2026 年 8 巻 1 号 p. 18-27
BACKGROUND
Polypharmacy management represents a public health concern; however, intervention by community pharmacists remains limited. Therefore, we evaluated the effectiveness of community pharmacy-based educational intervention for improving polypharmacy management using interrupted time series analysis (ITSA).
METHODS
We conducted the ITSA of an educational intervention implemented from June to September 2021 across pharmacies under Medical System Network Co., Ltd., Japan. We analyzed pharmacy claims data from April 2020 to May 2024, with no missing values. The intervention included lectures and workshops delivered via cascade training. The primary and secondary outcomes were the number of Fee for Medication Adjustment Support 2 (FMAS2) and Fee for Medication Adjustment Support 1 (FMAS1) per 100,000 patients, respectively. FMAS2 is claimed when pharmacists propose reducing medications for patients taking six or more medications from multiple institutions, and FMAS1 is claimed when the proposal leads to actual reduction.
RESULTS
The study included 404 pharmacies with an average of 569,909 patients per month. After intervention, FMAS2 claims immediately increased by 3.82 per 100,000 people (95% confidence interval (CI): 1.23 to 6.41, p = 0.005), with no significant trend change. FMAS1 claims first increased by 1.29 per 100,000 people (95% CI: 0.49 to 2.09, p = 0.002) but decreased thereafter (95% CI: −0.298 to −0.130, p < 0.001).
CONCLUSIONS
Educational intervention increased FMAS2 and FMAS1 claims in the short term. However, the clinical impact of this intervention was small, and its long-term effectiveness is limited. Future efforts should focus on developing sustainable programs and follow-up systems.
Owing to aging population trends, polypharmacy has become a global public health challenge1). Polypharmacy is associated with various health and socioeconomic issues, including rising healthcare costs, a risk of adverse events due to drug interactions, and decreased patient adherence2–5). Community pharmacists play a crucial role in addressing polypharmacy issues as they are optimally positioned to continuously monitor patient medication status within community healthcare settings and improve prescription practices.
A recent amendment of the medical fee system in Japan included revisions to dispensing fees, such as the implementation of the “Fee for Medication Adjustment Support” (FMAS) in April 2018, which was designed as a countermeasure to polypharmacy6). This additional incentive aims to reduce unnecessary medication use, particularly among elderly patients, by encouraging community pharmacists to actively engage in deprescribing initiatives. However, the implementation of polypharmacy management measures by community pharmacists remains insufficient. Moreover, successful approaches to promoting their implementation have not been clearly established. Currently, the number of FMAS claims remains low7), indicating limited engagement with polypharmacy management among community pharmacists. Furthermore, a lack of knowledge and skills among pharmacists has been cited as a barrier to deprescribing initiatives8); however, no studies have evaluated the impact of educational interventions specifically designed to overcome these challenges or established optimal educational methods. Despite reports of beneficial educational interventions for pharmacists in various fields9–11), most studies have only verified the short-term outcomes immediately after intervention, highlighting the need for longer-term studies.
Therefore, in this study, we aimed to evaluate the outcomes of a “polypharmacy reduction project” (PRP) conducted for pharmacies affiliated with Medical System Network Co., Ltd., Japan, using interrupted time series analysis (ITSA)12). The objective was to elucidate the effectiveness of the PRP for reducing polypharmacy and inform future educational interventions. The rationale was enabling a quantitative assessment of both the short- and long-term effects of intervention via ITSA, obtaining insights into promising educational methods for implementing polypharmacy management.
The aim of the PRP was to identify patients with potential polypharmacy, propose medication reduction, and reduce the number of medications. The project was conducted from June to September 2021 across six pharmacy operating companies affiliated with Medical System Network Co., Ltd.
