Incidence of serious infections, herpes zoster and hepatitis B virus reactivation among patients with SLE: a cohort study using a health claims database in Japan

Abstract

Objectives

To investigate the real-world incidence of serious infections (SI), herpes zoster (HZ) and hepatitis B virus (HBV) reactivation among patients with SLE.

Methods

This retrospective, non-interventional cohort study used an insurance claims database in Japan to calculate the incidence of SI requiring hospitalisation, HZ and HBV reactivation in patients with SLE from 1 April 2017 to 31 March 2021. As an exploratory analysis, HRs and 95% CIs were estimated for clinically relevant patient characteristics for SI and HZ, adjusting for age and sex.

Results

Among 10 865 patients with SLE (female: 83.1%, median age: 53.0 years), including those who had previously experienced each complication, the incidence rates (per 100 patient-years) for SI, HZ and HBV reactivation were 4.80 (95% CI 4.53 to 5.08), 0.52 (0.44 to 0.61) and 0.37 (0.30 to 0.45), respectively. For those who had no previous history of each complication, incidence rates were 3.67 (3.44 to 3.92), 0.48 (0.40 to 0.58) and 0.25 (0.20 to 0.32), respectively. Incidence rates in specific subgroups were as follows: history of each complication (SI: 37.93, HZ: 4.19, HBV reactivation: 72.01), history of SI (HZ: 1.51, HBV reactivation: 1.75) and glucocorticoid pulse therapy during follow-up (SI: 64.23, HZ: 3.10, HBV reactivation: 4.85).

Conclusions

These results provide data on the occurrence of selected infections in SLE and may inform safety management.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Infection events are key contributors to morbidity and mortality in SLE; however, contemporary large-scale real-world estimates with current risk profiles remain limited.

WHAT THIS STUDY ADDS

• This study described the incidence of serious infections (SI), herpes zoster and hepatitis B virus reactivation in patients with SLE using a large insurance claims database in Japan.

• Incidence rates were also described across clinically relevant subgroups, including those with a prior history of the corresponding complication or SI and those defined by glucocorticoid dose categories.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These real-world incidence estimates can inform safety management in SLE.

Introduction

SLE is a systemic autoimmune rheumatic disease that results in widespread tissue and organ damage.¹ In Japan, the prevalence of SLE is between 3.7 and 37.7 per 100 000 people,² and patients with SLE often have comorbidities that further affect clinical outcomes.³

Treatment options for SLE have expanded over the last decade following the approval of several targeted therapeutics. Hydroxychloroquine is recommended for all patients with SLE.⁴ In addition, biologic therapies that target specific pathways involved in SLE pathogenesis were also introduced. These advances have enabled more targeted therapeutic approaches to managing SLE. Despite recent therapeutic progress, the burden of disease remains high.⁵

Immunosuppressive therapies for SLE can increase the risk of infection, a risk that is especially high during the induction phase following SLE diagnosis.⁶ Infections are a major contributor to both morbidity and early mortality in patients with SLE.⁷ Both herpes zoster (HZ) and serious infections (SI) can be fatal in patients treated with immunosuppressants.⁸ Patients with a history of hepatitis B virus (HBV) infection also have an increased risk of mortality.⁹

In Japan, previous studies have reported that the prevalence of HZ in patients with SLE is 43%–47%.10,12 A claims database study using the Japan Medical Data Center claims database (2005–2014) reported an HZ incidence rate of 1.59 per 100 patient-years (/100 P-Y) among patients with SLE.¹³ A cross-sectional study indicated that the prevalence of HBV in patients with SLE is approximately 17%.¹⁴ However, there remains a lack of large-scale studies since 2020 investigating the incidence of SI requiring hospitalisation, HZ and HBV reactivation and the contributing risk factors among patients with SLE in Japan, which may inform safety management that reflects the current treatment context.

