Globally, approximately 250 million people are infected with the hepatitis B virus, with a high prevalence, morbidity, and mortality rate in Asia.^1,2)In Korea, although the number of cases has been declining due to the prevention of vertical transmission of hepatitis B, a significant portion of the population remains infected, with approximately 3% of the population affected and 65-75% of liver cancer cases attributed to chronic hepatitis B infection. Hepatitis B is a major cause of cirrhosis and liver cancer, making its treatment and management critical.^3,4,5,6,7)

The first choice of antiviral drugs used for treatment is nucleos(t)ide analogue (NA). Adverse events associated with NA include hypertriglyceridemia and hypercholesterolemia, and there is increasing evidence that metabolic factors such as diabetes, obesity, dyslipidemia, and metabolic syndrome are risk factors for cirrhosis and hepatocellular carcinoma.^{8,9,10,11,12,13)} Additionally, elevated blood lipid levels and weight gain are well-recognized major risk factors for cardiovascular disease.^14,15) Therefore, agents with weight loss or lipid-lowering effects may be crucial considerations for drug selection in the treatment of hepatitis B in clinical practice.

Among NAs, tenofovir disoproxil fumarate (TDF) is an orally available treatment that is preferred due to its low resistance and high efficacy. Tenofovir alafenamide fumarate (TAF) has been introduced to mitigate the side effects associated with long-term use of TDF, and several studies have reported worsening in body weight and blood lipid levels after switching. A randomized clinical trial in Korea showed a significant increase in body weight and blood lipid levels after switching from TDF to TAF. However, it remains unclear which drug was responsible for the changes in body weight and blood lipid levels, highlighting the need for further studies with long-term data.¹⁶⁾ D ue t o the unavailability of TAF data, this study focused on long-term data of TDF. Since the variety of data sources and the volume of data can affect the accuracy and diversity of the results in identifying associations, this study utilized two data sources to ensure reliability.¹⁷⁾

This study aims to use spontaneous adverse event reporting data from Korea and the United States to examine the association between TDF treatment and positive side effects of weight loss and lipid-lowering effects in patients with chronic hepatitis B.

Data sources and study design

This study utilized the raw data from the Korea Adverse Event Reporting System managed by the Korea Institute of Drug Safety & Risk Management (KIDS KAERS DB) (2307A0010), and the U.S. Food and Drug Administration Adverse Event Reporting System (FAERS DB). The KAERS DB does not collect identifiable patient information, ensuring the raw data remains anonymized. The data is distributed across nine tables, linked by a randomized report number. Similarly, the FAERS DB contains reports of adverse drug reactions, medication errors, and product quality issues submitted to the FDA, allowing for the detection of potential associations between drugs and reported adverse events. The FAERS DB is organized into seven tables linked by a randomized report number. The Kyung-Hee University Institutional Review Board determined that this study was exempted from ethical review [KHSIRB-24-262(EA)].

This study conducted an analysis using a case/non-case design, which is based on disproportionality, using raw data reported to the KAERS DB and FAERS DB from January 1, 2013, to December 31, 2022. The case/non-case design is suitable for exploring associations between drug exposure and adverse events in data whose population size is unknown.¹⁸⁾

Study population

The study population included chronic hepatitis B patients who reported adverse events to the adverse event reporting systems during the study period. The study drug was TDF, and the comparative drugs were adefovir, lamivudine, and both anhydrous and monohydrate forms of entecavir. These were based on the 2022 Korean Association for the Study of the Liver guidelines for chronic hepatitis B treatment. Defining patients based on indication variables resulted in a significant loss of data due to missing information, making it difficult to define chronic hepatitis B patients by disease code. Although NA drugs are also used for HIV treatment, where combination therapy is recommended, no co-administered drugs reported alongside TDF were confirmed as HIV treatments. Furthermore, according to drug approval information, the study drugs are approved for 'chronic hepatitis B' treatment; hence, patients administered with these drugs were defined as chronic hepatitis B patients. Indication information of data was further utilized in subgroup analysis to verify the results. Although the data sources have information on concomitant medications, this study focused on comparing TDF with other NA drugs, excluding data on non-NA drugs. To exclude duplicate reports, only the first report was analyzed. Reports with logical errors, such as missing drug codes or adverse event information, were also excluded.

