2024 Volume 47 Issue 6 Pages 1128-1135
Medication waste may be caused by medication oversupply; however, the degree of medication oversupply in Japan is unclear. This study aimed to quantify the degree of oversupply of chronic disease medications per patient, the proportion of oversupplied patients, and the excess days and costs of the oversupplied medications in Japan. This retrospective nationwide cohort study using a large insurance claims database from Japan was conducted in patients aged ≥55 years who received one or a combination of the following five classes of medications dispensed in FY 2019: third-generation calcium antagonists, angiotensin 2 receptor blockers, statins, dipeptidyl peptidase-4 inhibitors, and biguanides. Medications with the same ingredient having the same specification were treated as the same medication. Medication oversupply was defined as a medication possession ratio (MPR) during persistence >1.0. The proportions of oversupplied patients and excessively oversupplied patients with ≥30 excess days/year were approximately 16 and 1–2% for all drug classes, respectively. Three-quarters of the oversupplied patients had fewer excess day (≤14/year), and the median oversupplied medication cost was less than 1000 yen/year for all classes. However, there was a patient with oversupplied medication estimated as 983 excess days per year and a patient with oversupplied medication costs of nearly 90000 yen per year. Using the MPR and excess days as indicators, it is necessary to accelerate estimation of the oversupply per patient, as well as the development of patient intervention strategies and a national system to reduce medication oversupply.
Medication waste due to unused medications at home is a problem in many countries.1–3) The estimated cost of unused medications possessed by patients in Japan is enormous, at over tens of billions of yen annually.4) Poor patient adherence to medication protocols is a well-known cause of medication waste,1,3,5) but medication oversupply due to physician overprescription and early patient visits could also be factors.6,7) A study in Japan8) reported that multiple medications in the same class were dispensed to 8.8% of patients aged 0–70 years whose prescriptions were issued from multiple hospitals during 1 month of the study period.
However, no report has quantified the current status of oversupply of chronic disease medications in Japan. In addition, there has been little progress in determining how to estimate the degree of medication oversupply at the patient level and how to utilize these estimates to reduce oversupply in Japan. On the other hand, in other countries such as in the U.S., Sweden, and Thailand, this issue has been addressed in several studies.6,9–13) Those studies,6,9–11) under the assumption that medications in the same class are treated as the same medication, reported that approximately 20–50% of patients are oversupplied chronic disease medications.
This study aimed to quantify the degree of oversupply of third-generation calcium antagonists, angiotensin 2 receptor blockers, statins, dipeptidyl peptidase-4 inhibitors, and biguanides per patient, the proportion of oversupplied patients, and the excess days and costs of the oversupplied medications in patients aged ≥55 years in Japan. Ingredients with the same specification, not the same class, were treated as the same medication to avoid overestimation of oversupply, because we believe that it is first step for accurate estimation of oversupply.
This retrospective nationwide cohort study was conducted using the DeSC database created by DeSC Healthcare, Inc. for observational research purposes.14–16) The database includes claims from multiple insurance systems in Japan, namely Society-managed Health Insurance, National Health Insurance, and Medical Care System for the Elderly Aged ≥75 Years. The database is unique in that it contains a large number of claims from older patients, and it covered approximately 10 million people as of 2022, the time at which the data were provided to the authors.
The claims contain anonymized unique identifiers, sex, birth year and month, the type of insurance, type of claim (whether the claim was for medical, diagnosis procedure combination, dental, or prescription purposes and for an inpatient or outpatient), anonymized codes for hospitals, clinics, and pharmacies that issue claims, and medication information such as the name, dose, days supplied, and date dispensed.
This study was performed in accordance with the Act on the Protection of Personal Information, the Ethical Guidelines for Medical and Biological Research Involving Human Subjects, and the Data License Agreement of DeSC Healthcare, Inc. Ethics review was deemed unnecessary by the Ethics Committee of Tokushima Bunri University. Informed consent was waived because of the anonymized information in the claims data.
