Signal Detection of Acute Renal Failure Following the Use of SGLT-2 Inhibitors: Stratified Analysis and Time Trend Analysis in Japan and the United States

Abstract

The results of previous studies that have used databases to investigate the associations between sodium-glucose-cotransporter-2 (SGLT-2) inhibitors and acute renal failure (ARF) have differed, and the impact of biases such as the Weber effect and stimulated reporting has not been fully examined. This study aimed to determine the associations between SGLT-2 inhibitors and ARF using signal detection, the effects on signals of regulatory agency alerts for ARF, and the publication of prominent studies by measuring changes in signals over time. Data registered in the Food and Drug Administration’s Adverse Event Reporting System from January 2013 to March 2020 were downloaded, signals were detected, and reporting odds ratios (RORs) were calculated for each country of occurrence (Japan/the United States). Quarterly changes in the number of reports and RORs were examined. Although an association between SGLT-2 inhibitor use and ARF was suggested in the United States, this study did not suggest such an association in Japan. The number of reports and RORs fluctuated when regulatory alerts and prominent studies were published, and events affecting the number of reports and RORs varied by country. This study revealed the difference in the associations between SGLT-2i and ARF in Japan and the United States. Additionally, the signal was identified to be influenced by alerts and the publication of studies. Therefore, these results should be interpreted cautiously as there could be a possibility of overestimation due to alert biases and publication of studies.

INTRODUCTION

Sodium-glucose-cotransporter-2 inhibitors (SGLT-2is) are relatively new drugs, and their mechanism of action involves the promotion of urinary excretion of glucose by inhibiting SGLT-2-mediated reabsorption of glucose in the proximal convoluted tubule of the kidney. Regarding the safety of SGLT-2i, there are different views on the association of the regulatory response and clinical trials in each country with the development of acute renal failure (ARF). An example of a regulatory response for ARF is the warning issued by the U.S. Food and Drug Administration (FDA) on the onset of ARF in June 2016.¹⁾ In contrast, there is only a revision of labels on dehydration in the revision of precautions issued by the Pharmaceuticals and Medical Devices Agency (PMDA) in January 2015 and the alert on dehydration-related events issued by the Japanese Diabetes Society in May 2016 and August 2019.^2–5) Representative clinical trials, such as EMPA-REG OUTCOME (EMPA-REG), CANVAS Program (CANVAS), DECLARE-TIMI 58 (DECLARE-TIMI), and CREDENCE, all show negative views on the association between ARF and the use of SGLT-2i.^6–11)

Signal detection is a method of detecting signals. It provides information about the possibility of an association between a drug and an adverse event from spontaneous reports, which have not been identified until now. Reporting odds ratio (ROR), an indicator of signal detection, is calculated by combining two factors, “presence or absence of a specific drug” and “presence or absence of a specific adverse event.”¹²⁾ Previous studies using the FDA Adverse Event Reporting System (FAERS)¹³⁾ showed positive results regarding the association between the use of SGLT-2i and adverse events.¹⁴⁾ In contrast, a study conducted by the authors using the Japanese Adverse Drug Event Report¹⁵⁾ database did not suggest an association with ARF.¹⁶⁾ Another study conducted by the authors with the latest FAERS data to examine this difference in trends suggested the influence of the nephroprotective effect of the combination of angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers.¹⁷⁾ However, many cases were excluded from the analysis because of unconfirmed age and sex. Moreover, the influence of bias, such as the Weber effect and stimulated reporting, has not been sufficiently examined in this previous study as well as in a study using a spontaneous reporting database. The Weber effect refers to the phenomenon in which more adverse events are reported within the first two years after a drug is launched.¹⁸⁾ A relatively large number of non-serious adverse events were reported during this period, consequently hampering the detection of signals for serious adverse events of interest.¹⁹⁾ Stimulated reporting can lead to an increasing number of reports and reporting odds ratios (RORs) by drawing attention to specific adverse events generated from alerts, academic studies, and media articles.^20,21) The effects of these biases on other drugs and adverse events have been investigated by assessing changes in ROR over time,²²⁾ but whether ARF can be caused by SGLT-2i has yet to be determined.

This study aimed to clarify how the regulatory agencies’ alerts for ARF and the publication of prominent studies affect signals for a more accurate interpretation of signals; we determine the relationship between SGLT-2i and ARF by detecting signals using the FAERS.

MATERIALS AND METHODS

We downloaded and analyzed data published by the FDA registered in the FAERS between January 2013 and March 2020. As in the previous study,¹⁷⁾ adverse events and pharmaceuticals were extracted by the preferred term listed in the Medical Dictionary for Regulatory Activities (MedDRA) Version 23.0. ARF was defined as 50 types of events identified as a broad term in the standardized MedDRA query. SGLT-2i is a product of the name canagliflozin, dapagliflozin, empagliflozin, ertugliflozin, ipragliflozin, luseogliflozin, and tofogliflozin, or their generic name. In this study, primary suspect drugs and secondary suspect drugs were analyzed, and cases in which the abovementioned drug name was included in “DRUGNAME” or “PROD_AI” were defined as SGLT-2i cases. “DRUG NAME” and “PROD_AI” include the name of the medical product and the active ingredient, respectively. Moreover, to reduce the impact of duplicate reports, only the latest reports were analyzed based on the FDA recommended method.²³⁾ First, the ROR was calculated by dividing the diabetes treatment cases into cases whose country of occurrence was the U.S.A. and cases from Japan. A logistic regression model was used to determine the development of ARF using SGLT-2i, and then the ROR and 95% confidence interval (CI) were calculated. The presence of a signal was determined when the lower limit of the 95% CI >1. Diabetes treatment cases are cases in which the product name or generic name of a glucose-lowering drug currently approved in the U.S. or Japan is entered in “DRUG NAME” or “PROD_AI.” Subsequently, we calculated the number of quarterly reports of all adverse events and ARF, except for cases with unknown date reports. We also calculated the quarterly ROR and quarterly cumulative ROR for the ARF. IBM SPSS Statistics 26/27/28 (IBM Corp., Armonk, NY, U.S.A.) was used for this study.

RESULTS

Association between the Use of SGLT-2i and ARF

18581 cases of adverse events occurred in Japan, and 432366 cases of adverse events occurred in the U.S. (Fig. 1). Of the cases that used SGLT-2i, 33922 (71.57%) occurred in the U.S. and 2446 (11.10%) in Japan. Table 1 lists the other patient attributes.

Fig. 1. Data Extraction Flow Chart

Table 1. Demographic Distribution of Reports

Diabetes treatment cases	with SGLT-2is (N = 36368)		without SGLT-2is (N = 414579)		p-Value
	N	(%)	N	(%)
Age
Under 70 years old	15750	(43.31)	164589	(39.70)	<0.001
Over 70 years old	2870	(7.89)	68209	(16.45)	<0.001
Unknown	17748	(48.80)	181781	(43.85)	<0.001
Sex
Female	16091	(44.24)	206601	(49.83)	<0.001
Male	16063	(44.17)	166409	(40.14)	<0.001
Unknown	4214	(11.59)	41569	(10.03)	<0.001
U.S.A.	33922	(93.27)	398444	(96.11)	<0.001
Japan	2446	(6.73)	16135	(3.89)	<0.001
Outcomea)
DE	715	(1.97)	26486	(6.39)	<0.001
LT	844	(2.32)	6274	(1.51)	<0.001
HO	10792	(29.67)	96964	(23.39)	<0.001
DS	1154	(3.17)	5646	(1.36)	<0.001
CA	6	(0.02)	186	(0.04)	0.012
RI	61	(0.17)	168	(0.04)	<0.001
OT	8997	(24.74)	135682	(32.73)	<0.001
Unknown	18180	(49.99)	216000	(52.10)	<0.001
Reporter’s occupation
MD	11145	(31.45)	69746	(17.46)	<0.001
PH	1147	(3.24)	21999	(5.51)	<0.001
OT	4314	(12.17)	36683	(9.18)	<0.001
LW	36	(0.10)	8800	(2.20)	<0.001
CN	18631	(52.57)	259665	(64.99)	<0.001
Unknown	1095	(3.01)	17686	(4.27)	<0.001
Other glucose-lowering drugsb)
Sulfonylureas	2792	(7.68)	51996	(12.54)	<0.001
Biguanide	8786	(24.16)	120320	(29.02)	<0.001
α-Glucosidase inhibitors	349	(0.96)	3234	(0.78)	<0.001
Thiazolidinediones	655	(1.80)	65175	(15.72)	<0.001
Glinide class	182	(0.50)	2881	(0.69)	<0.001
DPP4i	3971	(10.92)	46496	(11.22)	0.086
GLP1 receptor agonists	2300	(6.32)	74305	(17.92)	<0.001
Insulin	2933	(8.06)	156733	(37.81)	<0.001

SGLT-2is: Sodium-glucose-cotransporter-2 inhibitors, DE: Death, LT: Life-Threatening, HO: Hospitalization – Initital or Prolonged, DS: Disability, CA: Congenital Anomaly, RI: Required Intervention to Prevent Permanent Impairment/Damage, OT: Other Serious, MD: Physician, PH: Pharmacist, OT: Other health-professional, CN: Consumer, DPP4i: Dipeptidyl Peptidase-4 inhibitors, GLP1: glucagon-like peptide-1. a) Some patients were reported more than 1 outcome. b) Some patients were prescribed more than 1 glucose-lowering drugs.

As a result of stratified analysis of the country of occurrence, when the country was Japan, there was a negative association with ARF across all the SGLT-2is. In contrast, when the country of occurrence was in the U.S., the results suggested an association with ARF (Table 2). Given the influence of the approval status in each country, it was not possible to calculate the ROR for the areas where the number of data points was 0.

Table 2. Reporting Odds Ratio of Acute Renal Failure with SGLT-2 Inhibitors by Diabetes Treatment Cases in Japan and U.S.A

	Diabetes treatment cases in Japan			Diabetes treatment cases in U.S.A.
	Total number of cases (n)	Acute renal failure (n)	ROR (95%CI)	Total number of cases (n)	Acute renal failure (n)	ROR (95% CI)
All types of SGLT-2is	2446	106	0.52 (0.43–0.64)	33922	3381	3.50 (3.36–3.64)
Canagliflozina)	1012	33	0.40 (0.28–0.57)	17005	2774	5.09 (4.87–5.31)
Dapagliflozina)	672	37	0.71 (0.51– 0.99)	4910	252	1.38 (1.21–1.57)
Empagliflozina)	623	18	0.36 (0.22–0.57)	8277	360	1.17 (1.05–1.30)
Ertugliflozina)	0	0	–	427	17	1.06 (0.66–1.73)
Ipragliflozina)	98	11	1.56 (0.83–2.93)	0	0	–
Luseogliflozina)	17	3	2.64 (0.76–9.20)	0	0	–
Tofogliflozina)	31	4	1.82 (0.64–5.22)	0	0	–
No SGLT-2is	16135	1291	–	398444	12216	–

SGLT-2is: Sodium-glucose-cotransporter-2 inhibitors, ROR: Reporting Odds Ratio, CI: confidence interval. a) Some patients were prescribed more than 1 SGLT2i during their course of treatment.

Influence of Alerts and Publication of Notable Studies on Number of Reports and ROR

In Japan, the number of reports of all adverse events and ARF tended to increase in Q1 2015, Q2–Q4 2016, and Q3 2018–Q4 2019. Focusing on all adverse events, a relatively strong increasing trend was observed with the number of reports doubling between Q2 2015 and Q3 2016 (Fig. 2). In the U.S., the number of reports of all adverse events and ARF tended to increase in Q1 2015–Q3 2015 and Q1 2017–Q3 2018. In Q2 2015, a sharp increase was noted due to an increase in the number of reports. (Fig. 3).

Fig. 2. The Number of Reported Adverse Events Related to the Use of SGLT-2 Inhibitors in Japan

PMDA: Pharmaceuticals and Medical Devices Agency, JDS: The Japanese Diabetes Society. ^a The revision of the label issued by the PMDA in January 2015.^{2,3) b} The alert issued by the Japan Diabetes Society in May 2016.^{4) c} Publication of studies related to the EMPA-REG OUTCOME Trial in July 2016.^{10) d} Publication of studies related to CANVAS Program in September 2018.^{7) e} Publication of studies related to the DECLARE-TIMI 58 trial in January 2019.^{8) f} Publication of studies related to CREDENCE Trial in April 2019.^{6) g} The alert issued by the Japan Diabetes Society in August 2019.⁵⁾

Fig. 3. The Number of Reported Adverse Events Related to the Use of SGLT-2 Inhibitors in the United States

FDA: The US Food and Drug Administration. ^a The alert issued by the FDA in June 2016.^{1) b} Publication of studies related to CANVAS Program in June 2017.^{11) c} Publication of studies related to EMPA-REG-OUTCOME in January 2018.^{9) d} Publication of studies related to CANVAS Program in September 2018.⁷⁾

As a result of analyzing the transition of ROR over time in Japan in both the quarterly analysis and quarterly cumulative analysis, a decreasing trend was observed in Q2 2015–Q2 2016, and an increasing trend was observed in Q2–Q4 2016 and Q3 2018–Q3 2019. In the quarterly analysis, an increasing trend was observed in Q2 2017–Q3 2018, but in the cumulative analysis, there was a slight decreasing or flattening trend (Fig. 4, Supplementary Table 1). In the U.S., both quarterly and quarterly cumulative analyses showed a decrease in Q2 2015 and an increase in Q3 2017 and Q3 2018. In the quarterly analysis, an increasing trend was observed in Q2–Q3 2016, but no clear change was observed in the cumulative analysis (Fig. 5, Supplementary Table 1).

Fig. 4. The Transition of Reporting Odds Ratio over Time in Japan

JDS: The Japanese Diabetes Society, ROR: Reporting Odds Ratio, CI: confidence interval. ^a The revision of the label issued by the PMDA in January 2015.^{2,3) b} The alert issued by the Japan Diabetes Society in May 2016.^{4) c} Publication of studies related to the EMPA-REG OUTCOME Trial in July 2016.^{10) d} Publication of studies related to CANVAS Program in June 2017.^{11) e} Publication of studies related to CANVAS Program in September 2018.^{7) f} Publication of studies related to the DECLARE-TIMI 58 trial in January 2019.^{8) g} Publication of studies related to CREDENCE Trial in April 2019.^{6) h} The alert issued by the Japan Diabetes Society in August 2019.⁵⁾

Fig. 5. The Transition of Reporting Odds Ratio over Time in the United States

FDA: The US Food and Drug Administration, ROR: Reporting Odds Ratio, CI: confidence interval. ^a The alert issued by the FDA in June 2016.^{1) b} Publication of studies related to the EMPA-REG OUTCOME Trial in July 2016.^{10) c} Publication of studies related to CANVAS Program in June 2017.^{11) d} Publication of studies related to CANVAS Program in September 2018.⁷⁾

DISCUSSION

The most crucial finding in this study is that the signal is substantially influenced not only by alerts but also by the publication of prominent studies. It was suggested that even if the study suggested a negative perspective on adverse events, it might help increase the ROR, and the impact could vary depending on the study design. These results provide helpful insights for a more accurate interpretation of signals from previous studies and also the current study.