Effective management of complications, including infections, remains critical for SLE treatment. Although some epidemiological data have been reported previously,10,16 they may no longer reflect the current incidence of complications and any associated risk factors. This is because the changing therapeutic landscape and new treatment recommendations⁴ have altered clinical practice. Therefore, we investigated the incidence of SI requiring hospitalisation, HZ and HBV reactivation among patients with SLE in Japan. Baseline patient characteristics for SI and HZ were also explored.

Methods

Study design and patients

This retrospective cohort study used the DeSC-IQVIA Integrated Claims Data for Japan,¹⁷ which combines data from the Health Insurance Society, the National Health Insurance and the Medical Care System for Elderly in the Latter Stage of Life and reflects estimates of real-world activity. As of 2022, this database contained information on approximately 30.8 million individuals, representing 25.3% of the Japanese population. It includes demographic, medical and pharmacy claims data, covering clinical diagnoses according to the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10) classification codes, drug prescriptions per the Anatomical Therapeutic Chemical classification and healthcare procedures. The ICD-10 codes applied in this study are listed in online supplemental tables S1 and S2.

Eligibility criteria were: patients aged ≥15 years with a diagnosis of SLE (ICD-10 codes M32.0, M32.1, M32.9 and F06.9), with ≥2 prescriptions for SLE treatment (glucocorticoids, hydroxychloroquine and biologics) and ≥1 anti-dsDNA antibody test (K-code: D014) prior to SLE treatment initiation. Patients who were treated with anifrolumab prior to the index date were excluded.

The index date was defined as the earliest date of the first prescription for an SLE treatment meeting all inclusion/exclusion criteria between 1 April 2017 and 31 March 2021 and was not intended to represent the date of SLE onset (online supplemental figure S1). The baseline period was defined as the 6 months prior to the index date. SLE duration was defined using a look-back approach as the time from the earliest recorded diagnosis date of SLE available in the patient’s claims history prior to the index date. The earliest SLE diagnosis date was identified using ICD-10 codes M32.0, M32.1, M32.9 and F06.9 (the latter for neuropsychiatric SLE only). Because the look-back period extended to dates preceding the study period, the calculated SLE duration could exceed the study period. Patients were followed from the index date and censored at the earliest of the following events: (1) onset of SI requiring hospitalisation, HZ or HBV reactivation; (2) discontinuation of SLE treatment; (3) initiation of anifrolumab treatment; (4) change or withdrawal of health insurance; (5) end of data provision from the database; or (6) the study end date (31 March 2022).

Study outcomes

The primary objective was to describe the incidence and cumulative incidence of SI requiring hospitalisation and HZ in patients with SLE treated with hydroxychloroquine, glucocorticoids, immunosuppressants or biologics. The key secondary objective was to estimate the crude incidence of HBV reactivation. Other secondary objectives included a subgroup analysis by clinically relevant factors for SI requiring hospitalisation, HZ and HBV reactivation. An exploratory objective was to estimate HRs for clinically relevant baseline patient characteristics in relation to SI requiring hospitalisation and HZ.

SI was identified by in-patient claims with infections or infestation diagnostic codes (online supplemental table S1); at least one prescription for intravenous antibiotics, antifungals, antivirals or receipt of antituberculosis drugs during hospitalisation; and a record of an immunological infection test or related tests, or a pathological diagnosis (code N), within the same month or 1 month prior to the claim with the diagnostic code.

HZ was identified with the diagnostic codes listed in online supplemental table S1 and at least one prescription of antivirals (acyclovir, vidarabine, famciclovir or valacyclovir). Serious HZ was identified with the diagnostic codes (online supplemental table S1) and prescription of antivirals on the in-patient claim. Serious HZ and COVID-19 were used to estimate the incidence of SI, but these were excluded from the exploratory analysis of clinically relevant baseline patient characteristics. Outcome definition of HBV reactivation is provided in the online supplemental methods.