Study PopulationData were included from patients who (1) were in the database as of April 1, 2018 and had the same insurance coverage until at least September 30, 2019 (minimum coverage duration: 1.5 years), (2) were aged ≥55 years in April 2019, and (3) received one or a combination of the following five representative classes of chronic disease medications during the observation period in FY 2019: third-generation calcium antagonists (CABs; amlodipine besilate and azelnidipine), angiotensin 2 receptor blockers (ARBs; C09C Anatomical Therapeutic Chemical Classification), statins (STAs; C10AA), dipeptidyl peptidase-4 inhibitors (DPPs; A10BH), and biguanides (BIGs; A10BA). Considering starting age of these five classes of lifestyle disease medications, this study targeted patients aged ≥55 years. Patients prescribed medications for temporary or rescue use purposes were excluded because it was difficult to identify the dates the medications were used. We treated the same ingredients with the same specification as the same medication. Thus, patients prescribed in FY 2019 a medication with multiple ingredients or a medication with a single ingredient but with multiple specifications were also excluded. This is because patients prescribed medications with multiple ingredients may have had their medications switched, or patients prescribed ingredients with multiple specifications may have had their dosage increased; such cases are not considered oversupplied patients but rather medication oversupply. Additionally, for calculating the medication possession ratio (MPR) and the proportion of days covered (PDC), as described below, patients who had only one dispensing date during the study period were excluded.
Given that oversupply is identified by the MPR, which is calculated based on overlapping days of supply as described below, the above five drug classes were selected among antihypertensive, lipid-modifying, and oral diabetic agents because they are used prevalently and are typically taken once daily. The reason for evaluating only medications taken once daily is that if a twice-daily medication is prescribed separately for morning and evening use, the days of supply of the medication is doubled. As an exception, even though BIGs are used two to three times per day, they were selected in this study to confirm the results.
Identification of Medication OversupplyMedication oversupply was identified using the MPR.6,7,9,10,12,13,17) Oversupply was defined as an MPR >1.0.12,17) The algorithm for identifying medication oversupply and estimating excess days of oversupply is illustrated in Fig. 1. The calculations of the MPR and PDC used in the algorithm are shown in Fig. 2.
MPR, Medication possession ratio; Rx, Prescription; PDC, Proportion of days covered. a) If MPR <1, PDC = 1 is impossible.
MPR, Medication possession ratio; Rx, Prescription; PDC, Proportion of days covered.
The MPR and PDC are often used as measures of medication adherence in studies based on prescription and dispensing records.17,18) The MPR is based on the total days of supply during a period, and PDC is based on the total days that a medication is covered during a period (Fig. 2). Thus, the numerator of the PDC is equal to the overlapping period subtracted from the value of the numerator used to calculate MPR (see Fig. 2). This means that the maximum PDC is 1.0. Drug persistence was identified using the permissible gap method to calculate the MPR and PDC during persistence.19) Regarding this method, a therapy is regarded as persistent if the gap (duration from the dispensing date plus period of medication supply to the next dispensing date) is within the permissible gap, and as discontinued otherwise. The permissible gap was defined as the median number of days of medication supply and was calculated for each individual.20) The persistence period excluding the days of medication supply during the last dispensing period (period A) was used as the denominator for the MPR and PDC calculations,17,18) because the duration of use of the last medication dispensed could not be measured.
The left side of Fig. 1 shows an example of a case of medication oversupply at an MPR >1.0. If MPR >1.0, the medication is supplied for excess days, which are the days exceeding period A. If PDC = 1, the denominator and numerator in the PDC calculation are equal, meaning that no gap exists. If MPR >1.0 and PDC = 1, the excess days are the overlapping days. If the MPR >1.0 and PDC <1.0, a gap exists, but the overlapping days exceed the gap days; thus, the excess days are equal to the overlapping days minus the gap days under the assumption that patients used the medication during the gap. The right side of Fig. 1 shows an example of a case with an MPR ≤1.0 in which medications are not oversupplied; there are no excess days because there are no overlapping days, or the gap period is longer than or equal to the number of overlapping days even if there are overlapping days.
