Associations between COVID-19 vaccination and incident psychiatric disorders after breakthrough SARS-CoV-2 infection: The VENUS Study

Background: The associations between COVID-19 vaccination and post-COVID psychiatric disorders are unclear. Furthermore, it is uncertain if these associations differ depending on the dominant SARS-CoV-2 variant at the time of infection. This retrospective cohort study aimed to clarify the associations between COVID-19 vaccination and incident psychiatric disorders after breakthrough infection according to the different variant periods in Japan. Methods: Medical claims data, COVID-19 case-related information, and vaccination records were collected from three Japanese municipalities. The study population comprised public insurance enrollees aged ≥ 65 years who developed COVID-19 between June 2021 and December 2022. The study exposure was each participant ’ s vaccination status 14 days before infection, and the outcomes were the occurrence of psychiatric disorders within three months of infection. Multivariable logistic regression analyses were performed to calculate the odds ratios (ORs) and 95 % confidence intervals (CIs) of vaccination for the occurrence of psychiatric disorders. Analyses were conducted for the Delta period (June to December 2021), Omicron BA.1/BA.2 period (January to June 2022), and Omicron BA.5 period (July to December 2022). Results: We analyzed 270 participants (vaccinated: 149) in the Delta period, 2,963 participants (vaccinated: 2,699) in the Omicron BA.1/BA.2 period, and 7,723 participants (vaccinated: 7,159) in the Omicron BA.5 period. During the Delta period, vaccinated participants had significantly lower odds for psychotic disorders (OR: 0.23, 95 % CI: 0.06 – 0.88, P = 0.032) than unvaccinated participants. During the Omicron BA.5 period, vaccinated participants had significantly lower odds for organic mental disorders (OR: 0.54, 95 % CI: 0.30 – 0.95, P = 0.033), psychotic disorders (OR: 0.31, 95 % CI: 0.19 – 0.53, P < 0.001), mood disorders (OR: 0.53, 95 % CI: 0.29 – 0.99, P = 0.046), and insomnia (OR: 0.48, 95 % CI: 0.32 – 0.72, P < 0.001) than unvaccinated participants. There were no significant differences in psychiatric disorders between the vaccinated and unvaccinated groups during the Omicron BA.1/BA.2 period. Conclusions: This is the first study to demonstrate that the associations between COVID-19 vaccination and post-COVID psychiatric disorders vary among the different variant periods. Future studies on these associations should be conducted with consideration to the prevalent circulating variants.


Introduction
The first cases of COVID-19 were reported in Wuhan, China in December 2019, and this local outbreak rapidly evolved into a global pandemic that infected millions and led to unprecedented health, social, and economic crises (Cucinotta and Vanelli, 2020).Many COVID-19 survivors were found to have various symptoms that persisted or developed after the initial infection, and these post-acute and chronic sequalae are collectively described as "long COVID" (Fernández-de-Las-Peñas, 2022).In addition to physical symptoms, long COVID also includes psychiatric conditions such as anxiety, depression, stress and adjustment disorders, cognitive decline, and sleep disorders (Xie et al., 2022).The US Centers for Disease Control and Prevention reported that older individuals aged ≥65 years have a higher risk of developing mental health conditions than those aged 18-64 years (Bull-Otterson et al., 2022).The psychiatric manifestations of long COVID not only affect individuals, but may also have serious socioeconomic repercussions.
Studies have reported that COVID-19 vaccination can exert a variety of effects on psychiatric disorders occurring after breakthrough infection.For example, Al-Aly et al. found that vaccinated COVID-19 survivors had a lower risk of post-acute mental health sequelae than their unvaccinated counterparts (Al-Aly et al., 2022).In contrast, Taquet et al. reported that COVID-19 vaccination did not reduce the six-month incidence of post-COVID anxiety disorder, depression, and mood disorder (Taquet et al., 2022a).Accordingly, current evidence on the associations between vaccination and specific psychiatric disorders after COVID-19 remains inconsistent.
Furthermore, the risks of psychiatric disorders occurring after breakthrough infection may differ depending on the dominant circulating SARS-CoV-2 variant at the time of infection.Antonelli et al. noted that COVID-19 cases during the Omicron wave were less likely to develop low mood and other long COVID symptoms than those occurring during the earlier Delta wave (Antonelli et al., 2022), but Taquet et al. reported similar risks for neurological and psychiatric outcomes between the two variants (Taquet et al., 2022b).Although the incidences of post-COVID psychiatric outcomes may differ between the Delta and Omicron periods, studies have yet to examine the associations between COVID-19 vaccination and incident psychiatric disorders according to these periods.
This study was conducted to clarify the associations between COVID-19 vaccination and incident psychiatric disorders after breakthrough infection according to the different dominant variant periods in Japan.