Definition of criteria

Adverse events of interest included weight loss and decreased blood lipid levels, classified as the case group, with TDF exposure defining the exposure group. Given the characteristics of the data sources, there were no control groups for patients who took the study drugs without experiencing adverse events. Therefore, the adverse events reported for other drugs were used as the nonexposure group. In this study, to compare TDF with other NA drugs, the non-exposure group was defined as patients treated with adefovir, entecavir, or lamivudine, which are used as treatments for hepatitis B.

The outcome variables are reductions in blood lipid levels and weight loss. To determine this, all adverse events that could explain the reductions in blood lipid levels and weight loss were identified by referring to previous studies (Table 1). Among the adverse events related to TDF and blood lipid levels, those identified in prior studies included reductions in blood cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, and triglycerides.

Table 1 Detailed information of adverse event related to weight loss in United States Food and Drug administration adverse events reporting system database

	Preferred Term	Lowest Level Term
Lipid-lowering Adverse events	Blood cholesterol decreased High density lipoprotein decreased Hypo HDL cholesterolaemia Hypocholesterolaemia Low density lipoprotein decreased Non-high-density lipoprotein cholesterol decreased Very low density lipoprotein decreased Total cholesterol/HDL ratio decreased Ldl/hdl ratio decreased Hypolipidaemia Lipids decreased Blood triglycerides decreased Hypotriglyceridaemia	Blood cholesterol decreased Cholesterol blood decreased Cholesterol blood lowered Cholesterol blood reduced Cholesterol levels low Cholesterol low Cholesterol total decreased HDL cholesterol decreased High density lipoprotein cholesterol decreased High density lipoprotein cholesterol low Hypo HDL cholesterolaemia Hypo HDL cholesterolemia Hypocholesterolaemia Hypocholesterolemia Low density lipoprotein cholesterol decreased Low density lipoprotein cholesterol low Non-HDL cholesterol decreased Non-high-density lipoprotein cholesterol decreased Very low density lipoprotein cholesterol decreased Very low density lipoprotein cholesterol low Total cholesterol/HDL ratio decreased Ldl/hdl ratio decreased Hypolipidaemia Lipids decreased Blood triglycerides decreased Hypotriglyceridemia Triglyceride decreased Triglycerides decreased Triglycerides low
Weight loss Adverse events	Abnormal loss of weight Underweight Weight decreased	Abnormal loss of weight Loss of weight Lost weight Underweight Weight decrease Weight decreased Weight loss

Statistical analysis

The association was measured using the reporting odds ratio (ROR) and its 95% confidence interval, with the ROR defined as the ratio of exposure odds in the case group to the exposure odds in the non-case group.¹⁸⁾ An association was considered significant if the lower limit of the 95% confidence interval for the ROR exceeded 1.¹⁹⁾ A 2x2 contingency table was used to calculate these metrics (Table 2). To compare two datasets, common variables containing the same content were extracted and utilized for analysis.

Table 2 Formulas and criteria for signal detection in data mining indices

2×2 contingency table for measures of disproportionality
	Specific AEs	All other AEs
TDF	A	B
All other NAs	C	D
Indices	Definition	Criteria of signal detection
ROR	A/CB/D	Lower limit of 95% confidence interval>1

Abbreviation: AE, Adverse Event; NA, Nucleotide Analogue; ROR, Reporting odds ratio; TDF, Tenofovir Disoproxil Fumarate.

Subgroup analysis

Other factors that may be associated with the drug-adverse event relationship can be adjusted through stratified analysis.^20,21) Therefore, in this study, a sensitivity analysis was conducted by dividing the data into four subgroups to examine the association between the drug and adverse events.

