical and medication data, and coordination and man- in community-dwelling older adults: a systematic review of rand- agement of the intervention within a larger collaboration omized controlled trials. J Am Geriatr Soc. 2016;64(6):1210–22. 11. Willeboordse F, Hugtenburg JG, van DL, Bosmans JE, de Vries between GPs and pharmacists are very important for suc- OJ, Schellevis FG, et al. Opti-Med: the effectiveness of optimised cessful implementation. clinical medication reviews in older people with ‘geriatric giants’ in general practice; study protocol of a cluster randomised con- Acknowledgements We would like to thank all GPs, GP employees trolled trial. BMC Geriatr. 2014;14:116. and expert team members who facilitated data collection for this pro- 12. Willeboordse F, Schellevis FG, Chau SH, Hugtenburg JG, Elders ject. Furthermore, thanks to Melek Dogdu (MD), Sabri Yigit (SY) for PJ. The effectiveness of optimised clinical medication reviews for their help with coding respectively the focus group and semi-structured 1 3 International Journal of Clinical Pharmacy (2018) 40:550–565 565 geriatric patients: opti-med a cluster randomised controlled trial. 22. Atlas.ti version 7.5.16 Gmbh [Computer software]. Berlin. Fam Pract. 2017;34(4):437–45. 23. Allard J, Hebert R, Rioux M, Asselin J, Voyer L. Efficacy of a 13. Meulendijk MC, Spruit MR, Drenth-van Maanen AC, Numans clinical medication review on the number of potentially inappro- ME, Brinkkemper S, Jansen PA, et al. Computerized decision priate prescriptions prescribed for community-dwelling elderly support improves medication review effectiveness: an experi- people. CMAJ. 2001;164(9):1291–6. ment evaluating the STRIP assistant’s usability. Drugs Aging. 24. Clyne B, Cooper JA, Hughes CM, Fahey T, Smith SM, Team 2015;32(6):495–503. O-Ss. A process evaluation of a cluster randomised trial to reduce 14. Chrischilles E, Rubenstein L, Van Gilder R, Voelker M, Wright potentially inappropriate prescribing in older people in primary K, Wallace R. Risk factors for adverse drug events in older adults care (OPTI-SCRIPT study). Trials. 2016;17(1):386. with mobility limitations in the community setting. J Am Geriatr 25. Kwint HF, Bermingham L, Faber A, Gussekloo J, Bouvy ML. Soc. 2007;55(1):29–34. The relationship between the extent of collaboration of general 15. Dros J, Maarsingh OR, Beem L, van der Horst HE, ter Riet G, practitioners and pharmacists and the implementation of recom- Schellevis FG, et al. Functional prognosis of dizziness in older mendations arising from medication review: a systematic review. adults in primary care: a prospective cohort study. J Am Geriatr Drugs Aging. 2013;30(2):91–102. Soc. 2012;60(12):2263–9. 26. Chau SH, Jansen AP, van de Ven PM, Hoogland P, Elders PJ, 16. Larson EB, Kukull WA, Buchner D, Reifler BV. Adverse drug Hugtenburg JG. Clinical medication reviews in elderly patients reactions associated with global cognitive impairment in elderly with polypharmacy: a cross-sectional study on drug-related prob- persons. Ann Intern Med. 1987;107(2):169–73. lems in the Netherlands. Int J Clin Pharm. 2016;38(1):46–53. 17. Ruby CM, Hanlon JT, Boudreau RM, Newman AB, Simonsick 27. Mast RM, Schouten GP, van Woerkom M. Niveau van medica- EM, Shorr RI, et al. The effect of medication use on urinary tiebeoordeling initiatieven in Nederland kan beter. (Room for incontinence in community-dwelling elderly women. J Am Geri- improvement in medication review initiatives in the Netherlands). atr Soc. 2010;58(9):1715–20. Pharm Weekbl Sci. 2010;4(11/12):189–94. 18. Tinetti ME, Inouye SK, Gill TM, Doucette JT. Shared risk factors 28. Meulendijk MC, Spruit MR, Willeboordse F, Numans ME, Brink- for falls, incontinence, and functional dependence. Unifying the kemper S, Knol W, et al. Efficiency of clinical decision support approach to geriatric syndromes. JAMA. 1995;273(17):1348–53. systems improves with experience. J Med Syst. 2016;40(4):76. 19. Woolcott JC, Richardson KJ, Wiens MO, Patel B, Marin J, Khan 29. Sinnige J, Korevaar JC, van LJ, Westert GP, Schellevis FG, KM, et al. Meta-analysis of the impact of 9 medication classes on Braspenning JC. Medication management strategy for older falls in elderly persons. Arch Intern Med. 2009;169(21):1952–60. people with polypharmacy in general practice: a qualitative 20. Willeboordse F, Grundeken LH, van den Eijkel LP, Schellevis study on prescribing behaviour in primary care. Br J Gen Pract. FG, Elders PJ, Hugtenburg JG. Information on actual medication 2016;66(649):e540–51. use and drug-related problems in older patients: questionnaire or 30. Reeve E, Andrews JM, Wiese MD, Hendrix I, Roberts MS, Shakib interview? Int J Clin Pharm. 2016;38(2):380–7. S. Feasibility of a patient-centered deprescribing process to reduce inappropriate use of proton pump inhibitors. Ann Pharmacother. 21. Williams M, Peterson GM, Tenni PC, Bindoff IK, Stafford AC. 2015;49(1):29–38. DOCUMENT: a system for classifying drug-related problems in community pharmacy. Int J Clin Pharm. 2012;34(1):43–52. 1 3
International Journal of Clinical Pharmacy – Springer Journals
Published: Mar 20, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera