The VAERS served as the data source for this study. VAERS, established in 1990 and jointly administered by the Centers for Disease Control and Prevention and the Food and Drug Administration, is a national passive surveillance system aimed at identifying potential safety signals and generating data on adverse events following vaccination (14). Due to the spontaneous reporting nature of the VAERS system, establishing causal links between vaccines and reported adverse events using VAERS data alone is generally not feasible (15).

VAERS collects voluntary reports from a variety of sources, including healthcare providers, vaccine manufacturers, vaccine recipients, and other source. Each report collects information pertaining to the vaccinated individual (such as age and sex), details of the administered vaccine, including concomitant vaccines, and the description of the adverse event experienced. Adverse events are systematically coded by trained personnel using the Medical Dictionary for Regulatory Activities (MedDRA), a clinically validated, internationally standardized terminology (16). In MedDRA, reported signs, symptoms, or diagnoses are assigned one or more preferred terms (PTs) within a system organ class (16). Reports submitted to VAERS are classified as “serious” if they meet one or more criteria defined by the US Code of Federal Regulations (17). These criteria encompass events leading to death, life-threatening illness, hospitalization or its extension, congenital anomalies, permanent disability, or requiring medical intervention to prevent these outcomes.

For this study, our analysis encompassed all documented safety signals associated with JEV from March 1993 to February 2025 in the VAERS database. Those JEV were classified four types (IXIARO, JE-VAX, J-VAX; NO Brand) in VAERS. These classifications are based on vaccine product names as recorded in VAERS and represent distinct vaccine types: IXIARO (a cell culture-derived, inactivated vaccine), JE-VAX (a mouse brain-derived, inactivated vaccine that was discontinued globally and replaced by IXIARO), and J-VAX (another mouse brain-derived, inactivated vaccine). The ‘NO Brand’ category includes VAERS reports where the specific JEV brand was not specified. Signals from this category warrant particular caution due to its inherent heterogeneity and potential for misclassification or incomplete reporting, which may reflect reporting artifacts rather than true vaccine-specific safety concerns. Therefore, RORs derived from the ‘NO Brand’ category should be interpreted with caution and are considered less reliable for specific signal detection. This classification scheme allows for the comprehensive analysis of all JEV-related adverse event reports within the VAERS database, including those from earlier, now discontinued, vaccine formulations. Adverse events were coded using preferred terms (PTs) from the MedDRA terminology. Vaccine manufacturers were identified using the Manufacturer of Vaccine variable. While JE-VAX and J-VAX have been discontinued, their inclusion in this study provides valuable historical safety profiles of JVE in the U. S. and allows for a comprehensive understanding of long-term post-marketing surveillance data. This historical context is relevant for identifying enduring safety signals, understanding evolving adverse event patterns, and informing regulatory considerations for future vaccine development.

Disproportionality analysis was conducted using the ROR. By intentionally omitting inclusion criteria related to patient demographics, adverse event severity, or reporting sources, we aimed to maintain population-level representativeness in this pharmacovigilance assessment. The reported data were categorized and analyzed based on key characteristics, including vaccine type, age, sex, seriousness of the event, and year of reporting. The ROR was calculated based on a 2 × 2 contingency table comparing the reporting of a specific adverse event with a specific JEV versus all other vaccines in VAERS. The formula for ROR is: ROR = (a/b)/(c/d), where ‘a’ is the number of reports for the specific AE-vaccine pair, ‘b’ is the number of reports for the specific vaccine but not the AE, ‘c’ is the number of reports for the AE but not the specific vaccine, and ‘d’ is the number of reports for neither the AE nor the specific vaccine. A minimum of three reports for the specific AE-vaccine pair (n ≥ 3) was required for ROR calculation. Statistical significance was defined by a lower bound of the 95% confidence interval (95% CI) of the ROR being greater than 1 (17). A p-value of 0.05 or less was used to determine statistical significance. Due to the large number of comparisons in disproportionality analysis, there is a risk of false positive signals from multiple testing. While we report all statistically significant RORs, we acknowledge that a formal False Discovery Rate (FDR) correction was not applied, as disproportionality analysis is primarily hypothesis-generating. Signals, especially those with small report counts, warrant further investigation through more robust epidemiological studies. Importantly, ROR estimates from passive surveillance systems like VAERS are susceptible to various reporting biases, including underreporting, differential reporting by event type or vaccine, and stimulated reporting (e.g., increased reporting following media attention). These biases can inflate or deflate ROR values, meaning observed associations warrant cautious interpretation and do not necessarily indicate causation. Descriptive statistical analysis was performed with R Studio (version 4.4.2) to summarize the key characteristics of the reports between all JEV and different categories. Line graphs were utilized to visualize trends in reporting over time.

