2023 Volume 46 Issue 4 Pages 592-598
In April 2014, sodium-glucose cotransporter 2 inhibitor (SGLT-2i) was introduced in Japan. In May 2015, the prescription limitation for SGLT-2i was lifted. Subsequently, SGLT-2i was shown to reduce cardiovascular events in patients with type 2 diabetes mellitus (T2DM). SGLT-2i prescription is expected to increase and consequently affect the prescription trends for other antidiabetic agents. Therefore, we evaluated the trends for antidiabetic agent prescriptions in Japan from April 2012 to March 2020. In this study, a dynamic cohort consisting of patients with T2DM derived from the Japan Medical Data Center health insurance database and with at least one antidiabetic agent prescription was investigated. The prescription rates were calculated monthly (/1000 person-months) for each class of antidiabetic agent. The eligible cohort comprised 34333 patients. The prescription rate for dipeptidyl peptidase-4 inhibitor increased from 424.0 in April 2012 to 656.3 in May 2015, and slightly decreased to 635.4 in March 2020. The prescription rate for biguanide consistently increased from 347.2 in April 2012 to 500.1 in March 2020. The prescription rate for sulfonylurea consistently decreased from 393.8 in April 2012 to 172.5 in March 2020. The prescription rate for SGLT-2i consistently increased from 4.1 in April 2014 to 363.1 in March 2020. SGLT-2i prescription increased and may affect the prescription trends for dipeptidyl peptidase-4 inhibitor and sulfonylurea after May 2015, when the prescription limitation for SGLT-2i was lifted. Biguanide prescriptions increased regardless of the introduction of SGLT-2i. The treatment of T2DM in Japan is clearly changing, with a focus on SGLT-2i and biguanide.
In 2009, dipeptidyl peptidase-4 inhibitor (DPP-4i) was introduced in Japan for the treatment of type 2 diabetes mellitus (T2DM). A previous study has shown that the prescription of DPP-4i immediately increased after the lifting of the prescription limitation for newly listed drugs (new drugs are limited to a maximum of 14 d per prescription during one year after the listing in Japan).1) The study suggested that prescribers preferred the profile of DPP-4i as it had a low risk for hypoglycemia, helped to maintain a steady weight, and preserved beta-cell function.
In the 2010s, two classes of antidiabetic agents, sodium-glucose cotransporter-2 inhibitor (SGLT-2i) and glucagon-like peptide-1 receptor agonist (GLP-1RA), were introduced. Both agents showed efficacy at reducing cardiovascular events in patients with T2DM when compared to a placebo in several clinical trials.2,3) For example, the EMPA-REG OUTCOME study reported the efficacy of SGLT-2i firstly in November 2015.4) In contrast, DPP-4i did not show a reduction in cardiovascular events when compared to the placebo.2,3) Recent guidelines have recommended the use of SGLT-2i and GLP-1RA as early lines of T2DM treatment in patients with cardiovascular or chronic kidney disease.5) Thus, the prescription rates of SGLT-2i and GLP-1RA were expected to increase, particularly the preference for SGLT-2i, as an oral antidiabetic, would impact the prescription rates of other oral antidiabetic agents. Consequently, the increase in prescriptions of both agents in accordance with relevant guidelines will reduce T2DM-associated complications including cardiovascular events.
Several studies have investigated the prescription trends of antidiabetic agents and showed that there is an increasing trend for the prescription of SGLT-2i.6–12) A study using a nationwide claims database covering >98% of claims in Japan showed increments in the prescription of SGLT-2i and biguanide (BG) and reductions in that of DPP-4i in patients receiving first-line treatment from 2014–2017.6) This information suggests that SGLT-2i and BG have gradually been prescribed as first-line treatments instead of DPP-4i. However, prescription trends after first-line treatment and how antidiabetic agents were combined during the 2010s remain unclear. Additionally, the interactions between the prescription trends for one class of antidiabetic agents with respect to the other agents are not sufficiently clear, as previous studies investigated the yearly prescription trends.