Study design and settings
This retrospective cohort study was conducted using data provided by the Longevity Improvement and Fair Evidence (LIFE) Study, which is a multi-region database project created by Kyushu University (Fukuoka, Japan) (Fukuda et al., 2022).The LIFE Study collects and compiles a variety of anonymized data types-including National Health Insurance claims data and Latter-Stage Older Persons Health Care System claims data-from participating municipalities.National Health Insurance is a public insurance system where municipal governments serve as insurers; enrollees (aged 0-74 years) include self-employed persons, unemployed persons, primary industry workers, and their dependents.The Latter-Stage Older Persons Health Care System is a public insurance system where prefectural governments serve as insurers; enrollees include residents aged 65-74 years with specific disabilities and all residents aged ≥75 years.The medical claims data for these insurance systems contain recorded diagnoses and treatment-related information for all insurancecovered clinical encounters.Diagnoses are recorded using International Classification of Diseases, 10th Revision (ICD-10) codes.Based on the LIFE Study, the Vaccine Effectiveness, Networking, and Universal Safety (VENUS) Study was created with the aim of investigating vaccine effectiveness and safety using real-world data.The VENUS Study utilizes an integrated database comprising claims data from the LIFE Study, information on COVID-19 cases from the Health Center Real-time Information-sharing System on COVID-19 (HER-SYS), and detailed vaccination records from the Vaccination Record System (VRS).The HER-SYS is a national information-sharing system on COVID-19 cases managed by the Ministry of Health, Labour and Welfare, and contains data on all COVID-19 patients until September 25, 2022.From September 26, 2022 onward, the HER-SYS only stores data on the following COVID-19 patients: (i) patients aged ≥65 years, (ii) patients requiring hospitalization, (iii) patients determined by physicians to be at risk of severe disease or require COVID-19 medications or oxygen therapy, and (iv) patients who are pregnant.The VRS is a municipal-level system that records the COVID-19 vaccination statuses of residents, and includes information on vaccination dates, doses, vaccine manufacturers, and lot numbers (Fukuda et al., 2023).
This study used a combination of National Health Insurance claims data, Latter-Stage Older Persons Health Care System claims data, HER-SYS data, and VRS data from three municipalities participating in the VENUS Study.The study population comprised National Health Insurance and Latter-Stage Older Persons Health Care System enrollees aged ≥65 years with a recorded diagnosis of COVID-19 between June 2021 and December 2022.Each participant's index date was set as the date of their first recorded COVID-19 diagnosis during the observation period.We excluded individuals who were not enrolled in either insurance system six months before their index dates.The observation period was divided into the following three periods according to the prevalent circulating SARS-CoV-2 variant/subvariant: Delta period (June to December 2021), Omicron BA.1/BA.2period (January to June 2022), and Omicron BA.5 period (July to December 2022).
This study was approved by the Kyushu University Institutional Review Board for Clinical Research (Approval No. 22114-02).

Exposure
The study exposure was COVID-19 vaccination between June 2021 and December 2022.Participants were categorized as vaccinated or unvaccinated based on their vaccination statuses 14 days before their index dates.Each participant's vaccination status and number of vaccine doses were ascertained using the VRS data.In Japan, vaccinations for the general public began in April 2021 for persons aged ≥65 years.By July 2021 (i.e., during the Delta period), 87 % of persons aged ≥65 years throughout Japan had completed the first dose of vaccination, and 77 % had completed the second dose (Digital Agency, 2023).Early vaccination was only available for persons aged ≥65 years, and was not contingent on any pre-existing conditions (including psychiatric disorders).A preliminary analysis found that 83.9 % of the study population with vaccine type data had received the BNT162b2 vaccine (Pfizer-BioNTech).Therefore, vaccine type was not included as a variable in our study.