The first event during the follow-up period was used to estimate the incidence rate and cumulative incidence for SI, HZ and HBV reactivation. For HZ, the onset date was the earlier date of antiviral prescription. The date for serious HZ was the first hospitalisation with the relevant diagnostic code. For SI, the onset date was the first hospitalisation date with the diagnostic codes flagged as confirmed. The onset date of HBV reactivation was the earliest date of testing or treatment. HBV-related hepatitis was treated as a baseline comorbidity and defined by ICD-10 diagnosis codes (online supplemental table S1), and HBV reactivation was identified using a claims-based algorithm described in the online supplemental methods.

Statistical methods

The full analysis set (FAS) included all patients who were eligible based on the inclusion and exclusion criteria. Continuous variables were summarised with descriptive statistics (number, mean, SD, median, range and IQR) and categorical variables with numbers and percentages.

In the primary analysis, the incidence rate (per 100 P-Y) and Poisson 95% CIs were estimated. The secondary analysis estimated the cumulative incidence over time with cumulative incidence curves and Nelson-Aalen 95% CIs. Forest plots for incidence rates and Poisson 95% CIs were generated for the FAS and each subgroup.

The incidence of SI requiring hospitalisation, HZ and HBV reactivation were also analysed in subgroups defined by clinically relevant patient characteristics, as listed in online supplemental table S3. SLE disease severity was defined using the method proposed by Garris et al,¹⁸ but with a criterion for severe disease of a glucocorticoid prescription of ≥40 mg/day instead of ≥60 mg/day to adjust the dosage per body weight in accordance with the clinical use in Japan (online supplemental table S2).

As an exploratory analysis for SI requiring hospitalisation and HZ, clinically relevant baseline patient characteristics were assessed using Cox proportional hazards models to estimate HRs and 95% CIs for the time to the first occurrence of each event, with adjustment for age and sex. These analyses were intended to descriptively assess associations with baseline characteristics and were not designed to support causal inference. Reference categories were prespecified as follows: sex (female), age (30–39 years), types of health insurance (Health Insurance Society), fiscal year (2017), duration of SLE (<1Q), comorbidities (no), medication history (no) and baseline SLE severity (mild).

Missing data or unknown data were not imputed. All analyses were conducted using SAS V.9.4.

Results

Patients

Of 52 639 adult patients with a diagnosis of SLE, 10 865 met the inclusion criteria and were included in the FAS (online supplemental figure S2). None of the patients who met the inclusion criteria met any exclusion criteria.

Most patients (9033/10 865, 83.1%) were female, and the median (range) age was 53.0 (15–98) years (table 1). The median (range) duration of SLE was 2.91 (0.00–53.02) years, with the first quantile of patients having received a diagnosis within 0.183 years prior to the index date. Most patients (93.0%) had received glucocorticoids.

Table 1. Patient characteristics.

	Patients with SLE (n=10 865)
Sex*
Male	1832 (16.9)
Female	9033 (83.1)
Age, years*
Mean±SD	53.5±17.4
Median (range)	53.0 (15–98)
Age group*
15–19 years	240 (2.2)
20–29 years	830 (7.6)
30–39 years	1414 (13.0)
40–49 years	2206 (20.3)
50–59 years	1976 (18.2)
60–69 years	1810 (16.7)
≥70s	2389 (22.0)
Health insurance type*
Health Insurance Society	5597 (51.5)
National Health Insurance	3768 (34.7)
Medical Care System for Elderly in the Latter Stage of Life	1500 (13.8)
Duration of SLE, years†
Mean±SD	6.255±7.918
Median (range) (Q1, Q3)	2.9 (0.00ؘ–53.02) (0.183, 9.755)
Comorbidities‡
Serious infection	702 (6.5)
Herpes zoster	95 (0.9)
COVID-19 infection	4 (0.0)
HBV reactivation§	51 (0.5)
HBV-related hepatitis¶	646 (5.9)
HCV-related hepatitis	296 (2.7)
Ischaemic heart disease	2015 (18.5)
Cerebrovascular disease	1118 (10.3)
Stroke	622 (5.7)
HIV infection/AIDS	34 (0.3)
Malignancy	1380 (12.7)
Hypertension	5349 (49.2)
Diabetes mellitus	4462 (41.1)
Chronic obstructive pulmonary disease	195 (1.8)
Bronchial asthma	1634 (15.0)
Depression	1024 (9.4)
Medication history*
Glucocorticoids	10 106 (93.0)
Hydroxychloroquine	1164 (10.7)
Mycophenolic acid mofetil	779 (7.2)
Tacrolimus	1759 (16.2)
Mizoribine	532 (4.9)
Azathioprine	534 (4.9)
Methotrexate	483 (4.4)
Cyclophosphamide	53 (0.5)
Belimumab	18 (0.2)
Rituximab	44 (0.4)
Janus kinase inhibitor	7 (0.1)
Trimethoprim–sulfamethoxazole	1777 (16.4)
Acyclovir	131 (1.2)
Valacyclovir	212 (2.0)
SLE disease severity‡
Mild	1355 (12.5)
Moderate	6284 (57.8)
Severe	3226 (29.7)

Data are n (%), median (range), (IQR) or mean±SD.

^*At the index date.

^†Defined using a look-back approach as the interval from the earliest recorded SLE diagnosis date available in the claims history to the index date; therefore, the duration may exceed the study period.

^‡During baseline period or all before the index date.

^§Indicates patients who experienced a claims-defined HBV reactivation (definition in the online supplemental methods).

^¶Indicates a baseline comorbidity defined by International Statistical Classification of Diseases and Related Health Problems 10th Revision codes (online supplemental table S1).

HBV, hepatitis B virus; HCV, hepatitis C virus; Q, quartile.

Serious infections

There were 1200 SI events, with an incidence rate (95% CI) of 4.80/100 P-Y (4.53 to 5.08) (figure 1). The cumulative incidences (95% CIs) of SI were 7.72% (7.21% to 8.26%) at 1 year, 10.40% (9.77% to 11.06%) at 2 years and 12.71% (11.98% to 13.48%) at 3 years (online supplemental figure S3A). The incidence rate (95% CIs) in patients without a history of SI at baseline was 3.67/100 P-Y (3.44, 3.92).

Figure 1. Incidence of serious infections overall and by subgroups. *First quartile: 0.183 years; second quartile: 2.910 years, third quartile: 9.755 years. CI, confidence interval; HBV, hepatitis B virus; HCV, hepatitis C virus; IR, incidence rate; MMF, mycophenolate mofetil; P-Y, patient-years; Q, quartile.

Figure 1 shows incidence of SI overall and by subgroups based on clinical characteristics. The incidence rate in patients with a history of SI at baseline was 37.93/100 P-Y (95% CI 33.83 to 42.39). In subgroups, notable incidence rates (≥10) were observed in aged ≥70 years (12.57/100 P-Y (95% CI 11.25 to 14.00)), short SLE disease duration (<1Q) (12.86 (10.97 to 14.97)), histories of SI, hepatitis C virus (HCV)-related hepatitis (11.75 (8.95 to 15.16)) or malignancies (11.75 (9.32 to 11.93)), or chronic obstructive pulmonary disease (11.67 (8.26 to 16.02)), a history of treatment with acyclovir (14.55 (10.14 to 20.24)) and treatment with pulsed glucocorticoid therapy (64.23 (57.33 to 71.73)) or glucocorticoid dose >225 mg/30 days (13.58 (12.68 to 14.53)).