Determining whether the treatment for T2DM was changed to a strategy that is supported by relevant guidelines and accumulated evidence contributes to identifying issues in recent treatments. This would improve T2DM treatment at the population level. Therefore, in this study, we have evaluated prescription trends for every line and combination of antidiabetic agents for patients with T2DM between April 2012 and March 2020 to clarify the overall changes in treatments for T2DM in Japan. Moreover, we evaluated the impact of introducing SGLT-2i on the prescription trends of other antidiabetic agents using an interrupted time-series analysis.
The data used in this investigation were sourced from the Japan Medical Data Center health insurance database (JMDC Inc., Tokyo, Japan). This database consists of basic information (including birth year and month, sex, and enrollment period in the database), health checkup data, and claims data for 11065484 individuals <75 years old in Japan (as of January 2021). Diagnoses are coded using the International Classification of Disease, 10th revision (ICD-10), and prescribed drugs are coded using the WHO’s Anatomical Therapeutic Chemical classification system (WHO-ATC). Unique identifiers are allocated to these anonymized data to enable continuous observations. Further details of this database have been presented in previous reports.13,14)
In this investigation, 1.7 million individuals were randomly sampled from the database due to data availability and used to define a dynamic cohort that included eligible patients at the earliest time point after April 2012 and excluded them at the earliest time point of death, exclusion from the database, or March 31, 2020. The cohort included patients who had ICD-10 codes of E11 or E14 as diagnoses and were prescribed at least one class of antidiabetic agents. Patients who met the following criteria were excluded: <18 years old in April 2012; diagnoses related to type 1 diabetes mellitus; or an ICD-10 code of E13 as diagnosis.
VariablesInformation on patient characteristics was collected when included in the cohort. The list of the variables and their definitions are presented in Supplementary eTable 1 of Supplementary Materials. The present study calculated prescription rates for alpha-glucosidase inhibitor (GI), BG, DPP-4i, glinide, GLP-1RA, insulin, SGLT-2i, sulfonylurea (SU), and thiazolidinedione (TZD). The WHO-ATC codes for these drugs are presented in Supplementary eTable 1.
Statistical AnalysisPatient characteristics were summarized using medians and interquartile ranges for age, T2DM duration, and the number of classes of prescribed antidiabetic agents, and frequencies and percentages for other variables.
The prescription rates (/1000 person-months) were calculated as follows: 1) prescription days for each class of antidiabetic agent were determined. For oral drugs, prescription days were based on the regulated dosage and administration, while for injection drugs, prescription days were calculated by dividing the amount of medicinal ingredients by the minimum dose per day. If the prescription days, based on the regulated dosage and administration, were longer than the intervals for the visiting prescribers when the same drugs were prescribed on the next visit, the prescription days were then replaced with the intervals of visiting prescribers; 2) we calculated the number of individual patients prescribed target drugs on at least one day in the month based on the calculated prescription days as the prescription frequency; 3) we calculated the total number of individual patients included in the cohort on at least one day in the month as the number at risk of prescription; 4) the prescription rates (/1000 person-months) were calculated using the following formula:
 |
where t indicates the t th month between April 2012 and March 2020. These indices indicate the rates at which 1000 individuals were observed in one month. The prescription rates for each class of antidiabetic agent were calculated from the overall population and by subgroups into which the cohort was divided based on the number of classes of prescribed antidiabetic agents (one, two, three, or ≥4). We defined these subgroups by considering changes in the number of antidiabetic agents between the observational periods for each patient. In each subgroup, the relative proportions (%) for the antidiabetic agents and combinations of these agents whose prescription rates were within the top ten in March 2020 were calculated.