Statistical AnalysesThe MPR, PDC, and excess days during period A were calculated for each patient. The oversupplied medication costs associated with the excess days were also calculated for each patient using the price of the dispensed formulation. For example, if the brand-name formulation was switched to the generic formulation at a certain time point, the total costs were calculated using the prices and periods of each formulation. Since the length of period A varied among the patients, excess days and costs were calculated on a yearly basis using the following formula:
 |
The proportions of patients who were oversupplied (defined as MPR >1.0) and excessively oversupplied (defined by three different indicators: MPR >1.1, MPR >1.2, and ≥30 excess days/year) were calculated. The definition of medication oversupply in previous studies6,7,9–13,17,21) varies, in which different MPR cutoffs of 1.0, 1.1, and 1.2 have been used. We employed MPR cutoffs of 1.1 and 1.2 for excessive oversupply. The statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, U.S.A.).
A flowchart of the study population is shown in Fig. 3. After applying the inclusion and exclusion criteria, 608864, 625381, 672959, 245199, and 119628 patients were recruited as the cohorts for the CAB, ARB, STA, DPP, and BIG medications, respectively. The cohort characteristics are shown in Table 1. The proportion of males was higher in the DPP and BIG cohorts. The proportion of patients aged ≥75 years ranged from 30 to 55% depending on the drug class and sex, and it tended to be lower among males regardless of the drug class and lower in the BIG cohort compared with the other drug classes. In all classes, the median drug persistence in FY 2019 was approximately 330 d (25–75th percentile: approx. 280–340 d), and most patients had only outpatient prescription during persistence. The proportions of patients with MPR <0.8 and PDC <0.8, which are generally considered to indicate poor medication adherence,17,19) were approximately 10% both for MPR and PDC in all classes.
ATC, Anatomical Therapeutic Chemical Classification; DPC, diagnosis procedure combination; CAB, third-generation calcium blocker; ARB, angiotensin II receptor blocker; STA, statin; DPP, DPP-4 inhibitor; BIG, biguanid. a) Clinical service identification codes 21–28 mean that the medications were dispensed for medical, DPC, or dental claims. b) Patients dispensed multiple medications of different classes were counted in each class.
| CAB n = 608864 n (%) | ARB n = 625381 n (%) | STA n = 672959 n (%) | DPP n = 245199 n (%) | BIG n = 119628 n (%) | |
|---|---|---|---|---|---|
| Male | 272665 (44.8) | 297143 (47.5) | 252183 (37.5) | 135323 (55.2) | 66509 (55.6) |
| 55–64 yearsa) | 35489 (13.0) | 40143 (13.5) | 33233 (13.2) | 17349 (12.8) | 11270 (16.9) |
| 65–74 yearsa) | 123270 (45.2) | 138360 (46.6) | 119466 (47.4) | 64165 (47.4) | 36240 (54.5) |
| 7–84 yearsa) | 85073 (31.2) | 90698 (30.5) | 79192 (31.4) | 42900 (31.7) | 16323 (24.5) |
| ≥85 yearsa) | 28833 (10.6) | 27942 (9.4) | 20292 (8.0) | 10909 (8.1) | 2676 (4.0) |
| Female | 336199 (55.2) | 328238 (52.5) | 420776 (62.5) | 109876 (44.8) | 53119 (44.4) |
| 55–64 yearsb) | 29225 (8.7) | 31046 (9.5) | 47145 (11.2) | 10959 (10.0) | 7677 (14.5) |
| 65–74 yearsb) | 121295 (36.1) | 122946 (37.5) | 187227 (44.5) | 44859 (40.8) | 26864 (50.6) |
| 75–84 yearsb) | 114925 (34.2) | 111781 (34.1) | 133590 (31.7) | 37779 (34.4) | 14915 (28.1) |