Training was conducted in three steps. In the first step, in June 2021, pharmacists belonging to each company’s educational division, along with pharmacists serving as district managers responsible for overseeing and managing pharmacies within their assigned areas, were trained. The training consisted of lectures and workshops. The lectures covered the necessity of polypharmacy countermeasures, communication skills in polypharmacy, and how to write a proposal for medication reduction for physicians. This proposed system, established under the Japanese insurance framework, allowed pharmacists to earn fees by identifying cases of polypharmacy and submitting medication-reduction proposals for physicians. The workshop’s component included practical exercises for drafting proposals.
In the second step, in July 2021, the same set of lectures was delivered to the managers of each pharmacy, and the same workshops from the first step were conducted with pharmacists from each company’s educational division and district managers as facilitators.
In the final step, during August–September 2021, lectures were delivered to all pharmacists at each pharmacy, and workshops were conducted with the managers of each pharmacy as facilitators. Post-training assignments were assigned at each step to ensure that all pharmacists had received the training (Fig. 1).
Data were obtained from two databases: one containing information on patients who visited pharmacies affiliated with Medical System Network Co., Ltd., and the other containing information on pharmacists working at these pharmacies, covering the period from April 2020 to May 2024. These data were collected from pharmacies affiliated with Medical System Network Co., Ltd., a company that operates pharmacies across Japan. Pharmacies use specialized electronic medical billing systems designed to generate claims and manage insurance reimbursements. These systems were used to collect the required data throughout the study period.
Patient data consisted of claim data from each pharmacy, including patient age, sex, whether prescriptions involved six or more medications, whether additional fees such as FMAS were claimed, and the number of pharmacies included in the analysis. These data were based on standardized health insurance claims required for reimbursement in Japan, which include only mandatory information. Therefore, there were no missing values. Pharmacist data were derived from internal company records and were also complete at the time of extraction. Thus, imputation was not required. In addition, the age and sex of pharmacists were collected to assess changes in workforce composition before and after the implementation of the project.
The study was conducted in accordance with the Declaration of Helsinki, and the use of data was approved by the Ethics Committee of the Hokkaido Institute for Pharmacy Benefit Co., Ltd. (2024029). Based on the judgment of the ethics committee, patient consent was waived because aggregated claim data without personal information were used.
STUDY PARTICIPANTSAll patients who visited the pharmacies recorded in the database between April 2020 and May 2024 were included, and no exclusion criteria were applied. As we were unable to access information on patients who were prescribed six or more oral medications from multiple medical facilities, we included the entire patient population visiting pharmacies in our analysis.
OUTCOME MEASURESThe primary outcome was the number of FMAS2 claims per 100,000 patients visiting the target pharmacies. The secondary outcome was the number of FMAS1 claims per 100,000 patients visiting the target pharmacies. FMAS is a reimbursement that can be claimed by community pharmacists in Japan when they engage in efforts to address issues such as duplicate prescriptions. FMAS2 can be claimed once every three months when a pharmacist submits a written proposal to prescribing physicians regarding the potential for unnecessary medications in patients prescribed six or more oral medications from multiple medical facilities13). Conversely, FMAS1 can be claimed monthly when a pharmacist submits a written proposal to a prescribing physician that leads to the discontinuation of two or more oral medications for a period of at least four weeks13). The numerator for these outcomes was the total number of FMAS claims, and the denominator was the total number of patients visiting pharmacies. We selected FMAS2 as the primary outcome because the aim was to measure behavioral changes among pharmacists resulting from educational intervention. We considered this approach appropriate for evaluating the pure effect of education, as deprescribing also involves patient and physician decision-making, as well as pharmacist behavior.
STUDY DESIGN AND STATISTICAL ANALYSISAn ITSA design was used to control for secular trends in the study results. This approach measures whether PRP mediated an immediate change in the level or existing trend (slope) of the outcome. In this analysis, the level change represents short-term effects observed immediately after the intervention, whereas the slope change indicates gradual or long-term effects observed over time after the intervention.