To evaluate the impacts of introducing SGLT-2i on the prescription trends of other antidiabetic agents, interrupted time-series analysis was conducted.15) This study design is used to evaluate the impacts of interventions introduced at a population level on outcomes including drug utilization over time. The impacts of interventions are evaluated based on whether a trend of an outcome in a time series has changed at a known time point. The present study considered two interventions in addition to the base trend for each drug (start from April 2012), namely, the trend change after the listing of SGLT-2i (start from April 2014) and the trend change after the lifting of the prescription limitation for SGLT-2i (start from May 2015). The incidence rate ratios (IRRs) and 95% confidence intervals (CIs) for the base trend and trend changes after the listing of SGLT-2i and after the lifting of its prescription limitation were estimated using quasi-Poisson regression models (link function; ln: distribution; quasi-Poisson). The constructed model was as follows:
 |
where T(t) represents the t th month between April 2012 (coded 0) and March 2020 (coded 95), T1 represents March 2014, one month before SGLT-2i was listed (coded 23), X1t is a dummy variable indicating the pre-listing of SGLT-2i (coded 0 before April 2014) or the post-listing of SGLT-2i (coded 1 on and after April 2014), T2 represents April 2015, one month before the prescription limitation for SGLT-2i was lifted (coded 36), X2t is a dummy variable indicating the pre-lifting of the prescription limitation for SGLT-2i (coded 0 before May 2015) or the post-lifting of the prescription limitation for SGLT-2i (coded 1 on and after May 2015), Yt is the prescription frequency in the t th month, and ln(Nt) is an offset variable to convert frequency into rate and the number at risk of prescription on a natural log scale in the t th month. Thus, β0 indicates the prescription rate on a natural log scale at T = 0 (April 2012), β1 is the IRR on a natural log scale for the base trends indicating the change in the prescription rate per month between the observational periods, β2 is the IRR on a natural log scale indicating the slope change on and after April 2014 by month due to the listing of SGLT-2i, and β3 is the IRR on a natural log scale indicating the slope change on and after May 2015 by month due to the lifting of the prescription limitation for SGLT-2i. The models were applied to the overall population and subgroups, respectively. If the Durbin–Watson test detected autocorrelations in the residuals, 95% CIs were adjusted according to Newey–West standard errors.
All statistical analyses were performed using R for Windows, version 4.1.2 (The R Foundation for Statistical Computing, Vienna, Austria).
Ethics Approval and Consent to ParticipateThe present study was reviewed and approved by the Ethics Committee of the National Institute of Public Health (Approval No. NIPH-IBRA#12315) and implemented under the “Ethical Guidelines for Medical and Biological Research Involving Human Subjects” in Japan. The need for informed consent was waived by the Ethics Committee of the National Institute of Public Health as the study used a commercial database, which had accumulated administrative claims data.
The eligible cohort included a total of 34333 patients, and their characteristics are shown in Table 1. At inclusion to the cohort, the median age was 56 years, 30.1% of the patients were female, and 1.1% had a history of cardiovascular disease.
| Variable | Overall (N = 34333) |
|---|---|
| Age, years | 56 (49, 62) |
| ≥65 | 5195 (15.1) |
| Sex (female) | 10334 (30.1) |
| T2DM duration, years | 3.0 (0.6, 6.5) |
| Chronic complications in DM | 8184 (23.8) |
| Number of antidiabetic agents prescribed | 2 (1, 2) |
| Concomitant drugs | |
| CCB | 12836 (37.4) |
| ACEi | 1552 (4.5) |
| ARB | 13162 (38.3) |
| Aldosterone receptor antagonists | 671 (2.0) |
| Other diuretics | 2951 (8.6) |
| Beta-blockers | 3264 (9.5) |
| Dyslipidemia agents | 16713 (48.7) |
| Antiplatelet agents | 3107 (9.0) |
| Anticoagulants | 755 (2.2) |
| History of cardiovascular disease | 372 (1.1) |
| PCI or CABG | 268 (0.8) |
| Coronary artery disease | 155 (0.5) |
| Stroke | 108 (0.3) |
| History of heart failure | 2928 (8.5) |
Abbreviations. ACEi: angiotensin-converting enzyme inhibitors; ARB: angiotensin receptor blockers; CABG: coronary artery bypass grafting; CCB: calcium channel blocker; DM: diabetes mellitus; PCI: percutaneous coronary intervention; T2DM: type 2 diabetes mellitus. The age, T2DM duration, and number of antidiabetic agents prescribed are presented as medians and interquartile ranges, and the other variables are presented as frequencies and percentages (%).