To investigate the potential relationship between public interest and the reporting of adverse events following vaccination, we conducted a Google Trends analysis. This ecological analysis explores patterns of public interest in relation to reported adverse events, rather than establishing causal links, which provide additional contextual insights into public perception trends. Google Trends¹ is an online platform that provides data on the relative search volume of specific terms or topics across different regions and time periods. The relative search volume is presented on a scale from 0 to 100, where each data point represents the search interest for a given term or topic relative to the highest point in the selected time series, which is normalized to 100 (17). Analysis can be performed using either specific search terms or broader topics, with the latter offering wider coverage by encompassing a group of related terms that share the same concept. In this study, we utilized search topics to capture a comprehensive range of related queries. The time period for data collection was aligned with the period of adverse event reports from VAERS, specifically from 2004 to 2025. The geographical focus of the search was set to worldwide to correspond with the scope of the adverse event data. We selected “All categories” and “Web search” for the analysis settings. The specific search topic used in this analysis was “Japanese Encephalitis Vaccine.” Google Trends data were accessed on April 5, 2025. Descriptive analysis was conducted to examine the temporal changes in the relative search volume for the selected topic during the study period. These trends in online search queries were then compared with the annual number of adverse event reports obtained from VAERS to visually assess potential correlations.

This study was based on publicly available de-identified data from VAERS and Google Trends; therefore, ethical approval and informed consent were not required.

Over the 32-year period from 1993 to 2025, a total of 1,384 adverse event reports related to different JEV brands, corresponding to 6,596 vaccine-event pairs (3,452 IXIARO, 1368, JE-VAX, 698, J-VAX, 1078, NO Brand) were retrieved in VAERS, with Figure 1 illustrating the annual number of these reports. Reports associated with J-VAX were observed predominantly in the earlier years of the reporting period, with a peak of 32 reports in 1999 and a general decline thereafter, with sporadic reports in later years. The reports of JE-VAX were most prominent from 2003 to 2008, reaching a high of 53 reports in 2005, before decreasing significantly in subsequent years. Reports classified as “NO Brand” were present throughout the study period, with varying numbers, reaching a peak of 20 reports in 2018. However, the reports of IXIARO, show a distinct pattern, with a gradual increase starting around 2010 and a substantial rise from 2017 onwards, with annual report numbers consistently above 50 reports and reaching a peak of 74 reports in 2018. Overall, the temporal distribution of JEV vaccine adverse event reports in VAERS exhibit a shift in the dominant reporting brand over the study period, with IXIARO accounting for the majority of reports in recent years.

Table 1 presents the detail distribution of adverse event reports for JEV in the VAERS database (N = 1,384), of which 284 (20.5%) were classified as serious events, categorized by age, sex, along with the serious events and deaths. The highest proportion of reports originated from the 18–29 years age group, accounting for 39.52% (n = 547) of the total, followed by the 30–59 years age group at 34.68% (n = 480). Conversely, the <6 months and ≥60 years age groups contributed the lowest percentages of reports, at 0.43% (n = 6) and 3.4% (n = 47), respectively. Reports from females (52.9%) slightly outnumbered those from males (47.1%) across all age groups, excluding unknown sex. Serious events were distributed differently across age groups; the 18–29 years age group had the highest number of serious events (n = 41 in females, n = 55 in males), and the unknown age group also showed a substantial number of serious events (n = 22 in females, n = 31 in males). Fatal outcomes were reported in the 6 months–5 years (n = 1), 6–17 years (n = 3), 18–29 years (n = 1), 30–59 years (n = 1), and unknown age groups (n = 3).