| ≥85 yearsb) | 70754 (21.0) | 62465 (19.0) | 52814 (12.6) | 16279 (14.8) | 3663 (6.9) |
| Insurance | |||||
| Society-managed Health Insurance | 16859 (2.8) | 18780 (3.0) | 22778 (3.4) | 6494 (2.6) | 4254 (3.6) |
| National Health Insurance | 283726 (46.6) | 304143 (48.6) | 353557 (52.5) | 125494 (51.2) | 75721 (63.3) |
| Medical Care System for the Elderly Aged ≥75 Years | 308279 (50.6) | 302458 (48.4) | 296624 (44.1) | 113209 (46.2) | 39653 (33.1) |
| Median days of persistence (25–75th percentile) | 327 (287–339) | 328 (294–340) | 326 (290–339) | 327 (291–339) | 322 (280–337) |
| Inpatient/outpatient prescription during persistence | |||||
| Outpatient prescription only | 567958 (93.3) | 596228 (95.3) | 635402 (94.4) | 224983 (91.8) | 112262 (93.8) |
| Inpatient prescription only | 6951 (1.1) | 3906 (0.6) | 3097 (0.5) | 2201 (0.9) | 920 (0.8) |
| Both | 33955 (5.6) | 25247 (4.0) | 34460 (5.1) | 18015 (7.3) | 6446 (5.4) |
| Adherence during persistence | |||||
| Poor adherence (MPR <0.8) | 55984 (9.2) | 53238 (8.5) | 79042 (11.7) | 20455 (8.3) | 10531 (8.8) |
| Poor adherence (PDC <0.8) | 63719 (10.5) | 60157 (9.6) | 87069 (12.9) | 22730 (9.3) | 11779 (9.8) |
CAB, third-generation calcium blocker; ARB, angiotensin II receptor blocker; STA, statin; DPP, DPP-4 inhibitor; BIG, biguanid; MPR, medication possession ratio; PDC, proportion of days covered. a) Percentage of males. b) Percentage of females.
The proportion of oversupplied patients, defined as MPR >1.0, was approximately 16% for all drug classes (Table 2). The proportion of excessively oversupplied patients with an MPR >1.1, MPR >1.2, or ≥30 excess days/year was approximately 1, 0.5, or 1–2%, respectively. Of the five medication classes, BIGs were associated with the lowest proportion of oversupplied patients but the highest proportion of excessively oversupplied patients.
| CAB n = 608864 n (%) | ARB n = 625381 n (%) | STA n = 672959 n (%) | DPP n = 245199 n (%) | BIG n = 119628 n (%) | |
|---|---|---|---|---|---|
| Oversupply | |||||
| MPR >1.0 | 105201 (17.3) | 103133 (16.5) | 105386 (15.7) | 41927 (17.1) | 18108 (15.1) |
| Excessive oversupply | |||||
| MPR >1.1 | 7562 (1.2, 7.2a)) | 5757 (0.9, 5.6a)) | 5385 (0.8, 5.1a)) | 2603 (1.1, 6.2a)) | 1620 (1.4, 8.9a)) |
| MPR >1.2 | 2742 (0.5, 2.6a)) | 1751 (0.3, 1.7a)) | 1354 (0.2, 1.3a)) | 878 (0.4, 2.1a)) | 766 (0.6, 4.2a)) |
| ≥30 excess days/year | 10037 (1.6, 9.5a)) | 7880 (1.3, 7.6a)) | 7628 (1.1, 7.2a)) | 3524 (1.4, 8.4a)) | 2025 (1.7, 11.2a)) |
MPR, medication possession ratio; CAB, third-generation calcium blocker; ARB, angiotensin II receptor blocker; STA, statin; DPP, DPP-4 inhibitor; BIG, biguanid. a) Percentage of oversupplied patients.
The MPR, excess days, and oversupplied medication costs of the oversupplied patients are shown in Table 3, and the respective boxplots are shown in Fig. 4. Most of the oversupplied patients had an MPR of slightly more than 1.0 (median for all drug classes: 1.0) and few excess days (median: 6–7 d/year; 75th percentile: 13–14 d/year). However, the maximum excess days exceeded 500 d/year for all classes, and the highest value was 983 d/year, observed for ARB. The median oversupplied medication cost was lowest for CABs (134 yen/year) and STAs (148 yen/year) and highest for DPPs (856 yen/year). The maximum oversupplied medication cost was lowest for STAs (36998 yen/year) and highest for DPPs (87357 yen/year).