The prescription trends for the antidiabetic agents in the overall population and by subgroup based on the number of classes of prescribed antidiabetic agents, respectively, are shown in Figs. 1 and 2. The values for the monthly prescription rates are shown in Supplementary eTables 2–6 of Supplementary Materials.
Abbreviations. BG: biguanide; DPP-4i: dipeptidyl peptidase-4 inhibitor; GI: glucosidase inhibitor; GLP-1RA: glucagon-like peptide-1 receptor agonist; SGLT-2i: sodium-glucose cotransporter-2 inhibitor; SU: sulfonylurea; TZD: thiazolidinedione. The dashed vertical lines indicate the following time points: April 2014, listing of SGLT-2i and May 2015, lifting of the prescription limitation for SGLT-2i.
Abbreviations. BG: biguanide; DPP-4i: dipeptidyl peptidase-4 inhibitor; GI: glucosidase inhibitor; GLP-1RA: glucagon-like peptide-1 receptor agonist; SGLT-2i: sodium-glucose cotransporter-2 inhibitor; SU: sulfonylurea; TZD: thiazolidinedione. Each panel indicates the results of the different subgroups based on the number of classes of prescribed antidiabetic agents. The dashed vertical lines indicate the following time points: April 2014, listing of SGLT-2i and May 2015, lifting of the prescription limitation for SGLT-2i.
The prescription rate of SGLT-2i increased from 4.1 in April 2014 to 363.1 in March 2020 through the trend change around May 2015. From April 2014 to March 2020, the prescription rates for SGLT-2i increased in all subgroups: one class, 0.5 to 184.4; two classes, 1.6 to 328.8; three classes, 5.8 to 564.4; and ≥4 classes, 25.8 to 777.5.
The prescription rate of BG increased from 347.2 in April 2012 to 500.1 in March 2020. From April 2012 to March 2020, increasing prescription trends for BG were observed in all subgroups: one class, 153.9 to 206.6; two classes, 387.1 to 583.8; three classes, 679.5 to 785.1; and ≥4 classes, 855.6 to 912.8.
The prescription rate of DPP-4i increased until around May 2015, after which it was considered stable or slightly decreased. Its prescription rates were 424.0 in April 2012, 656.3 in May 2015, and 635.4 in March 2020. For the different subgroups, the prescription rates for DPP-4i in April 2012, May 2015, and March 2020 were as follows: one class, 339.1, 558.5, and 489.5; two classes, 479.8, 737.2, and 760.0; three classes, 619.7, 849.2, and 860.6; and ≥4 classes, 791.4, 903.7, and 849.8.
The prescription rates for SU decreased from 393.8 in April 2012 to 172.5 in March 2020 in the overall population. From April 2012 to March 2020, decreasing prescription trends for SU were observed in all subgroups, as follows: one class, 171.2 to 23.9; two classes, 485.2 to 105.6; three classes, 754.3 to 302.3; and ≥4 classes, 844.9 to 547.7.
The prescription rates for alpha-GI and TZD decreased from April 2012 to March 2020 in the overall population, and similar prescription trends were observed in all subgroups. The prescription rates of glinide, insulin, and GLP-1RA were stable or slightly increased from April 2012 to March 2020 in the overall population, and similar prescription trends were observed in all subgroups.
Relative Proportions for Antidiabetic Agents and CombinationsThe changes in the relative proportions for the antidiabetic agents and the combinations of these agents by subgroup are shown in Fig. 3. The values for the frequencies and relative monthly proportions are shown in Supplementary eTables 7–10 of Supplementary Materials.