The fatal outcome rate of 0.80% (11 out of 1,384 reports) observed in the VAERS database represents the proportion of death reports among all adverse event reports for JEV (Table 1). This rate should not be confused with the mortality rate of Japanese encephalitis disease (typically 20–30% in symptomatic cases) (4, 5), as VAERS reports are submitted for events following vaccination. These deaths were distributed across several age groups: one death was reported in the 6 months–5 years age group, three deaths in the 6–17 years age group (two males, one female), three deaths in the 18–29 years age group (two males, one female), one death in the 30–59 years age group (male), and three deaths in the unknown age group (males). Notably, no deaths were reported in the <6 months and ≥60 years age groups. The 6–17 years, 18–29 years, and unknown age groups each reported three deaths. However, because the unknown age group has fewer reports (n = 148) than the 18–29 years group (n = 547), the death rate was higher in the unknown group (2.03%) than in the 18–29 years group (0.55%). The 6–17 years age group had the highest death rate (3.57%) among all groups.

The higher death rate in children and adolescents (6–17 years) might indicate an increased vulnerability in this age group, possibly due to immune factors or the nature of adverse events experienced. The relatively high number of deaths in the unknown age group may be due to underreporting of age information in severe cases, making this group worth further investigation. The relatively low death rate in the 18–29 years age group, despite accounting for the largest proportion of reports (39.52%) and serious events (n = 96), could suggest that young adults either experience less severe adverse events or have better access to care and recovery. However, these observations are based on a limited number of deaths and the inherent limitations of passive reporting systems like VAERS, and thus require cautious interpretation without implying causality. However, it is crucial to emphasize that these observations are based on a limited number of deaths and the inherent limitations of passive reporting systems like VAERS. Without access to detailed clinical narratives, or complete medical histories—which are typically unavailable in the public VAERS data used here—it is impossible to determine whether the vaccination played a causal role in these fatalities. Therefore, these findings should be interpreted with caution and do not imply causation.

This analysis was conducted on 6,596 vaccine-event pairs linked to JEV. We compared these JEV with other vaccines in the database. Table 2 lists the top 20 reported adverse events associated with JEV that showed a statistically significant disproportionality signal in the VAERS database. The top five most frequently reported events with positive signals were nausea (n = 134, ROR = 1.32, 95%CI: 1.11–1.57), dizziness (n = 132, ROR = 1.33, 95%CI: 1.12–1.58), pruritus (n = 127, ROR = 2.33, 95%CI: 1.95–2.77), rash (n = 126, ROR = 1.81, 95%CI: 1.51–2.16), and urticaria (n = 122, ROR = 2.89, 95%CI: 2.42–3.46). Among the listed events, urticaria showed the highest ROR (2.89), followed by pruritus (2.33), suggesting stronger disproportionality signals for these allergic reactions. Other reported events with statistically significant RORs included vomiting (n = 95, ROR = 1.62), paraesthesia (n = 79, ROR = 2.03), syncope (n = 63, ROR = 1.81), and less frequent but potentially more serious events like loss of consciousness (n = 39, ROR = 1.83) and seizure (n = 28, ROR = 2.56). The findings underscore a spectrum of adverse events linked to JEV in VAERS, exhibiting different frequencies and signal strengths.

We then analyzed each JEV in comparison with the other JEV. Table 3 presented the 20 most frequently reported adverse events with statistically significant disproportionality signals for each JEV brand (IXIARO, JE-VAX, J-VAX, and NO Brand) in the VAERS database, along with their report counts (N) and RORs with 95% CI. The most reported JEV was IXIARO, which resulted in statistical significance in 135 out of 1,050 vaccine reaction pairs (12.86%). Table 3 presents IXIARO’s data, highlighting the top 20 statistically significant AEFIs, most of which are already documented in the comprehensive JEV characteristics summary. For IXIARO, commonly reported events with positive signals included dizziness (n = 77, ROR = 1.48), rash (n = 51, ROR = 1.39), vomiting (n = 43, ROR = 1.4), and paraesthesia (n = 41, ROR = 2.01). Notably, IXIARO also showed a strong signal for immediate post-injection reaction (n = 25, ROR = 4.67) and potentially serious events such as loss of consciousness (n = 32, ROR = 2.88) and seizure (n = 17, ROR = 2.97).