In the exploratory Cox analysis, several baseline patient characteristics showed HRs whose 95% CI lower limits exceeded 1 for the first occurrence of SI, including a history of SI (adjusted HR: 7.07 (95% CI 6.18 to 8.10)), a history of rituximab use (4.68 (2.96 to 7.38)) and severe SLE disease severity at baseline (4.09 (3.09 to 5.41)) (online supplemental figure S4). Further analyses revealed no follow-up subgroups in which HRs whose 95% CI lower limits exceeded 1 for these characteristics relative to the respective reference categories (data not shown).

The cumulative incidences of SI by subgroup are shown in online supplemental table S4. The cumulative incidence (95% CIs) of SI in patients without a history of SI at baseline was 5.39% (4.94% to 5.88%).

Herpes zoster

There were 138 HZ events, with an incidence rate (95% CI) of 0.52/100 P-Y (0.44 to 0.61) (figure 2). The cumulative incidences (95% CIs) of HZ were 0.74% (0.58% to 0.93%) at 1 year, 1.37% (1.14% to 1.66%) at 2 years and 1.62% (1.35% to 1.94%) at 3 years (online supplemental figure S3B). The incidence rate (95% CI) in patients without a history of HZ at baseline was 0.48/100 P-Y (0.40 to 0.58).

Figure 2. Incidence of herpes zoster overall and by subgroups. *First quartile: 0.183 years; second quartile: 2.910 years, third quartile: 9.755 years. CI, confidence interval; HBV, hepatitis B virus; HCV, hepatitis C virus; IR, incidence rate; MMF, mycophenolate mofetil; P-Y, patient-years; Q, quartile.

Figure 2 shows incidence of HZ overall and by subgroups. The incidence rate in patients with a history of HZ at baseline (n=95) was 4.19/100 P-Y (95% CI 2.01 to 7.70). In subgroups, notable incidence rates (≥1) were observed in aged 15–19 years (1.29/100 P-Y (95% CI 0.52 to 2.65)), histories of SI (1.51 (0.91 to 2.36)), HBV-related hepatitis (1.10 (0.62 to 1.82)), or HCV-related hepatitis (1.23 (0.49 to 2.53)), or a history of treatment with acyclovir (1.83 (0.59 to 4.27)) and treatment with pulsed glucocorticoid therapy (3.10 (1.74 to 5.12)), or glucocorticoid dose >225 mg/30 days (1.01 (0.79 to 1.28)).

In the exploratory Cox analysis, several baseline patient characteristics showed HRs whose 95% CI lower limits exceeded 1 for the first occurrence of non-serious HZ, including a history of HZ (adjusted HR: 8.94 (95% CI 4.67 to 17.11)); age 15–19 years (2.91 (1.18 to 7.14)); National Health Insurance (1.97 (1.30 to 2.99)); histories of SI (2.57 (1.53 to 4.31)), HBV-related hepatitis (1.89 (1.06 to 3.37)) or malignancy (1.60 (1.00 to 2.56)); and use of trimethoprim-sulfamethoxazole (1.57 (1.05 to 2.33)), acyclovir (3.74 (1.53 to 9.15)) or hydroxychloroquine (1.71 (1.07 to 2.71)) (online supplemental figure S5). Further subgroup analyses indicated that among patients aged 15–19 years, in whom the number of patients was limited, HRs whose 95% CI lower limits exceeded 1 were observed for several clinically relevant patient characteristics, including severe SLE disease severity (adjusted HR: 8.79 (95% CI 1.77 to 43.73)), glucocorticoid use (3.57 (1.44 to 8.86)), glucocorticoid dose >150 mg/30 days (3.33 (1.26 to 8.80)), >150–225 mg/30 days (8.18 (1.64 to 40.91)) and ≤225 mg/30 days (7.39 (1.84 to 29.57)), use of hydroxychloroquine (10.96 (2.72 to 44.06)) and use of two different immunosuppressants (7.55 (1.52 to 37.45)) (online supplemental table S5). For patients with a history of HZ, there were no follow-up subgroups with HRs whose 95% CI lower limits exceeded 1 relative to the respective reference categories (data not shown).