Outcomes
The study outcomes were the occurrence of psychiatric disorders in a participant within three months after the index date.Using recorded diagnoses in the claims data, we analyzed the following five groups of psychiatric disorders (ICD-10 codes): organic mental disorders (F00-F09), psychotic disorders (F20-F29), mood disorders (F30-F39), anxiety disorders (F40-F49), and insomnia (F51.0,G47.0).Each group of psychiatric disorders was analyzed separately.For each analysis of a psychiatric disorder group, we excluded cases who had a recorded diagnosis from the same target disorder group during the six-month period before their index dates.For example, the analysis of organic mental disorders excluded cases with a recorded diagnosis of any organic mental disorder six months before their index dates; however, cases with a recorded diagnosis of any non-target disorder (e.g., mood disorder) were included in that analysis.
To examine the associations between COVID-19 vaccination and incident psychiatric disorders within three months after breakthrough infection, we constructed multivariable logistic regression models using each target psychiatric disorder group as the dependent variable and vaccination status as the independent variable of interest.The models also included sex, age, and comorbidities as covariates.The odds ratios (ORs) and 95 % confidence intervals (CIs) of vaccination for the occurrence of post-COVID psychiatric disorders were calculated.The regression analyses were conducted according to SARS-CoV-2 variant period.Moreover, we calculated the risk difference per 1000 people at one year based on the difference between the estimated incidence rates in the vaccinated and unvaccinated cohorts.
To examine the associations between the number of vaccine doses and incident psychiatric disorders within three months after breakthrough infection, we constructed multivariable logistic regression models using each target psychiatric disorder group as the dependent variable and the number of vaccine doses before the index dates as the independent variable of interest.
The HER-SYS began storing data of high-risk individuals only instead of all COVID-19 patients from September 26, 2022.To remove the potential confounding exerted by this shift in recorded cases, we conducted a sensitivity analysis of the associations between COVID-19 vaccination and incident psychiatric disorders during the Omicron BA.5 period after excluding patients who developed COVID-19 on September 26, 2022 or later.Given the potential overlap between psychiatric disorders, we conducted another sensitivity analysis in which the multivariable logistic regression model for each psychiatric disorder group also adjusted for psychiatric symptoms other than the target outcome.For example, the regression model for organic mental disorders included psychotic disorders, mood disorders, anxiety disorders, and insomnia as comorbidities during the six-month period before the index dates.
Statistical analyses were performed using Stata version 17.0 (Stata Corp, College Station, TX, US), and significance was set at P < 0.05 (twotailed).