Compared to other JEV in VAERS, JE-VAX demonstrated statistical significance in 58 out of 387 vaccine-reaction pairs. The top 20 most reported PT of JE-VAX was list (Table 3). JE-VAX reports highlighted signals for headache (n = 50, ROR = 1.36), dizziness (n = 31, ROR = 1.51), and erythema (n = 28, ROR = 2.48). Of note, the disproportionality analysis detected strong signals for JE-VAX and more severe events, including encephalitis (n = 5, ROR = 15.69) and angioneurotic oedema (n = 4, ROR = 68.17). While these high RORs warrant attention due to the potential severity of the events, it is crucial to note that they represent signals from passive surveillance necessitating further investigation and do not, by themselves, establish causation.

For J-VAX, statistically significant disproportionalities were observed in 46 out of 166 vaccine-reaction pairs. Table 3 shows the 20 more reported AEFIs for J-VAX. J-VAX showed strong signals for pruritus (n = 40, ROR = 7.2), rash (n = 28, ROR = 3.88), and urticaria (n = 26, ROR = 5.93), indicating a propensity for allergic reactions. Furthermore, J-VAX had extremely high RORs for injection site hypersensitivity (n = 18, ROR = 16.43) and laryngospasm (n = 11, ROR = 185.73), although the number of reports for laryngospasm was relatively low. The extremely high ROR for laryngospasm, while statistically significant, is based on a small number of reports and may be influenced by reporting artifacts or biases. As with all signals from disproportionality analysis, this finding is exploratory and requires confirmation.

Analyses of the ‘NO Brand’ category should be interpreted with the understanding that this group is inherently less precise, and signals identified within it may be confounded. Disproportionalities for NO Brand were statistically significant in 51 out of 553 vaccine-reaction pairs. The reports classified as “NO Brand,” pyrexia was the most frequent event with a positive signal (n = 47, ROR = 1.44). Table 3 presented the top 20 more reported AEFIs for NO Brand. This category also exhibited strong signals for seizure (n = 11, ROR = 6.18), febrile convulsion (n = 6, ROR = 9.05), and upper respiratory infection (n = 6, ROR = 22.4). Notably, skin infection showed a very high ROR (n = 3, ROR = 61.34 [95% CI, 19.69–191.1]); however, the small number of cases (n = 3) and wide confidence interval (19.69–191.1) urge caution in interpretation, as such disproportionate reporting might be influenced by coding artifacts or reporting bias rather than a true safety signal. Therefore, RORs for the ‘NO Brand’ category should be regarded with particular skepticism due to the heterogeneity and potential misclassification within this group, rendering them less reliable for specific signal detection compared to branded vaccines. In summary, these brand-specific analyses reveal distinct adverse event profiles and varying signal strengths across different JEV products.

Figure 2 compares the annual number of JEV-related adverse event reports in VAERS (blue line, left y-axis) with the relative search volume for JEV on Google Trends (red line, right y-axis) from 1993 to 2025. Google search data are only available from 2004 onward (Google’s founding year); hence, values before 2004 are zero. The top related queries retrieved by Google Trends included terms related to JEV side effects, indicating public interest in vaccine safety. The VAERS reports show fluctuating numbers, generally staying below 60 reports per year before 2017, followed by a sharp rise peaking at over 75 reports in 2017–2018, a decline, and another increase in 2023–2024. The Google search volume for JEV displays a general upward trajectory after 2004. Notably, the temporal patterns of the two data streams show increasing alignment after 2004, with similar peaks and troughs (e.g., the rise around 2017–2018 and 2023–2024).

The primary significance of this analysis lies in exploring the potential influence of public awareness and media attention—proxied here by search volume—on spontaneous adverse event reporting. The observed co-movement suggests that heightened public interest, possibly driven by news or online discussions about JEV safety, may coincide with increased reporting to VAERS. This underscores a key limitation of passive surveillance systems: reporting rates can be influenced by external factors unrelated to the actual safety profile of the vaccine. Consequently, spikes in reports should not be interpreted as evidence of increased risk without further investigation. Future studies incorporating media event data could help elucidate the drivers behind these correlations. Data for 2025 are incomplete as of the analysis date (February 1, 2025).