For subgroup analysis, chronic hepatitis B was defined using the indication variable (DSAS_CD) from the KAERS database. Disease codes were confirmed according to the Korean Standard Classification of Diseases, with codes B18.0 and B18.1 applied in the analysis (Table 3). Only data confirmed as ‘Chronic hepatitis B’ in the MedDRA English version explicitly indicating ‘chronic’ and ‘hepatitis B’ were extracted.

Table 3 Definitions of disease codes used in this study

Code	Name
B18.0	Chronic viral hepatitis B with delta- agent
B18.1	Chronic viral hepatitis B without delta- agent

Among the causality assessment in the data source, only those corresponding to ‘Certain’, ‘Probable’, or ‘Possible’ were extracted to identify adverse events. The raw data from the spontaneous reporting system include variables assessed by experts at pharmacovigilance centers for the causality between the drug and adverse events. By utilizing these variables, high-causality signals can be identified.²²⁾ Therefore, subgroup analyses were conducted reflecting the causality assessment, and the differences in results were compared.

The third analysis was conducted on whether the ROR changed based on the year of TAF approval. Some previous studies have documented increased blood lipid levels and weight gain following the switch from TDF to TAF. To determine if there were any changes in the reporting of adverse events related to TDF after the introduction of TAF, subgroup analyses were performed based on the year TAF was approved. In South Korea, TAF was approved in 2017, which was used as a cutoff to compare data from before and after this year. In the United States, 2016 year was used.

Additionally, a subgroup analysis based on age was conducted. The primary outcome variable in this study is a symptom that can be influenced by various factors such as age and lifestyle. Among these factors, age information, which was available in the data, was used to compare its impact. The results were classified and examined by each age group.

Characteristics

Between 2013 and 2022, the Korea Adverse Event Reporting System (KAERS) recorded a total of 2,681,219 adverse event reports. Among these, 8,512 cases were associated with the use of TDF and comparator drugs. After excluding follow-up reports, 6,688 initial reports remained. Reports with missing drug names or adverse event information were also excluded, resulting in 11,245 drug-adverse event pairs for final analysis.

The FDA Adverse Event Reporting System (FAERS) recorded 14,567,307 adverse event reports during the same period. Of these, 76,249 cases involved TDF and comparator drugs. Excluding follow-up reports left 36,384 initial reports, and after excluding those with missing drug names or adverse event information, 35,656 reports remained, with 126,060 drug-adverse event pairs included in the final analysis (Fig. 1).

Fig. 1. Flow chart for target population selection.
Abbreviation: AE, Adverse event; FAERS, United States Food and Drug Administration adverse event reporting system database; KAERS DB, Korea adverse event reporting system database; NA, Nucleos(t)ide analogue.

To describe the study population, gender, age, and reporter information were linked from the raw data (Table 4). In KAERS, males reported 4,064 cases (60.8%), and females reported 2,078 cases (31.1%), showing that male reports were approximately twice as high. In FAERS, the trend was similar with 12,200 cases (34.2%) from males and 7,260 cases (20.4%) from females (Table 4). Age distribution in KAERS showed 12 cases (0.2%) in the 0- 29 years age group, 46 cases (0.7%) in the 30-39 years group, and the highest number of reports, 1,056 cases (15.8%), in the 60- 69 years group. The ‘unknown’ category comprised more than half the reports (3,960 cases, 59.2%). In FAERS, the 50-59 years age group had the most reports (3,802 cases, 10.7%), with the ‘unknown’ category accounting for 21,771 cases (61.1%), similar to KAERS. Regarding reporter information, KAERS reports were primarily submitted by physicians (5,041 cases, 75.4%), followed by consumers (589 cases, 8.81%) and pharmacists (509 cases, 7.61%). No reports were submitted by lawyers. FAERS showed a more even distribution, with the highest number of reports from consumers (10,715 cases, 30.1%), followed by other healthcare professionals excluding physicians and pharmacists (7,652 cases, 21.5%), physicians (5,983 cases, 16.8%), and the fewest reports from pharmacists (1,587 cases, 4.5%).