Abbreviations. BG: biguanide; DPP-4i: dipeptidyl peptidase-4 inhibitor; GI: glucosidase inhibitor; GLP-1RA: glucagon-like peptide-1 receptor agonist; SGLT-2i: sodium-glucose cotransporter-2 inhibitor; SU: sulfonylurea; TZD: thiazolidinedione. Each panel indicates the results of the different subgroups based on the number of classes of prescribed antidiabetic agents.
In the subgroup prescribed one class of antidiabetic agent, SGLT-2i was the third most prescribed (18.5%) at the end of the observational period. In the subgroup prescribed two classes, SGLT-2i was the second most prescribed in combination with DPP-4i (18.6%) and the third most prescribed with BG (10.6%) at the end of the observational period. In the subgroup prescribed three classes, SGLT-2i was most prescribed in combination with BG and DPP-4i (36.3%), the fourth most with DPP-4i and SU (5.6%), and the seventh most with BG and GLP-1RA (2.1%) at the end of the observational period. In the subgroup prescribed ≥4 classes, eight of the top ten combinations at the end of the observational period included SGLT-2i.
Impact of SGLT-2i IntroductionThe IRRs and 95% CIs of the interrupted time-series analyses in the overall population and by subgroup are shown in Table 2. The plots of observed and estimated prescription trends for the antidiabetic agents are shown in Supplementary eFigs. 2–6 of Supplementary Materials. Regarding the prescription rate for DPP-4i in the overall population, the IRR (95% CI) of the lifting of the prescription limitation for SGLT-2i was 0.993 (0.987 to 0.999), suggesting that previous prescription trends for DPP-4i changed to 0.993 times per month after the lifting of the prescription limitation for SGLT-2i. For the subgroups, the prescription trends for DPP-4i changed to be stable or slightly decreased after both the listing of SGLT-2i and the lifting of its prescription limitation. For the other antidiabetic agents, there were some statistically significant trends, but clear changes could not be observed visually (Supplementary eFigs. 2–6 of Supplementary Materials).
| Population | Class | Base trend | Trend change after listing SGLT-2i | Trend change after lifting prescription limitation for SGLT-2i |
|---|---|---|---|---|
| Overall eligible patients | alpha-GI | 0.990 (0.989 to 0.991) | 1.005 (1.002 to 1.009) | 0.997 (0.993 to 1.001) |
| BG | 1.004 (1.002 to 1.006) | 1.002 (0.996 to 1.008) | 0.997 (0.990 to 1.004) | |
| DPP-4i | 1.014 (1.011 to 1.017) | 0.992 (0.984 to 1.0003) | 0.993 (0.987 to 0.999) | |
| Glinide | 1.005 (0.993 to 1.016) | 1.012 (0.974 to 1.052) | 0.983 (0.947 to 1.020) | |
| GLP-1RA | 0.996 (0.982 to 1.011) | 1.032 (0.997 to 1.067) | 0.990 (0.967 to 1.014) | |
| Insulin | 1.000 (0.999 to 1.001) | 1.005 (0.999 to 1.011) | 0.993 (0.987 to 0.9998) | |
| SU | 0.991 (0.989 to 0.994) | 1.002 (0.993 to 1.011) | 0.997 (0.987 to 1.007) | |
| TZD | 0.995 (0.993 to 0.996) | 0.995 (0.993 to 0.998) | 1.000 (0.998 to 1.002) | |
| Subgroup prescribed one class of antidiabetic agents | alpha-GI | 0.985 (0.983 to 0.987) | 0.991 (0.968 to 1.014) | 1.013 (0.980 to 1.047) |