The time series was divided into three segments: pre-intervention (April 2020 to May 2021), “phase-in” (June 2021 to September 2021), and post-intervention (October 2021 to May 2024). The pre-intervention segment comprised the outcomes before the intervention. The period from June 2021 to September 2021 was considered a “phase-in” period, where the effect of the PRP was assessed by comparing the trend in the pre-intervention period with that in the post-intervention period. The post-intervention segment comprised post-intervention outcomes. The Durbin-Watson test was used to assess autocorrelation; upon confirming autocorrelation, the Prais-Winsten regression model was used14). This method uses generalized least squares to estimate the parameters of a linear regression model in which the errors are assumed to follow a first-order autoregressive process15).
As part of the sensitivity analyses, the number of FMAS2 claims per 100,000 patients visiting pharmacies without a phase-in period was analyzed to assess the immediate effect of the intervention and validate the appropriateness of the phase-in period used in the primary analysis. Additionally, we analyzed the number of FMAS2 claims per 100,000 patients taking six or more oral medications in a single prescription. Furthermore, we analyzed the fee for inhalation guidance as a negative control. The fee for inhalation guidance is claimed once every three months when a pharmacist instructs a patient with asthma or chronic obstructive pulmonary disease on inhalation techniques using documents and practice inhalers under the instruction of a physician, confirms that the patient is using inhaled medications correctly, and submits a report to the prescribing physician13). This outcome was selected because it was considered to be unaffected by the PRP. All statistical analyses were performed using Stata/BE software (version 18.0; StataCorp LLC, USA).
During the study period, 28,495,496 patients were analyzed. The study cohort included an average of 569,909 patients per month from 404 pharmacies across Japan. The average number of patients in the pre-intervention period was 527,851; this increased modestly to 590,771 in the post-intervention period. The mean age of patients decreased slightly from 57.5 (standard deviation: 0.85) years in the pre-intervention period to 55.9 (standard deviation: 0.72) years in the post-intervention period. The number of pharmacies increased by an average of 15 from the pre-intervention to the post-intervention period. No other pharmacy characteristics showed substantial changes between the two periods (Table 1, Fig. 2).
| All | Pre intervention Apr 2020 to May 2021 |
During intervention June 2021 to Sep 2021 |
Post intervention Oct 2021 to May 2024 |
|
|---|---|---|---|---|
| Months | 50 | 14 | 4 | 32 |
| Number of patients per month, Mean (SD) | 569,909 (40,568) | 527,851 (30,224) | 550,231 (8,626) | 590,771 (30,215) |
| Age, years, Mean (SD) | 56.4 (1.01) | 57.5 (0.85) | 56.8 (0.50) | 55.9 (0.72) |
| Female, n (%) | 321,449 (56.4) | 297,174 (56.3) | 311,961 (56.7) | 333,255 (56.4) |
| Percentage of patients receiving six or more oral medications, Mean (SD)a) | 13.7 (0.35) | 14.0 (0.49) | 13.8 (0.08) | 13.6 (0.25) |
| Number of Community Pharmacies per month, Mean (SD)b) | 404 (8.03) | 395 (0.99) | 396 (1.50) | 410 (4.78) |
| Region of Community Pharmacies, Mean (SD)b) | ||||
| Hokkaido | 123 (1.3) | 122 (1.3) | 121 (0.6) | 123 (1.0) |
| Tohoku | 24 (1.0) | 22 (0.5) | 24 (0.5) | 24 (0.5) |
| Kanto | 98 (5.4) | 91 (1.3) | 94 (0.8) | 102 (3.1) |
| Chubu | 26 (0.3) | 26 (0.5) | 26 (0.0) | 26 (0.2) |
| Kinki | 72 (1.6) | 74 (0.6) | 73 (0.0) | 71 (0.7) |
| Chugoku | 15 (0.0) | 15 (0.0) | 15 (0.0) | 15 (0.0) |
| Shikoku | 6 (0.3) | 6 (0.0) | 6 (0.0) | 6 (0.3) |
| Kyusyu, Okinawa | 41 (2.7) | 38 (0.5) | 38 (0.0) | 42 (2.0) |
| Number of pharmacists per month, Mean (SD)c) | 1,905 (36.1) | 1,868 (29.0) | 1,925 (6.4) | 1,919 (28.8) |
| Pharmacists Age, years, Mean (SD)c) | 39.6 (11.6) | 39.4 (11.8) | 39.2 (11.7) | 39.7 (11.6) |
| Pharmacists, Female, n (%)c) | 1,311 (68.8) | 1,285 (68.8) | 1,334 (69.3) | 1,319 (68.7) |
a) “Patients receiving six or more oral medications” refers to patients prescribed six or more oral medications in a single prescription.