Table 4 Baseline characteristics of the study population

Characteristics	KAERS (n=6,688)		FAERS (n=35,656)
	N	%	N	%
Sex
Male	4064	60.8	12200	34.2
Female	2078	31.1	7260	20.4
Unknown	546	8.2	16196	45.4
Age group
0-9	2	0.0	491	1.4
10-19	4	0.1	352	1.0
20-29	6	0.1	918	2.6
30-39	47	0.7	1980	5.6
40-49	246	3.7	2502	7.0
50-59	571	8.5	3802	10.7
60-69	1056	15.8	2372	6.7
70-79	570	8.5	1179	3.3
80+	226	3.4	289	0.8
Unknown	3960	59.2	21771	61.1
Reporter type
Physician	5041	75.4	5983	16.8
Pharmacist	509	7.6	1587	4.5
Other Professional	312	4.7	7652	21.5
Lawyer	0	0.0	5183	14.5
Consumer	589	8.8	10715	30.1
Unknown	237	3.5	4536	12.7
Reporting year
2013	1411	22.1	927	2.6
2014	558	8.7	845	2.4
2015	462	7.2	2580	7.2
2016	510	8.0	3167	8.9
2017	447	7.0	3002	8.4
2018	634	9.9	3088	8.7
2019	411	6.4	3664	10.3
2020	458	7.2	11667	32.7
2021	689	10.8	3953	11.1
2022	802	12.6	2763	7.7

Association with weight loss and lipid-lowering effects

Primary results indicated that TDF was not significantly associated with adverse events related to weight loss or lipid-lowering effects. In the KAERS data, no reports of lipid-lowering events were identified, and the analysis for weight loss-related adverse events resulted in 19 reports, that did not meet the criteria (ROR=1.3[0.9-1.8])(Table 5). In the FAERS data, two reports of lipid-lowering events (ROR=0.2[0.1-0.7]) and 51 reports of weight loss-related adverse events did not meet the criteria (ROR=0.3 [0.2-0.4]).

Table 5 Disproportionality and sensitivity analysis of adverse drug reactions related to weight loss and lipid-lowering effects of tenofovir disoproxil fumarate

	KAERS				FAERS
	case	%	ROR	95%CI	case	%	ROR	95%CI
Lipid-lowering AE	0	-	-	-	2	0.00	0.2	0.1-0.7
Weight loss AE	19	0.17	1.3	0.9-1.8	51	0.04	0.3	0.2-0.4
ASSESSMENT
Lipid-lowering AE	0	-	-	-	0	-	0.0	-
Weight loss AE	0	-	-	-	39	0.04	0.3	0.2-0.3
DISEASE CODE
Lipid-lowering AE	0	-	-	-	0	-	-	-
Weight loss AE	2	0.17	2.0	0.7-6.0	3	0.04	1.3	0.4-3.7
YEAR<2017	YEAR<2016
Lipid-lowering AE	0	-	-	-	0	-	-	-
Weight loss AE	9	0.16	1.8	1.0-3.0	11	0.07	0.9	0.5-1.6
YEAR>2017	YEAR>2016
Lipid-lowering AE	0	-	-	-	2	0.00	0.2	0.0-0.7
Weight loss AE	9	0.19	1.1	0.7-1.7	40	0.04	0.3	0.2-0.4

Abbreviation: AE, Adverse event; CI, Confidence interval; FAERS, United States Food and Drug administration adverse events reporting system; KAERS, Korea adverse event reporting system; ROR, Reporting Odds Ratio.

Sensitivity analysis

Subgroup analyses based on causality assessment, disease codes, TAF market introduction year, and age consistently showed no significant association.