| CAB n = 105201 | ARB n = 103133 | STA n = 105386 | DPP n = 41927 | BIG n = 18108 | |
|---|---|---|---|---|---|
| MPR during persistence | |||||
| Median (25–75th percentile) | 1.0 (1.0–1.0) | 1.0 (1.0–1.0) | 1.0 (1.0–1.0) | 1.0 (1.0–1.0) | 1.0 (1.0–1.0) |
| Max | 2.6 | 3.7 | 2.7 | 2.7 | 3.0 |
| Excess days/year | |||||
| Median (25–75th percentile) | 7 (2–14) | 6 (2–13) | 6 (2–13) | 6 (2–13) | 7 (2–14) |
| Max | 580 | 983 | 619 | 630 | 736 |
| Oversupplied medication costs (yen/year) | |||||
| Median (25–75th percentile) | 134 (52–342) | 239 (93–649) | 148 (60–364) | 856 (325–1847) | 207 (82–582) |
| Max | 43914 | 73561 | 36998 | 87357 | 60871 |
CAB, third-generation calcium blocker; ARB, angiotensin II receptor blocker; STA, statin; DPP, DPP-4 inhibitor; BIG, biguanid.
MPR, medication possession ratio; CAB, third-generation calcium blocker; ARB, angiotensin II receptor blocker; STA, statin; DPP, DPP-4 inhibitor; BIG, biguanid.
This study quantified medication oversupply per patient for five representative classes (CABs, ARBs, STAs, DPPs, and BIGs) of chronic disease medications dispensed in Japan. This is the first estimation of medication oversupply using the MPR and excess days in Japan. In this study, which was designed to avoid overestimation of oversupply, the proportions of oversupplied patients and patients with more than one excess month/year were approximately 16 and 1–2% for all five representative drug classes. Half of the oversupplied patients had less than 7 excess days/year, and three quarters less than 14 d/year, for all five drug classes. Given that there were patients who could not visit a hospital as scheduled, these small numbers of excess days might be acceptable. The small numbers may be indicative of the proper implementation of insurance screening in the Japanese health insurance system. However, there were patients with 983 excess days/year and costs of approximately 90000 yen/year for oversupplied medications. Thus, the proportion of patients with disproportionate amounts of excess days and/or oversupplied medication costs was relatively small, but the total waste of all types of chronic disease medication across the nation could be substantial. It is hard to estimate amount of the totaled waste from the results in this study because this study was analyzed from only five classes of lifestyle disease medications.
Ichinowatari et al. interviewed patients at 44 community pharmacies in Japan about unused medications in their home and reported that the occurrence reasons of unused medications and the proportions were 43% for forgetting to take medication, 25% for patients’ own adjustment of medications and 9% for the gap between the visit date and the number of days prescribed.22) As the medication oversupply in this study is calculated from the number of overlapping days, it is caused from other factors such as duplicate prescriptions by multiple healthcare facilities, in addition to the gap between the visit date and the number of days prescribed. Thus, the proportion of oversupply in unused medications would exceed 9%, but this needs to be clarified in future studies.
A report in the U.S.21) found that publicly insured individuals were more likely to have oversupplied medications than were privately insured individuals. Under the national universal health insurance system in Japan, all insured individuals have access to any healthcare facility or pharmacy, irrespective of time or geographical region, which may contribute to medication oversupply. Therefore, it is necessary to clarify the triggers of medication oversupply and the characteristics of patients likely to be oversupplied. Additionally, developing tools that support physicians and pharmacists to thoroughly check prescription intervals and duplicate prescription could be useful to improve clinical practice. Furthermore, it is essential to develop strategies for patient interventions and to establish a national-level system to reduce medication oversupply. The indicators employed in this study, i.e., MPR and excess days, will be useful to identify patients at risk of medication oversupply and to evaluate the effectiveness of this new system. Kumamaru et al. reported that previously measured medication adherence predicted an individual’s future adherence.23) Their research focused on poor adherence, but future medication oversupply could also be predicted by previous measures.