The results from the U.S. suggested that all SGLT-2i were associated with ARF in patients treated for diabetes, with the exception of ertugliflozin, which has a limited number of patients, but we found no such association in Japan. However, it should be noted that, in this study, drug combinations that might reduce renal function, such as non-steroidal anti-inflammatory drugs, furosemide may have affected the ROR.^24–27) We believe that a combination of differences in patient background information, such as laboratory values and concomitant medications, as well as biases inherent in spontaneous reporting databases, led to the difference in the association between the use of SGLT-2i and the occurrence of ARF in Japan and the U.S. In order to accurately interpret the signals calculated in this study, including the differences in trends in the stratified analysis, we aim to discuss here our observations on the time trend analysis. The results of the time trends analysis of the number of cases and ROR in Japan and the U.S. showed fluctuations that could be attributed to stimulated reporting and the Weber effect; however, the types of alerts and publications which affected the signals were different in each country.

First, fluctuations that could be considered the Weber effect were observed from Q2 2015–Q3 2016 in Japan and Q1 2015–Q3 2015 in the U.S. The Weber effect in Japan did not show a sharp peak as confirmed in the cases reported from the U.S. We believe that the reason for this continuous increase is that the restriction on the prescription days for each SGLT-2i in Japan had been gradually lifted after one year on the market. Especially in Japan, for the first year after the launch of a new drug, the number of prescription days per prescription was limited to 14 d. The first SGLT-2i to lift this restriction was ipragliflozin in May 2015. Since then, in 2015, the restrictions on the number of prescription days have been lifted for many SGLT-2i, and its dissemination in clinical practice has accelerated along with this ease of restrictions. We believe that this spread in clinical practice is one of the causes of the increase in the number of adverse event reports. According to the outpatient drug efficacy group prescription quantity in the National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB) open data²⁸⁾ released by the Ministry of Health, Labour, and Welfare, there were no data on the actual prescription quantity of Suglat tablets 50 mg^® (medicinal ingredient name, ipragliflozin) because it was not in the top 30 in terms of prescription days by drug efficacy. However, we confirmed that there is a strong increasing trend with 25913528 tablets prescribed in 2015, 34863644 tablets in 2016, and 43851514 tablets in 2017. In contrast, in the U.S., the number of reported cases of all adverse events peaked in Q2 2015 and decreased sharply, and this rapid fluctuation was like the change in the number of reports caused by the Weber effect of exenatide and sitagliptin observed in a previous study.²²⁾

Second, the fluctuations that were presumed to be stimulated reporting were caused by alerts and the publication of studies in Japan. In contrast, in the U.S., the fluctuations caused by warnings were limited, and those by the publication of studies were substantial. The bias caused by the publication of studies does not always coincide with the views expressed by the study. We believe that this is because even if a nephroprotective effect is observed, it will not necessarily be reported to regulatory authorities due to the system of adverse event reporting. Therefore, we cannot deny that the number of reports on ARF might have increased due to increased attention to changes in renal function associated with the publication of studies. The impact on cumulative ROR associated with the publication of studies related to the CANVAS^7,11) was limited in Japan compared to the U.S. We believe that this was related to the extent to which the clinical trial, which was the basis of the published studies, proceeded according to a plan that matched the medical conditions in the country. Asian individuals accounted for 12.7% of CANVAS^7,11) subjects. There is limited data on Japanese individuals, and some doses used in clinical trials exceeded the approved dose in Japan, making it difficult to apply in clinical practice in Japan. The percentage of Asian individuals included in the results of other representative clinical trials was 21.6% for EMPA-REG,^9,10) 13.7% for DECLARE-TIMI,⁸⁾ and 19.9% for CREDENCE.⁶⁾ Moreover, all tests were within the approved dose range in Japan, and we believe that these trials were easier to refer to in Japanese medical practice compared with the CANVAS.^7,11) Although no significant fluctuations were observed in the cumulative analysis, fluctuations in the quarterly analysis were observed between Q2 and Q3 2016. An increasing trend was observed in the quarterly analysis of this period, which might have been caused by stimulated reporting due to the FDA alert in 2016 and the EMPA-REG study published in 2016.^1,10) However, since no clear changes in ROR were observed in the cumulative analysis with abundant data, we believe that the FDA alert¹⁾ and the publication of the study related to EMPA-REG¹⁰⁾ had a very limited effect on ROR. Regarding the effect of FDA alerts, in a previous study, ROR significantly increased after the alert was issued, affirming the effect of alerts on ROR.^1,14) The following can be considered as the background for suggesting different trends between this study and previous studies. The RORs in parts of Q2 of 2016 and Q3 of 2016, which were defined as post-alert groups in a previous study,¹⁴⁾ were higher in the present study. In a previous study, the period presumed to be strongly affected by the alert was identified as the post-alert group. Therefore, we cannot rule out the possibility that the ROR of the post-alert group might have been overestimated. Conversely, regarding the presence or absence of the effect of the alert on ROR, there are affirmative studies and negative studies on other drugs and adverse events.^22,29–31) This might be because multiple factors, such as the seriousness of the adverse event in question, the scientific basis for the association between the drug and the adverse event, the inclusion of concrete countermeasures in warnings, and media attention are involved in altering RORs. As such, in this study, we inferred that the effects of the alert were limited in the U.S.; however, it is necessary to interpret the results, keeping in mind that, although limited, the effect of the alert was present to some extent.