| BG | 1.001 (0.999 to 1.004) | 1.003 (0.997 to 1.008) | 0.999 (0.995 to 1.003) | |
| DPP-4i | 1.014 (1.011 to 1.018) | 0.997 (0.988 to 1.006) | 0.986 (0.979 to 0.994) | |
| Glinide | 0.977 (0.973 to 0.981) | 1.024 (1.010 to 1.037) | 0.987 (0.976 to 0.998) | |
| GLP-1RA | 0.943 (0.921 to 0.966) | 1.010 (0.946 to 1.077) | 1.070 (1.023 to 1.119) | |
| Insulin | 1.001 (0.988 to 1.013) | 0.974 (0.949 to 1.0003) | 1.022 (1.002 to 1.042) | |
| SU | 0.975 (0.971 to 0.979) | 1.010 (1.001 to 1.020) | 0.995 (0.989 to 1.001) | |
| TZD | 0.987 (0.985 to 0.989) | 0.988 (0.983 to 0.994) | 1.009 (1.004 to 1.015) | |
| Subgroup prescribed two classes of antidiabetic agents | alpha-GI | 0.980 (0.978 to 0.983) | 1.012 (1.007 to 1.016) | 0.993 (0.990 to 0.997) |
| BG | 1.003 (1.002 to 1.004) | 1.009 (1.003 to 1.016) | 0.992 (0.985 to 0.998) | |
| DPP-4i | 1.017 (1.015 to 1.018) | 0.989 (0.982 to 0.996) | 0.995 (0.987 to 1.003) | |
| Glinide | 1.006 (1.0003 to 1.011) | 0.987 (0.972 to 1.003) | 0.999 (0.983 to 1.015) | |
| GLP-1RA | 1.004 (0.997 to 1.010) | 0.980 (0.965 to 0.995) | 1.030 (1.015 to 1.045) | |
| Insulin | 0.991 (0.987 to 0.995) | 1.011 (1.0003 to 1.022) | 0.991 (0.981 to 1.001) | |
| SU | 0.989 (0.984 to 0.993) | 1.000 (0.992 to 1.008) | 0.992 (0.987 to 0.996) | |
| TZD | 0.989 (0.987 to 0.991) | 1.003 (0.998 to 1.008) | 0.993 (0.988 to 0.998) | |
| Subgroup prescribed three classes of antidiabetic agents | alpha-GI | 0.992 (0.989 to 0.994) | 1.008 (0.995 to 1.022) | 0.989 (0.974 to 1.004) |
| BG | 1.003 (1.001 to 1.004) | 0.997 (0.994 to 0.9997) | 1.002 (1.001 to 1.004) | |
| DPP-4i | 1.012 (1.008 to 1.017) | 0.989 (0.980 to 0.997) | 0.999 (0.995 to 1.004) | |
| Glinide | 1.006 (0.998 to 1.013) | 1.021 (0.997 to 1.044) | 0.969 (0.949 to 0.990) | |
| GLP-1RA | 1.005 (0.987 to 1.024) | 1.021 (0.980 to 1.063) | 0.987 (0.962 to 1.011) | |
| Insulin | 0.996 (0.991 to 1.0003) | 1.013 (1.005 to 1.022) | 0.988 (0.981 to 0.994) | |
| SU | 0.991 (0.991 to 0.992) | 1.006 (1.004 to 1.009) | 0.991 (0.989 to 0.993) | |
| TZD | 0.991 (0.989 to 0.993) | 0.996 (0.991 to 1.001) | 1.001 (0.997 to 1.004) | |
| Subgroup prescribed ≥4 classes of antidiabetic agents | alpha-GI | 0.995 (0.994 to 0.997) | 1.002 (0.998 to 1.007) | 0.996 (0.992 to 1.001) |
| BG | 1.003 (1.002 to 1.004) | 0.994 (0.992 to 0.996) | 1.004 (1.002 to 1.006) | |
| DPP-4i | 1.006 (1.005 to 1.007) | 0.994 (0.992 to 0.995) | 1.000 (0.998 to 1.001) | |
| Glinide | 1.015 (1.004 to 1.026) | 1.012 (0.989 to 1.035) | 0.976 (0.963 to 0.989) | |
| GLP-1RA | 1.008 (0.761 to 1.336) | 1.072 (0.229 to 5.008) | 0.943 (0.174 to 5.120) | |
| Insulin | 1.007 (0.997 to 1.016) | 0.996 (0.980 to 1.012) | 0.995 (0.986 to 1.004) | |
| SU | 0.996 (0.995 to 0.997) | 0.995 (0.992 to 0.998) | 1.005 (1.001 to 1.008) | |
| TZD | 0.997 (0.995 to 0.999) | 0.992 (0.986 to 0.997) | 1.001 (0.996 to 1.007) |
Abbreviations. BG: biguanide; DPP-4i: dipeptidyl peptidase-4 inhibitor; GI: glucosidase inhibitor; GLP-1RA: glucagon-like peptide-1 receptor agonist; SU: sulfonylurea; TZD: thiazolidinedione. Each cell value indicates the incidence rate ratio (95% confidence interval).