Results
The study population comprised 299 participants in the Delta period, 3,584 participants in the Omicron BA.1/BA.2period, and 9,319 participants in the Omicron BA.5 period.Among these, the number (percentage) of vaccinated participants was 166 (55.5 %) in the Delta period, 3,255 (90.8 %) in the Omicron BA.1/BA.2period, and 8,662 (92.9 %) in the Omicron BA.5 period.
Table 1 presents the characteristics of participants without organic mental disorders six months before the index date.The mean age (standard deviation) of the unvaccinated participants was 75.6 (6.7) years in the Delta period, 78.2 (8.1) years in the Omicron BA.1/BA.2period, and 78.3 (8.8) years in the Omicron BA.5 period.The mean age (standard deviation) of the vaccinated participants was 76.2 (7.1) years in the Delta period, 77.4 (8.2) years in the Omicron BA.1/BA.2period, and 77.2 (7.9) years in the Omicron BA.5 period.Among the unvaccinated participants, the proportion of women was 59.5 % in the Delta period, 58.0 % in the Omicron BA.1/BA.2period, and 61.3 % in the Omicron BA.5 period.Among the vaccinated participants, the proportion of women was 51.7 % in the Delta period, 55.4 % in the Omicron BA.1/BA.2period, and 56.0 % in the Omicron BA.5 period.Diabetes without chronic complications was the most prevalent comorbidity in all periods for both unvaccinated (Delta: 33.1 %, Omicron BA.1/BA.2:23.5 %, Omicron BA.5: 24.5 %) and vaccinated (Delta: 27.5 %, Omicron BA.1/BA.2:27.3 %, Omicron BA.5: 25.3 %) participants.The characteristics of participants without psychotic disorders, mood disorders, anxiety disorders, and insomnia during the six-month period before their index dates are presented in Table A .1,Table A.2,Table A.3,and Table A.4,respectively (Appendix).
Table 4 shows the incidences of psychiatric disorders occurring within three months after breakthrough infection according to the number of vaccine doses.Among the one-dose participants, the incidence proportions of mood disorders and insomnia were highest during the Delta period (mood disorders: 6.5 %, insomnia: 13.0 %), whereas the incidence proportion of organic mental disorders was highest during the Omicron BA.5 period (9.1 %) and lowest during the Omicron BA.1/BA.2period (5.9 %).The incidence proportions of psychotic disorders and anxiety disorders were 0 % in all periods.Among the two-dose participants, the incidence proportions of all psychiatric disorders except organic mental disorders were highest during the Delta period (psychotic disorders: 4.6 %, mood disorders: 3.1 %, anxiety disorders: 2.6 %, insomnia: 5.8 %) and lowest during the Omicron BA.5 period (psychotic disorders: 1.6 %, mood disorders: 1.6 %, anxiety disorders: 1.1 %, insomnia: 4.7 %).Among the three-dose participants (only in the Omicron BA.1/BA.2 and BA.5 periods), the incidence proportions of all psychiatric disorders were highest during the Omicron BA.1/BA.2period (organic mental disorders: 3.3 %, psychotic disorders: 2.7 %, mood disorders: 1.9 %, anxiety disorders: 2.2 %, insomnia: 5.3 %) and lowest during the Omicron BA.5 period (organic mental disorders: 2.3 %, psychotic disorders: 1.3 %, mood disorders: 1.0 %, anxiety disorders: 1.4 %, insomnia: 3.2 %).Table 5 presents the results of the logistic regression analyses of the associations between the number of vaccine doses and incident psychiatric disorders within three months after breakthrough infection.During the Delta period, there were no significant associations between the number of vaccine doses and incident psychiatric disorders.During the Omicron BA.1/BA.2period, three-dose participants had significantly lower odds for organic mental disorders (adjusted OR: 0.32, 95 % CI: 0.13-0.81,P = 0.016) than unvaccinated participants.During the Omicron BA.5 period, three-dose participants had significantly lower odds for mood disorders (adjusted OR: 0.48, 95 % CI: 0.25-0.9,P = 0.029),

Table 2
Incidences of psychiatric disorders within three months after breakthrough infection.Values are presented as number (%) unless otherwise indicated.

Discussion
This study evaluated the associations between COVID-19 vaccination and the occurrence of psychiatric disorders after breakthrough infection in Japan using a multi-source database comprising medical claims data, HER-SYS data, and VRS data.When compared with unvaccinated participants, vaccinated participants had significantly lower odds of developing psychotic disorders during the Delta period, as well as developing organic mental disorders, psychotic disorders, mood disorders, and insomnia during the Omicron BA.5 period.In contrast, there were no significant differences in psychiatric disorders between the vaccinated and unvaccinated groups during the Omicron BA.1/BA.2period.
To the best of our knowledge, this is the first study to examine the associations between COVID-19 vaccination and incident psychiatric disorders with consideration to the dominant circulating SARS-CoV-2 variant periods.Taquet et al. analyzed electronic health records from patients mostly in the US, and found that COVID-19 vaccination did not significantly reduce the six-month incidence of psychotic disorders, mood disorders, and anxiety disorders after breakthrough infection as part of a composite outcome (including death as the other component) between January and August 2021 among individuals aged ≥ 60 years (Taquet et al., 2022a).That study period corresponded to the Delta and Omicron BA.1/BA.2periods in the US, and their results were similar to our observations that vaccination was not associated with post-COVID psychiatric disorders during the Omicron BA.1/BA.2period.In contrast, our findings differed in that we observed a significantly reduced risk of psychotic disorders in similarly aged vaccinated adults during the Delta period.However, this may have been influenced by our small sample size of cases with psychotic disorders during that period.The Delta variant is associated with higher acute-phase disease severity (Jassat et al., 2022) than the Omicron variant, which could have led to a higher proportion of cases with severe disease during the Delta wave.Studies have also shown that disease severity during the acute phase of COVID-19 can affect the occurrence of post-COVID conditions (Jassat et al., 2022;Tsampasian et al., 2023).Similarly, a Japanese study reported that disease severity during acute COVID-19 was proportional to the frequency of subsequent psychotic disorders among hospitalized