We believe that the results of this study, which were analyzed for population with high proportion of elderly patients, were indicative of the current status of the medication oversupply for elderly in Japan, because the lifestyle disease medications are frequently used by the elderly. However, burden rate of healthcare copayment that patients pay at healthcare facilities or pharmacies varies according to types of insurers and income of insured persons in Japan. The burden rate of healthcare copayment may influence the proportion of oversupplied patients. In addition, most of this study population had only outpatient prescriptions during persistence and few had inpatient prescriptions. This study identified patients by defining the same ingredients with the same specifications as the same medications in order to avoid overestimation of oversupply. As a result, many patients with stable medical conditions who medication therapy did not change were included in this study and fewer patients had inpatient prescriptions. If the definition of the same medications is changed (specifically, including multiple ingredients or different specifications), many patients with inpatient prescriptions would be included. However, it would overestimate oversupply, because overlapping days occur due to ingredient switching and dose increase, which is regarded as oversupply. Further studies are needed to confirm the differences in estimate of oversupply when the definition of same medications is changed.
Many previous foreign studies examining the oversupply of chronic disease medications used a less stringent MPR of >1.2 to define medication oversupply.6,9–11,21) The proportion of oversupplied patients with an MPR >1.2 reported in those studies6,9–11) was 20–50%, which was much higher than the proportion (0.5%) reported in this study for all five drug classes. The primary reason for this discrepancy could be that this study treated the same ingredient with the same specifications as the same medication, while the previous studies treated the same drug class as the same medication.
Additionally, most previous studies of medication oversupply used only the MPR6,7,9,10) as an indicator to quantify the degree of medication oversupply, with an MPR cutoff of 1.2, applying the same 20% differential as an MPR cutoff of 0.8 to indicate poor medication adherence.7,9) This study employed the number of excess days in addition to the MPR as indicators for two reasons. First, the MPR depends on the length of period A (the denominator in the MPR calculation); for example, an MPR >1.2 during 1 year means that the number of excess days/year is >73 (365 × 1.2 − 365 = 73). In other words, as many as 73 d/year of medication remain on hand; i.e., the longer period A is, the greater the number of days allowed. Second, it is easier to understand the quantity of oversupply using the number of excess days than using MPR.
Some limitations exist in this study. First, it is possible that we overestimated oversupply, especially of medications such as BIGs, which are taken multiple times a day, because medications of that type are occasionally prescribed as separate medications rather than as multiple daily doses of the same medication (e.g., morning and nightly doses of the same drug are prescribed as two different medications rather than as twice daily doses of the same drug). Among the five drug classes, BIGs were associated with the lowest proportion of oversupplied patients (MPR >1.0) but with the highest proportion of excessively oversupplied patients (MPR = 1.1, MPR = 1.2, or 30 excess days). These results may be because the MPR increases if a medication taken multiple times a day is prescribed as different medications. Second, the five medication classes focused in this study are often concomitantly used. The similar values for patient proportion of oversupply and excessive oversupply between classes in this study may have been influenced by the concomitant use of the classes. Third, the quantification methods employed in this study are based on the assumption that patients took the prescribed medications during the gap period. If the patients did not take the medications during the gap period, the oversupply quantity would have been underestimated. Forth, the MPR defining medication oversupply was quantified according to overlapping days, and the calculation of MPR does not take the daily dose into consideration (Fig. 2). Therefore, an excessive daily dose was not included in the evaluation of oversupply in this study. Fifth, it is unclear whether patients actually take the medications in this study using claims data. Medication oversupply could occur if patients with poor medication adherence repeatedly visited to healthcare facilities, but oversupply due to medication poor adherence cannot be analyzed in this study.
Despite these limitations, this study quantified medication oversupply per patient using the MPR and excess days as indicators of oversupply. The proportions of oversupplied patients and patients with ≥30 excess days/year in Japan were approximately 16 and 1–2%, respectively, for all five drug classes. Most of the oversupplied patients had few excess days, but one patient had oversupplied medication estimated as 983 excess days per year, and one patient had oversupplied medication costs of nearly 90000 yen per year. Based on our results, it is necessary to accelerate development of patient intervention strategies and a national-level system to reduce medication oversupply.
We sincerely thank DeSC Healthcare, Inc. for providing the data for our study. We also thank CY for her assistance in processing the data.
The authors declare no conflict of interest.