The results of our analysis show that DPP-4i was the most prescribed antidiabetic agent at the end of the observational period in the overall population and all subgroups, except for the group prescribed ≥4 classes of antidiabetic agents. The prescription rates of DPP-4i were stable in the subgroups prescribed two or three classes or decreased in the subgroups prescribed one or ≥4 classes after approximately May 2015, when the prescription limitation for SGLT-2i was lifted. The prescription rates of SGLT-2i rapidly increased in the overall population and all subgroups through the positive slope change in the prescription trends around May 2015. Moreover, the interrupted time-series analyses showed that either the listing of SGLT-2i or the lifting of its prescription limitation changed the prescription trends for DPP-4i in the overall population and all subgroups to the stable or decreasing trends. Thus, SGLT-2i may impact the prescription trends for DPP-4i.
However, the direct replacement of DPP-4i with SGLT-2i was likely to only proceed in the subgroup prescribed one class of antidiabetics. The extent of the prescription increases for SGLT-2i was larger in the subgroups prescribed more classes of antidiabetic agents. SGLT-2i was majorly used in combination with DPP-4i and/or BG. In contrast to SGLT-2i, the prescription rates of alpha-GI, SU, and TZD decreased in the overall population and all subgroups over the entire observational period. The relative proportions of the combinations including these agents also decreased. These results suggest that SGLT-2i directly replaced alpha-GI, SU, and TZD, which were chosen at later lines for the treatment of T2DM after its listing or the lifting of its prescription limitation. That is, alpha-GI, SU, and TZD were replaced with DPP-4i until the introduction of SGLT-2i, after which these were replaced with SGLT-2i. Moreover, DPP-4i may have been directly replaced with SGLT-2i in the subgroup prescribed one class because the prescription rates for DPP-4i decreased in this subgroup. The increased prescription of SGLT-2i was probably attributed to these replacements.
A decrease in the prescription rates for DPP-4i was also observed in the subgroup prescribed ≥4 classes. However, DPP-4i was used in combination with SGLT-2i rather than being replaced by SGLT-2i. In this subgroup, the prescription rates for GLP-1RA were relatively high and seemed to be inversely related to the rates of DPP-4i. Prescribers may have switched from DPP-4i to GLP-1RA as both agents affect the glucagon-like peptide-1 receptor. Furthermore, the stable or slightly increasing prescription trends for GLP-1RA indicated that this agent did not replace DPP-4i or other antidiabetic agents in the earlier lines of treatment of T2DM, even though GLP-1RA has been shown to have efficacy in reducing cardiovascular events.2,3)
The prescription rates of BG increased in the overall population and all subgroups. Relevant guidelines recommend that BG should be considered as a first-line treatment due to its low cost and efficacy, which includes the possibility of preventing cardiovascular events.5,16) Thus, the increase in the prescription rates of BG was considered a favorable trend, which is expected to continue.
This study suggests that treatments for T2DM in Japan between April 2012 and March 2020 were changing with a focus on SGLT-2i and BG.