As a result of comprehensively examining the number of reports and changes in ROR in this study, the following points were suggested. The impact of bias varies depending on the extent of marketing regulations and how much the study design reflects the actual medical practice of each country of interest. Because the signals obtained in this study include these biases, it was not possible to conclude the relationship between SGLT-2i and ARF or the differences in the trend of ARF occurrence between Japan and the U.S. from the results of this study. However, from the viewpoint of drug risk management, we believe that the risk can be minimized using the signal detection research method and carefully monitoring the signal-detected adverse events from an early stage, even if the causal relationship is unclear.

Study Limitations

This study had the following limitations. First, the study might include duplicate reports from different reporters and incomplete or incorrect information. Second, the effects of drug interactions with concomitant drugs cannot be eliminated completely. Third, when assessing quarterly accumulated changes in RORs, changes in ROR became difficult to detect due to the amount of accumulated data for more recent quarters. Finally, since this study was conducted to determine the impact of alerts related to ARF, it is necessary to consider the possibility that the precautions taken with other adverse events related to SGLT-2i might have also affected ROR.

Owing to these restrictions, the signal detection method used in this study cannot clarify the causal relationship between drugs and adverse events. As such, further clinical trials and a comprehensive interpretation of existing clinical trials are required to clarify the causal relationship.

CONCLUSION

This study suggests the difference in the associations between SGLT-2i and ARF in Japan and the U.S. However, it is necessary to interpret these results carefully because of the possibility of overestimation suggested in the present study due to the alert biases and the publication of studies.

Conflict of Interest

Yukari Katsuhara is an employee of Takeda Pharmaceutical Company Limited. Shunya Ikeda has no conflict of interest.

Supplementary Materials

This article contains supplementary materials.