In the group reflecting the causality assessment of the KAERS data, no drug-adverse event pairs related to weight loss or reduction in blood lipid levels were identified. In the FAERS data, no drug-adverse event pairs related to the reduction of blood lipid levels were identified, and while 39 drug-adverse event pairs related to weight loss were observed, no significant association was found (ROR=0.3[0.2-0.3]).

After applying chronic hepatitis B-related disease codes to both data sources, adverse events related to weight loss or reduction in blood lipid levels were examined, but no signal information was identified. When examining adverse events related to weight loss, the indicator values in KAERS increased when classified by disease code but did not meet the threshold (ROR=2.0[0.7-6.0]). Similar results were observed in the FAERS data (ROR=1.3[0.4- 3.7]).

Both data sources were examined for signal information based on the approval year of TAF. Adverse events related to weight loss were checked in the KAERS data before and after 2017, but the threshold was not met (pre-2017: ROR=1.8[1.0-3.0]; post- 2017:ROR=1.1[0.7-1.7]). When the FAERS data were examined before and after 2016, 40 cases of weight loss-related adverse events were identified after 2016, showing an increase compared to the previous period. However, as the total number of reports also increased, the indicator values were calculated to be lower (pre-2016: ROR=0.9[0.5-1.6]; post-2016: ROR=0.3[0.2-0.4]). In the FAERS database, adverse events related to the reduction of blood lipid levels were all identified after the approval of TAF (ROR=0.2[0.0-0.7]). No significant differences in signal information were observed in either data source before and after the approval of TAF.

Subgroup analyses by age were conducted (Table 6). Adverse events related to decreased blood lipid levels were not subjected to further analysis as no cases were reported in KAERS, and the two cases identified in FAERS were both found in the 'Unknown' age group. In both data sources, the most cases were found in the group with an unknown age (15 cases in KAERS, 26 cases in FAERS). The next most frequently reported cases were in the 40s age group in both data sources (2 cases in KAERS, 8 cases in FAERS). The indicator values for association did not meet the criteria in any group. Weight loss is a symptom that can be more easily identified compared to other adverse events, so the reporters of this adverse event can be widely distributed. To examine weight loss according to the type of reporter, we created subgroups and analyzed the reporting patterns. As a result, in KAERS, 13 reports (68.4%) were made by physicians, and 5 reports (26.3%) were made by consumers. In FAERS, 19 reports (37.3%) were made by physicians, and 13 reports (25.5%) were made by consumers (Table 7).

Table 6 Sensitivity analysis of adverse drug reactions related to weight loss effects of tenofovir disoproxil fumarate stratified by age

age	KAERS				FAERS
	case	%	ROR	95%CI	case	%	ROR	95%CI
0-9	0	-	-	-	0	-	-	-
10-19	0	-	-	-	0	-	-	-
20-29	0	-	-	-	1	0.03	0.4	0.1-2.4
30-39	0	-	-	-	3	0.04	1.2	0.4-3.5
40-49	2	0.47	-	-	8	0.07	1.2	0.6-2.2
50-59	1	0.11	2.3	0.6-9.3	6	0.03	0.5	0.2-1.0
60-69	0	-	0	-	5	0.06	1.1	0.5-2.4
70-79	1	0.1	2.3	0.5-11.6	2	0.05	0.6	0.2-2.4
80≤	0	-	0	-	0	-	0	-
unknown	15	0.22	1.0	0.7-1.5	26	0.04	0.2	0.2-0.3

Abbreviation: AE, Adverse event; CI, Confidence interval; FAERS, United States Food and Drug administration adverse events reporting system; KAERS, Korea adverse event reporting system; ROR, Reporting Odds Ratio.

Table 7 Subgroup analysis of adverse drug reactions related to weight loss with tenofovir disoproxil fumarate stratified by reporter

Reporter	KAERS (n=19)		FAERS (n=51)
	N	%	N	%
Physician	13	68.4	19	37.3
Phamacist	1	5.3	4	7.8
Other profesional	0	0.0	6	11.8
Lawyer	0	0.0	2	3.9
Consumer	5	26.3	13	25.5

Abbreviation: FAERS, United States Food and Drug administration adverse events reporting system; KAERS, Korea adverse event reporting system.