The increase in the prescription rates of SGLT-2i and BG suggests that prescribers may be emphasizing the efficacy of antidiabetic agents for reducing cardiovascular events. A study reported that the number of certified cardiologists was positively correlated with the increase in prescription of SGLT-2is.7) Another study reported that the publication of clinical trials evaluating the possible efficacy of SGLT-2is in reducing cardiovascular events increased the prescription of SGLT-2is for patients with a history of cardiovascular disease more compared to those without.17) These studies showed that prescribers chose antidiabetic agents to reduce cardiovascular events in accordance with the relevant guidelines and accumulated evidence.2–5,15) Additionally, SGLT-2i showed other various efficacies related to renal protection, weight loss, and prevention of heart failure.5) These favorable profiles could also lead to an increase in its prescription rates. Particularly, in the subgroup prescribed one class, the prescription rates for SGLT-2i exceed those of BG within several years as the newest guidelines indicate that SGLT-2i should be considered a first-line treatment.5) However, the recommendation is limited to patients with a history or risk factors of cardiovascular disease and patients with a history of heart failure or chronic kidney disease. SGLT-2i should be selected based on patient characteristics because it was reported that SGLT-2i was prescribed for patients in which obtaining the effects of reducing cardiovascular events was uncertain.17) Additionally, the drug costs of SGLT-2i were still higher than those of other oral antidiabetic agents, and thus, some patients may prefer lower drug costs. Prescribers may need to select antidiabetic agents based on not only patient characteristics but also their preferences.
The present study clarified the prescription trends for antidiabetic agents in Japan between April 2012 and March 2020, but there were several limitations due to data availability. First, this study could not investigate the prescription trends in patients aged ≥75 years because the Japan Medical Data Center health insurance database does not cover them. A previous study reported that prescribers tended to avoid prescribing SGLT-2i for elderly patients due to concerns about safety.6) Thus, the prescription trends for antidiabetic agents that were affected by age could change when the elderly are included. Second, although the same study reported that there were regional differences in prescription trends,6) we could not estimate these impacts, as regional information was unavailable. The results of this study could thus not reflect the overall prescription trends in Japan. Third, the sample size was too small to stably estimate the prescription trends for some of the antidiabetic agents, including glinide, GLP-1RA, and insulin. For injection drugs, the prescription days were overestimated since this study partly calculated those based on the amount of medicinal ingredients by the minimum dose per day. Further, we did not investigate the differences in prescription trends between patients naïve to T2DM treatment and those already prescribed T2DM treatment since the sample size for the naïve patients was not sufficiently obtained. This study focused on prescription trends obtained, rather than distinguishing them.
However, the present study can have depicted current trends as the prescription trends of one class in the present study were similar to those in the previous study based on a nationwide claims database covering >98% of the claims in Japan.6) Future studies are required to investigate the prescription trends after March 2020 rather than the trends between the same periods in this study. In the 2020s, additional indications for heart failure and chronic kidney disease were approved for SGLT-2i and the oral GLP-1RA, namely semaglutide, was listed in Japan. It is thus crucial that we continue to evaluate prescription trends considering these factors.
While DPP-4i was still the most prescribed antidiabetic agent, the prescription of SGLT-2i was found to increase rapidly, and that of BG increased gradually for every line of T2DM treatment in Japan between April 2012 and March 2020. The increased use of SGLT-2i and BG will likely be in accordance with the relevant guidelines and accumulated evidence, and consequently, will reduce T2DM-associated complications, including cardiovascular disease, at the population level. Prescribers should continue to focus on SGLT-2i and BG, but also select antidiabetic agents in accordance with patients’ characteristics and preferences in cases where SGLT-2i or BG is not suitable.
We would like to thank Dr. Keiko Konomura for useful discussions. This work was supported by JSPS KAKENHI Grant No. JP20K16102. The funders had no role in this work.
RI and SI both contributed to planning the study design, analyzing, and interpreting the data, and writing the manuscript. All authors have read and approved the final manuscript.
The authors declare no conflict of interest.
This article contains supplementary materials.