REFERENCES

• 1) Food and Drug Administration. “FDA drug safety communication: FDA strengthens kidney warnings for diabetes medicines canagliflozin (Invokana, Invokamet) and dapagliflozin (Farxiga, Xigduo XR).”: ‹https://www.fda.gov/drugs/drug-safety-and-availability/fda-drug-safety-communication-fda-strengthens-kidney-warnings-diabetes-medicines-canagliflozin›, accessed 29 September, 2021.
• 2) PMDA. “PMDA revises a label tofogliflozin. 9 Jan 2015.”: ‹https://www.info.pmda.go.jp/kaiteip/20150109A002/03.pdf›, accessed 29 September, 2021.
• 3) PMDA. “PMDA revises labels ipragliflozin, dapagliflozin, luseogliflozin, canagliflozin, and empagliflozin.”: ‹https://www.info.pmda.go.jp/kaiteip/20150109A002/02.pdf›, accessed 29 September, 2021.
• 4) The committee on the proper use of SI. “Recommendations 2016.5. on the proper use of SGLT2 inhibitors.”: ‹https://www.sagacde.gr.jp/cms/wp-content/uploads/2020/05/iryou20160512-1.pdf›, accessed 29 September, 2021.
• 5) The Committee on the Proper Use of SI. “Recommendations 2019.8. on the proper use of SGLT2 inhibitors.”: ‹https://www.nittokyo.or.jp/uploads/files/recommendation_SGLT2_190806.pdf›, accessed 29 September, 2021.
• 6) Perkovic V, Jardine MJ, Neal B, et al. Canagliflozin and renal outcomes in type 2 diabetes and nephropathy. N. Engl. J. Med., 380, 2295–2306 (2019).
• 7) Perkovic V, de Zeeuw D, Mahaffey KW, Fulcher G, Erondu N, Shaw W, Barrett TD, Weidner-Wells M, Deng H, Matthews DR, Neal B. Canagliflozin and renal outcomes in type 2 diabetes: results from the CANVAS Program randomised clinical trials. Lancet Diabetes Endocrinol., 6, 691–704 (2018).
• 8) Wiviott SD, Raz I, Bonaca MP, Mosenzon O, Kato ET, Cahn A, Silverman MG, Zelniker TA, Kuder JF, Murphy SA, Bhatt DL, Leiter LA, McGuire DK, Wilding JPH, Ruff CT, Gause-Nilsson IAM, Fredriksson M, Johansson PA, Langkilde AM, Sabatine MS. Dapagliflozin and cardiovascular outcomes in type 2 diabetes. N. Engl. J. Med., 380, 347–357 (2019).
• 9) Wanner C, Lachin JM, Inzucchi SE, Fitchett D, Mattheus M, George J, Woerle HJ, Broedl UC, von Eynatten M, Zinman B. Empagliflozin and clinical outcomes in patients with type 2 diabetes mellitus, established cardiovascular disease, and chronic kidney disease. Circulation, 137, 119–129 (2018).
• 10) Wanner C, Inzucchi SE, Lachin JM, Fitchett D, von Eynatten M, Mattheus M, Johansen OE, Woerle HJ, Broedl UC, Zinman B. Empagliflozin and progression of kidney disease in type 2 diabetes. N. Engl. J. Med., 375, 323–334 (2016).
• 11) Neal B, Perkovic V, Mahaffey KW, de Zeeuw D, Fulcher G, Erondu N, Shaw W, Law G, Desai M, Matthews DR. Canagliflozin and cardiovascular and renal events in type 2 diabetes. N. Engl. J. Med., 377, 644–657 (2017).
• 12) Urushihara H. Basic dos and don’ts in applying signal detection methods to spontaneous reporting systems databases. Jpn. J. Drug Inform., 21, 135–141 (2020).
• 13) Food and Drug Administration. “FDA Adverse Event Reporting System.”: ‹https://www.fda.gov/drugs/questions-and-answers-fdas-adverse-event-reporting-system-faers/fda-adverse-event-reporting-system-faers-latest-quarterly-data-files›, accessed 4 July, 2020
• 14) Perlman A, Heyman SN, Matok I, Stokar J, Muszkat M, Szalat A. Acute renal failure with sodium-glucose-cotransporter-2 inhibitors: analysis of the FDA adverse event report system database. Nutr. Metab. Cardiovasc. Dis., 27, 1108–1113 (2017).
• 15) PMDA. “JADER.”: ‹https://www.pmda.go.jp/safety/info-services/drugs/adr-info/suspected-adr/0004.html›, accessed 29 September, 2021.