b) “Community Pharmacies” refers to all pharmacies included in the study.
c) “Pharmacists” refers to all pharmacists employed by the pharmacy operating companies included in the study.
SD: standard deviation
Before the intervention, the number of FMAS2 claims gradually increased. After the intervention, the number of FMAS2 claims showed an immediate increase. The level change was 3.82 per 100,000 people (95% confidence interval (CI): 1.23 to 6.41, p = 0.005), indicating a significant short-term increase. The slope change was −0.205 per 100,000 people per month (95% CI: −0.466 to 0.056, p = 0.120), indicating no significant long-term change after intervention (Table 2, Fig. 3).
| Level Change | Slope Change | |||
|---|---|---|---|---|
| Coefficient (95% CI) | p-Value | Coefficient (95% CI) | p-Value | |
| primary outcome | ||||
| Fee for Medication Adjustment Support 2 per 100,000 patients | 3.82 (1.23 to 6.41) | 0.005 | −0.205 (−0.466 to 0.0557) | 0.120 |
| secondary outcome | ||||
| Fee for Medication Adjustment Support 1 per 100,000 patients | 1.29 (0.490 to 2.09) | 0.002 | −0.214 (−0.298 to −0.130) | <0.001 |
| negative control | ||||
| Fee for Inhalation Guidance per 100,000 patients | −23.4 (−88.5 to 41.7) | 0.473 | −1.14 (−7.63 to 5.34) | 0.725 |
| sensitivity analysis | ||||
| Fee for Medication Adjustment Support 2 per 100,000 patients receiving six or more oral medicationsa) | 28.7 (9.68 to 47.7) | 0.004 | −1.46 (−3.37 to 0.460) | 0.133 |
| Fee for Medication Adjustment Support 2 per 100,000 patients without phase-in period | 1.92 (−0.813 to 4.66) | 0.164 | −0.259 (−0.570 to 0.0530) | 0.102 |
a) “Patients receiving six or more oral medications” refers to patients prescribed six or more oral medications in a single prescription.
ITSA: interrupted time series analysis, CI: confidence interval
A significant increase in FMAS1 claims was also observed immediately after intervention, with a level change of 1.29 per 100,000 people (95% CI: 0.49 to 2.09, p = 0.002). Moreover, the slope change after intervention was −0.214 per 100,000 people per month (95% CI: −0.298 to −0.130, p < 0.001), indicating a significant decrease in the long-term trend after intervention (Table 2, Fig. 4).
Regarding the number of claims for the fee for inhalation guidance, we observed no significant changes in the level change (−23.4 per 100,000 people, 95% CI: −88.5 to 41.7, p = 0.473) or slope change (−1.14 per 100,000 people per month, 95% CI: −7.63 to 5.34, p = 0.725) (Table 2, Supplementary Fig. 1).
Sensitivity analysesIn the sensitivity analysis limited to patients prescribed six or more oral medications, we observed a larger level change (28.7 per 100,000 people, 95% CI: 9.68 to 47.7, p = 0.004) for FMAS2 claims, whereas the slope change (−1.46 per 100,000 people per month, 95% CI: −3.37 to 0.460, p = 0.133) showed no significant difference (Table 2, Supplementary Fig. 2). When excluding the phase-in period, we observed no significant changes in either the level change (1.92 per 100,000 people, 95% CI: −0.813 to 4.66, p = 0.164) or slope change (−0.259 per 100,000 people per month, 95% CI: −0.570 to 0.0530, p = 0.102) (Table 2, Supplementary Fig. 3).