The adverse events related to ‘weight loss’ identified in each database and their respective odds ratios were examined. In the FAERS database, ‘weight decreased’ was the most frequently reported adverse event, with 46 cases identified (Table 8). However, ‘Abnormal loss of weight’ exhibited the highest ROR value. In the KAERS database, only two adverse events were identified; among them, ‘Weight decrease’ was the most frequently reported, but ‘Weight loss’ demonstrated a higher ROR value (Table 9). Overall, none of the identified adverse events were statistically significant.

Table 8 Detailed information of adverse event related to weight loss in United States Food and Drug administration adverse events reporting system database

Adverse event	Count	ROR	CI
Abnormal loss of weight	3	2.0	0.4-8.6
High density lipoprotein decreased	2	1.0	0.2-4.2
Underweight	2	1.3	0.2-7.0
Weight decreased	46	0.4	0.3-0.5

Abbreviation: CI, Confidence interval; ROR, Reporting Odds Ratio.

Table 9 Detailed information of adverse event related to weight loss in Korea adverse event reporting system database

Adverse event	Count	ROR	CI
Weight decrease	14	1.1	0.6-2.0
Weight loss	5	2.5	0.7-9.2

Abbreviation: CI, Confidence interval; ROR, Reporting Odds Ratio.

This study utilized spontaneous adverse event reporting data from South Korea and the United States to identify adverse events related to weight loss and lipid-lowering in chronic hepatitis B patients treated with TDF. According to two adverse event reporting databases, the reporting rate was higher in men, predominantly in middle-aged individuals. A 2021 epidemiological study in Korea showed that the proportion of male hepatitis B patients was 58.1%, higher than females, with an average age of 55.6 years.⁴⁾ According to the Centers for Disease Control and Prevention (CDC) report, the prevalence in men in the U.S. was also higher at 5.3% compared to 3.4% in women.²³⁾ Spontaneous adverse event reports reflect actual clinical cases, which is a significant advantage for pharmacovigilance. This study confirmed that the trends in disease prevalence and adverse event reporting were consistent. While KAERS data were mainly reported by physicians, FAERS data were collected from various reporters. A patient survey indicated that 87% of patients discussed potential adverse events with their doctors, but doctors were more likely to dismiss them.²⁴⁾ Consumer-submitted reports play a crucial role in the early detection of safety signals.²⁵⁾ Spontaneous adverse event reporting systems are valuable because they capture symptoms and interpretations from diverse reporters in real-world settings. Promoting and encouraging active reporting in adverse event reporting systems is necessary to minimize reporter bias.

In KAERS, there were no cases of lipid level reduction, and 19 cases of weight loss, while FAERS reported 2 cases of lipid level reduction and 51 cases of weight loss. The absolute number of adverse reactions in FAERS was higher, likely including more reports of the controversial adverse events. KAERS data showed higher disproportionality analysis values for weight loss compared to FAERS, with ROR exceeding 1 in KAERS. This suggests more reports of weight loss related to TDF in KAERS compared to other NA drugs, though differing reporting characteristics limit the absolute comparison of analysis indicators. Consequently, although previous literature highlighted significant weight loss and lipid level reduction due to TDF, this study did not show significant results. Very few reports of lipid-lowering adverse events in KAERS and FAERS (0cases, 2cases) contradicted the hypothesis that TDF reduces lipid levels. Hepatitis B treatment primarily occurs in outpatient settings, and lipid level reduction may not be frequently confirmed via blood tests and thus not reported as an adverse event.²⁶⁾ Additionally, due to the characteristics of the data sources and the fact that weight loss and lipidlowering might be perceived as positive responses rather than adverse effects, there may be underreporting in spontaneous adverse event data.²⁷⁾ Additional analysis divided into four subgroups did not reveal signal information, indicating that lipidlowering and weight loss adverse events were unaffected by indications, causality, TAF approval, or age.