• 16) Katsuhara Y, Ogawa T. Acute renal failure, ketoacidosis, and urogenital tract infections with SGLT2 inhibitors: signal detection using a Japanese spontaneous reporting database. Clin. Drug Investig., 40, 645–652 (2020).
• 17) Katsuhara Y, Ikeda S. Correlations between SGLT-2 inhibitors and acute renal failure by signal detection using FAERS: stratified analysis for reporting country and concomitant drugs. Clin. Drug Investig., 41, 235–243 (2021).
• 18) Hartnell NR, Wilson JP. Replication of the Weber effect using postmarketing adverse event reports voluntarily submitted to the United States Food and Drug Administration. Pharmacotherapy, 24, 743–749 (2004).
• 19) Wang HW, Hochberg AM, Pearson RK, Hauben M. An experimental investigation of masking in the US FDA adverse event reporting system database. Drug Saf., 33, 1117–1133 (2010).
• 20) Pariente A, Gregoire F, Fourrier-Reglat A, Haramburu F, Moore N. Impact of safety alerts on measures of disproportionality in spontaneous reporting databases: the notoriety bias. Drug Saf., 30, 891–898 (2007).
• 21) Poluzzi E, Raschi E, Piccinni C, Ponti FD. Data mining techniques in pharmacovigilance: analysis of the publicly accessible FDA adverse event reporting system (AERS). Data Mining Applications in Engineering and Medicine, Chapter 12, IntechOpen Ltd., London (2012). DOI: 10.5772/50095
• 22) Raschi E, Piccinni C, Poluzzi E, Marchesini G, De Ponti F. The association of pancreatitis with antidiabetic drug use: gaining insight through the FDA pharmacovigilance database. Acta Diabetol., 50, 569–577 (2013).
• 23) Sakaeda T, Tamon A, Kadoyama K, Okuno Y. Data mining of the public version of the FDA Adverse Event Reporting System. Int. J. Med. Sci., 10, 796–803 (2013).
• 24) Patek TM, Teng C, Kennedy KE, Alvarez CA, Frei CR. Comparing acute renal failure reports among antibiotics: a pharmacovigilance study of the FDA Adverse Event Reporting System (FAERS). Drug Saf., 43, 17–22 (2020).
• 25) Fan Q, Ma J, Zhang B, Li Q, Liu F, Zhao B. Assessment of acute renal failure related to small-molecule protein kinase inhibitors using the FDA adverse event reporting system. Cancer Chemother. Pharmacol., 86, 655–662 (2020).
• 26) Welch HK, Kellum JA, Kane-Gill SL. Drug-associated acute renal failure identified in the United States Food and Drug Administration Adverse Event Reporting System database. Pharmacotherapy, 38, 785–793 (2018).
• 27) Yue Z, Shi J, Li H, Li H. Association between concomitant use of acyclovir or valacyclovir with NSAIDs and an increased risk of acute kidney injury: data mining of FDA adverse event reporting system. Biol. Pharm. Bull., 41, 158–162 (2018).
• 28) The Ministry of Health, Labour, and Welfare. “National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB) open data.”: ‹https://www.mhlw.go.jp/stf/seisakunitsuite/bunya/0000177182.html›, accessed 29 September, 2021.
• 29) Hoffman KB, Demakas AR, Dimbil M, Tatonetti NP, Erdman CB. Stimulated reporting: the impact of US Food and Drug Administration-issued alerts on the adverse event reporting system (FAERS). Drug Saf., 37, 971–980 (2014).
• 30) Neha R, Subeesh V, Beulah E, Gouri N, Maheswari E. Existence of notoriety bias in FDA Adverse Event Reporting System database and its impact on signal strength. Hosp. Pharm., 56, 152–158 (2021).
• 31) Takeyama M, Sai K, Imatoh T, Segawa K, Hirasawa N, Saito Y. Influence of Japanese regulatory action on denosumab-related hypocalcemia using Japanese adverse drug event report database. Biol. Pharm. Bull., 40, 1447–1453 (2017).