Table 4
Incidences of psychiatric disorders within three months after breakthrough infection according to the number of vaccine doses.Values are presented as number (%) unless otherwise indicated.
F. Murata et al.
patients during the Delta period (Nakao et al., 2023).Furthermore, mRNA vaccines have shown to be effective in reducing symptomatic infections during the Delta period (Mimura et al., 2022).In this way, vaccination may have reduced the incidence of post-COVID psychotic disorders by preventing severe acute disease in patients infected with the Delta variant.
Our study found that COVID-19 vaccination was significantly associated with a lower incidence of all psychiatric disorders except anxiety disorders when the Omicron BA.5 subvariant was dominant in Japan.This represents a novel finding as previous studies did not assess the associations between vaccination and post-COVID psychiatric disorders during the Omicron BA.5 period.A UK study based on self-reported data from a mobile application found that persons infected with the Omicron variant were less likely to experience long COVID (including physical and mental symptoms) than those infected with the Delta variant (Antonelli et al., 2022).Although mRNA vaccines are less effective against the Omicron variant than the Delta variant (Mimura et al., 2022), vaccination is still able to confer protection against severe outcomes for Omicron BA.5 subvariant cases (Davies et al., 2023).A study conducted in the UK, US, and Sweden found that experiencing more than five symptoms in the first week of illness was a strong predictor of long COVID (Sudre et al., 2021).Our observation that vaccinated participants had a reduced risk of developing psychiatric disorders during the Omicron BA.5 period may have been due to the combined influence of the relatively lower incidence of long COVID for this subvariant and the vaccine-induced protection against severe outcomes during the