In this study, we demonstrated that educational intervention for community pharmacists aimed at reducing polypharmacy immediately increased both the FMAS2 and FMAS1 claims. Although the intervention had no effect on the long-term trend of the FMAS2 claims, the FMAS1 claims showed a decreasing trend. These findings were consistent across multiple sensitivity analyses, and the use of negative controls enhanced the validity of our results. These findings suggest that strengthening polypharmacy management strategies for community pharmacists exhibited short-term effects. However, maintaining long-term effectiveness remains challenging.
This study provides new insights into the impact of educational interventions among community pharmacists on polypharmacy reduction. Although the effectiveness of pharmacist education has been reported in various fields9),10),16), to our knowledge, this is the first study to demonstrate the effects of educational interventions that specifically target polypharmacy reduction. Moreover, previous studies of polypharmacy intervention models have been limited to small-scale interventions17). Therefore, by analyzing a large pharmacy chain network, our study provides more generalizable results. Furthermore, the time series study design enabled a detailed analysis of temporal changes in intervention effects. Previous reports have been limited to the immediate effects of pharmacist-led polypharmacy interventions17),18), whereas our study tracked intervention impacts over an extended period, revealing declining effectiveness and providing important insights into their long-term impact.
The immediate increase in the number of FMAS claims after intervention can be primarily attributed to enhanced pharmacist awareness and practical skill acquisition following educational intervention. The training program highlighted the importance of appropriate medication adjustments for patients taking multiple medications and providing specific implementation guidance. Moreover, the PRP likely helped mitigate previously identified challenges to polypharmacy management related to poor knowledge and skills among pharmacists8). Previous research suggests that workshops are more effective than lectures alone19), which supports the efficacy of our educational intervention, incorporating both lectures and workshops; however, we did not compare different educational formats in this study. Therefore, future studies should also determine optimal educational methods.
Several factors may explain the lack of long-term effects and decreasing trends. First, a single educational intervention may not have sustained long-term effects as motivation and application of new knowledge tend to decline over time20). Achieving sustained effects requires regular follow-up training and organizational support systems to maintain motivation. Second, the number of patients eligible for FMAS2 and FMAS1 claims may have decreased. Educational intervention may have led pharmacists to intervene more actively for patients at high risk of polypharmacy, resulting in medication-reduction proposals and implementation during the initial intervention phase. Third, pharmacists may struggle to find the time to implement FMAS1 claims. FMAS1 claims require reducing patient prescriptions by two types of medications and demand complex efforts, including building consensus with patients and prescribing physicians while considering patient safety and efficacy. Along with the considerable time and effort involved in implementing claims, community pharmacists face increasing workloads because of expanding patient care duties21), further reducing the time available for FMAS1 claims. Fourth, the annual turnover rate of pharmacists in the group companies included in this study is approximately 7%, suggesting that approximately 20% of trained pharmacists may have left their positions during the analysis period. This turnover may have contributed to the observed diminishing long-term impact. Additionally, newly hired pharmacists did not receive training equivalent to those who participated in the initial PRP. Implementing comparable training for newly hired pharmacists could help maintain intervention sustainability.
From multiple perspectives, the impact of our intervention warrants careful consideration. From an individual patient perspective, benefits such as reduced adverse events, lower out-of-pocket expenses, and improved medication adherence might be expected; however, the overall effect across the entire patient population appears limited. From the perspective of healthcare providers, while pharmacist awareness increased, actual prescription adjustments remained minimal, with only a small fraction of eligible patients receiving medication reviews or reductions. This restricted effect suggests that educational interventions alone may be insufficient to address the complex challenge of polypharmacy effectively.