In the subgroup analysis reflecting causality assessment, the KAERS results showed zero cases for both adverse events of interest. A limitation of spontaneous adverse event reporting data is that it is biased from the reporting stage, as reporters are likely to report adverse events they suspect are related to the drug.²⁸⁾ Since it is difficult to assess causality for symptoms that have not yet been confirmed as drug-related adverse events, this may have resulted in zero reported cases. In KAERS, there were 9 cases of weight loss adverse events reported before TAF was marketed in 2017, and 9 cases after 2017. It was also confirmed that the lower limits of the 95% confidence intervals for the ROR decreased accordingly (ROR=1.8→1.1). Similarly, in FAERS, when analyzed based on the year TAF was marketed, the analysis indicator values were found to have decreased (ROR=0.9→0.3). All analysis indicator values suggest that there is no association between the drug and the adverse events, and that the reporting of the adverse events of interest was not influenced by TAF. On the other hand, in the FAERS results, 2 cases of adverse events related to decreased blood lipid levels were reported after 2016. However, since this is a small number of cases, it is insufficient to make definitive conclusions or generalizations. Therefore, it is necessary to monitor the trend of adverse event reporting in the future. Analysis of weight loss reports by age revealed that most cases were reported in individuals under 60 years old, suggesting that these instances of weight loss may not be related to agerelated changes. Although weight loss is a symptom that can be easily recognized by individuals, it was more frequently reported by healthcare professionals. In both FAERS and KAERS, weight loss was most commonly reported by physicians, which differs from the overall pattern of adverse event reporting.

This study has some limitations. First, co-reported concomitant drugs in the raw database make it difficult to ascertain if the drug of interest caused the adverse event. Additionally, underreporting in spontaneous reporting data and incomplete collection of all adverse event information are limitations, including the fact that positive symptoms may be underreported. However, the purpose of a spontaneous adverse event reporting system is to propose new hypotheses based on real-world data, which this study fulfills. Future studies need to verify the results related to weight and lipid levels and reaffirm the association between the drug and adverse events based on the findings. Second, confounding factors such as underlying diseases, concomitant medications or dosing were not considered in the study. Weight and lipid levels are influenced by various factors. However, due to data limitations, only restricted information was obtained, and reflecting such information might result in data loss, complicating signal information identification. Observational studies should be conducted using data sources like medical records or claims data that can adequately reflect factors related to weight and lipid levels. Lastly, the study could not confirm adverse events related to TAF, developed to improve TDF’s side effects, due to data source limitations. Most of previous studies have examined weight and lipid changes after switching from TDF to TAF. Additional research on TAF is needed to enhance reliability by comparing with previous study results.

Despite these limitations, this study has significance in identifying controversial adverse events using data mining from KAERS and FAERS. Chronic disease medications require long-term use, necessitating the identification of long-term adverse events. With over 10 years since the drug's market introduction and the use of recent data, this study presents a good opportunity to reassess drug adverse events. The study provides reference information on drug safety and serves as a foundation for identifying future adverse drug reactions. This is the first study to report TDF adverse events using two data sources, comparing KAERS and FAERS, and confirming controversial adverse evnets. It also derives drug safety information, providing basic data for causal studies and policy development for safe hepatitis B drug use.

This study identified reports of weight loss and lipid level reduction associated with the use of tenofovir disoproxil fumarate in chronic hepatitis B patients using a spontaneous adverse event reporting database. However, the signal information did not suggest a potential association between the adverse events and the drug. Further research is needed to verify these results using longterm data and sources that can determine the adverse event onset period. Additionally, since outcome variables can be influenced by factors such as age and gender, well-designed observational studies are required to confirm causality.

This research was supported by a grant (21153MFDS601) from Ministry of Food and Drug Safety in 2024.

All authors declare that they have no conflict of interest.

Hae Sun Suh : Professor

Ju Young Kim : Graduate student