Table 5
Results of the logistic regression analyses of the associations between the number of COVID-19 vaccine doses and incident psychiatric disorders within three months after breakthrough infection.et al., 2022).This disparity may have been influenced by a difference in follow-up periods for the outcomes.Al-Aly et al. used a follow-up period of six months, whereas our study used a follow-up period of three months.Further investigations with longer follow-up periods in Japan may be warranted.Our analysis also evaluated the associations between the number of vaccine doses and psychiatric disorders after breakthrough infection.During the Delta period, there were no significant differences between the number of vaccine doses and incident psychiatric disorders when compared with unvaccinated participants.In contrast, the odds of organic mental disorders and psychotic disorders were significantly lower among three-dose participants than unvaccinated participants during the Omicron BA.1/BA.2period.During the Omicron BA.5 period, three-dose participants had significantly lower odds of developing psychotic disorders, mood disorders, and insomnia than unvaccinated participants, whereas four-dose participants had significantly lower odds of developing organic mental disorders, psychotic disorders, and insomnia.When analyzed according to the number of vaccine doses, some groups of participants had no incidences of psychiatric disorders after breakthrough infection.Accordingly, these results should be interpreted with caution due to the small sample sizes.To the best of our knowledge, no studies have evaluated the associations between the number of vaccine doses and incident psychiatric disorders after breakthrough infection.A previous study had examined composite outcomes comprising physical and psychiatric symptoms, and reported that the incidence of symptoms decreased with an increasing number of vaccine doses (Azzolini et al., 2022).Further research is needed to confirm if the number of vaccine doses has an impact on the development of psychiatric disorders.
The sensitivity analysis that excluded patients after September 26, 2022 produced similar results to the base case analysis.This indicated that our results were robust even after excluding the high-risk patients recorded in the HER-SYS after that date.Furthermore, the results of the sensitivity analysis that adjusted for psychiatric disorders other than the target outcome were also similar to those of the base case analysis.
This study had the following limitations.First, the study population was derived from only three municipalities, which lowers the generalizability of our findings.Moreover, the study population consisted of persons enrolled in National Health Insurance or the Latter-Stage Older Persons Health Care System.In 2020, 20.8 % of the Japanese population were enrolled in National Health Insurance, and 14.3 % were enrolled in the Latter-Stage Older Persons Health Care System (Ministry of Health, Labour and Welfare, 2023).As our study population did not include persons enrolled in employment-based insurance systems (i.e., persons employed in medium and large companies), it may not be fully representative of the Japanese population.Second, the actual SARS-CoV-2 variant infecting each COVID-19 case could not be identified from the claims data.Therefore, the COVID-19 cases during each period may have included patients who were infected by a non-dominant variant.Third, the study was conducted using claims data, which do not include detailed clinical information such as diagnostic criteria or test results.Consequently, we could not examine the differences in the severity of the initial infection or subsequent psychiatric disorders.Fourth, we were unable to account for the impact of household characteristics and socioeconomic factors on vaccination status or the occurrence of post-COVID psychiatric disorders.Socioeconomic status has been previously identified as a risk factor of long COVID (Subramanian et al., 2022).Next, more than 80 % of Japanese people aged 65-79 years cited concerns about adverse reactions as a reason for avoiding COVID-19 vaccination (Okubo et al., 2021).Living alone, low income, low educational level, and distrust toward the government and their COVID-19 policies are also associated with vaccine hesitancy.In our study population, the unvaccinated cohort had a slightly higher age and proportion of women than the vaccinated cohort during the Omicron BA.5 period, but these differences were not observed in earlier periods.We could not determine if there were any other sociodemographic differences between the cohorts that could influence vaccination status.Fifth, our study focused on older persons aged ≥ 65 years.Therefore, our results do not provide insight into the associations in younger persons.In addition, persons aged ≥ 65 years are more likely to have multiple comorbidities, and may have conditions other than those adjusted for in the model.Sixth, we could not ascertain if some participants had previously experienced COVID-19 before the observation period.Nevertheless, it is unlikely that a large proportion of cases had developed COVID-19 multiple times during the early part of the pandemic.Similarly, some patients may have experienced a psychiatric disorder seven months or more before their index dates.As we were unable to account for past diagnoses of psychiatric disorders, a portion of the outcomes may represent the recurrence of dormant conditions.
Despite the above limitations, our study was able to show that COVID-19 vaccination was associated with the reduction of psychotic disorders during the Delta period, as well as the reduction of all psychiatric disorders except anxiety disorders during the Omicron BA.5 period.

Table 6
Sensitivity analysis: results of the logistic regression analyses of the associations between COVID-19 vaccination and incident psychiatric disorders within three months after breakthrough infection during the Omicron BA.5 period after excluding patients from September 26, 2022.

Conclusion
This study evaluated the associations between COVID-19 vaccination and the three-month incidence of psychiatric disorders after breakthrough infection during the Delta and Omicron waves in Japan.Vaccination was only associated with a reduced risk of psychotic disorders during the Delta period, but appeared to lower the risk of organic mental disorders, psychotic disorders, mood disorders, and insomnia during the Omicron BA.5 period.As other variants continue to emerge and gain dominance, future studies on these associations should be conducted with consideration to the prevalent circulating variants.

Table 1
Characteristics of participants without organic mental disorders six months before the index date.
Values are presented as number (%) unless otherwise indicated.F.Murata et al.

Table 7
Sensitivity analysis: results of the logistic regression analyses of the associations between COVID-19 vaccination and incident psychiatric disorders within three months after breakthrough infection that adjusted for psychiatric disorders other than the target outcome.