Whether the observed FMAS2 and FMAS1 improvements are sufficient remains debatable, even for patients taking six or more medications. We propose several possible explanations for these limited results. First, optimal treatment for conditions such as heart failure, rheumatoid arthritis, cancer, and diabetes requires multiple medications22–26), rendering medication reduction clinically inappropriate in some cases. This highlights the need for screening tools to identify patients who are eligible for medication reduction. Second, patient-related barriers, including medication-reduction anxiety27) and limited trust in pharmacists compared to that in physicians28),29), may hinder intervention effectiveness. Third, time constraints in community pharmacy settings may limit the ability of pharmacists to thoroughly engage in medication-reduction activities30). Therefore, future interventions may need to consider additional factors beyond simply improving pharmacist education.
Our educational intervention shows minor short-term effects, but these are not clinically significant. Additionally, since the long-term effects do not persist, the intervention should not be conducted as a one-time event. Based on these findings, there is a need to restructure the educational interventions for pharmacists.
Our study had several limitations. First, slight changes in patient characteristics before and after intervention may have affected the results. To address the gradual increase in monthly patient numbers from 2020 to 2024, we used the number of FMAS1 and FMAS2 claims per 100,000 patients visiting the target pharmacies as the study outcomes. The mean patient age decreased from 57.5 years (pre-intervention) to 55.9 years (post-intervention), suggesting a potential reduction in the number of patients who could be charged additional medical fees. This may have led to an underestimation of intervention effects. Second, owing to data aggregation constraints, we allowed new pharmacies to enter the study during the observation period. However, this increase was limited to an average of 15 pharmacies between pre-intervention and post-intervention periods and led to only gradual increases in patient numbers. Although this change may explain the slope change, it does not account for the level change in our findings. Third, we were unable to accurately identify patients prescribed six or more oral medications at multiple medical facilities. In our sensitivity analysis, patients prescribed six or more medications in a single prescription were used as the denominator. However, this approach has inherent limitations in terms of accuracy, as it does not account for medications that may have been prescribed by multiple medical facilities. The consistency between our primary and sensitivity analyses suggests that this limitation did not affect our conclusions. Fourth, detailed prescription data were unavailable, making it impossible to identify specific medication types. Fifth, other existing initiatives or interventions may not have been fully captured. Although we confirmed no implementation of company-wide training programs other than the PRP before, during, or after the intervention period, we could not track unofficial training activities at individual pharmacies. Sixth, our study period included the COVID-19 pandemic, which may have influenced the results. For example, the pandemic may have reduced patient-pharmacist communication time needed for FMAS interventions and increased healthcare burdens, potentially limiting pharmacists’ information sharing with physicians. However, we observed no major changes in the characteristics of patients visiting the pharmacies. Seventh, the fee for inhalation guidance used by us as our negative control is not ideal for comparison with FMAS2 due to differences in both target populations and claim volumes, but no other suitable control options were available. However, the results still support our primary findings.
In this study, we demonstrated that a multistep educational intervention for community pharmacists increased the number of FMAS claims intended as incentives for polypharmacy management. These findings suggest that pharmacist education may contribute to the short-term implementation of polypharmacy strategies. However, the clinical impact of this intervention was small, and the long-term effectiveness of this intervention method is limited. Future efforts should focus on developing educational programs and follow-up systems to ensure sustained interventional effects.
The authors declare no conflicts of interest related to this manuscript.
This research received no specific grants from any funding agency in the public, commercial, or not-for-profit sectors.
We are deeply grateful to Professor Shunichi Fukuhara of the Kyoto University Graduate School of Medicine for his valuable comments on this research.
We would like to thank Editage (www.editage.jp) for English language editing.
SS and RY conceived this study. SS and TS collected the data. SS, RY, and TK analyzed the data. SS and RY drafted the manuscript. All authors have read and approved the final manuscript.