The cumulative incidences of HZ by subgroup are shown in online supplemental table S4). The cumulative incidence (95% CIs) of HZ in patients without a history of HZ at baseline was 0.70% (0.54% to 0.89%).

HBV reactivation

There were 99 events of HBV reactivation, with an incidence rate of 0.37/100 P-Y (95% CI 0.30 to 0.45) (figure 3). Cumulative incidences (95% CIs) of HBV reactivation were 0.87% (0.71% to 1.07%) at 1 year, 0.93% (0.76% to 1.14%) at 2 years and 0.97% (0.79% to 1.19%) at 3 years (online supplemental figure S3C). The incidence rate (95% CIs) in patients without a history of HBV reactivation at baseline was 0.25/100 P-Y (0.20 to 0.32).

Figure 3. Incidence of HBV reactivation overall and by subgroups. *First quartile: 0.183 years; second quartile: 2.910 years, third quartile: 9.755 years. CI, confidence interval; HBV, hepatitis B virus; HCV, hepatitis C virus; IR, incidence rate; MMF, mycophenolate mofetil; P-Y, patient-years; Q, quartile.

Figure 3 shows incidence of HBV reactivation overall and by subgroups. The incidence rate in patients with a history of HBV reactivation at baseline (n=51) was 72.01/100 P-Y (95% CI 48.93 to 102.22). In subgroups, notable incidence rates (≥1) were observed in male (1.02/100 P-Y (95% CI 0.73 to 1.39)), histories of SI (1.75 (1.09 to 2.65)), HBV-related hepatitis (4.42 (3.34 to 5.74)), malignancy (1.38 (0.98 to 1.90)), a history of treatment with acyclovir (4.96 (2.64 to 8.48)) and treatment with pulsed glucocorticoid therapy (4.85 (3.07 to 7.28)).

The cumulative incidences of HBV reactivation by subgroup are shown in online supplemental table S4. The cumulative incidence (95% CIs) in patients without a history of HBV reactivation at baseline was 0.58% (0.45% to 0.75%).

Discussion

The main findings of this study were the crude incidence rates of SI, HZ and HBV reactivation among patients with SLE using a large-scale insurance-based claims database in Japan. These data should provide useful information for clinicians managing SLE in real-world practice and could help inform the safety profiles of new therapeutics.

It is interesting to note that elevated incidence rates of SI, HZ and HBV reactivation were observed with increased glucocorticoid doses as well as glucocorticoid pulse therapy during the follow-up period. These findings underscore the importance of minimising glucocorticoid exposure in SLE treatment. This is consistent with the recent European Alliance of Associations for Rheumatology recommendation which states that glucocorticoid doses should be minimised to ≤5 mg/day,⁴ and with the SLE guideline from the American College of Rheumatology, which strongly recommends tapering glucocorticoid doses to ≤5 mg/day within 6 months.¹⁹

This study applied inclusion criteria reflecting clinical practice, including both treatment of SLE and anti-dsDNA antibody measurements in addition to diagnosis of SLE to identify eligible patients. The demographic composition, including age and sex, was consistent with expectations based on the Number of Recipient Certificates Issued for Specific Disease Treatment in Japan. The proportion of patients aged <60 years among holders of specific medical expenses in Japan was 66.2%–66.7% during fiscal years 2017–2020²⁰ and 61.4% in this study, suggesting similarities between the two populations. The median age and sex ratio in this study were also similar to those reported previously.^{21 22}

The prescription rate of hydroxychloroquine in this study was lower at baseline (10.7%) than that reported in recent Japanese nationwide database studies (21.4%).²³ This discrepancy likely reflects differences in the study period and study design: patients in this study were indexed from 2017, shortly after hydroxychloroquine was approved for SLE in Japan (2015) and included a substantial proportion of patients who were already receiving treatment at the index date (5854 of the 10 865 patients were enrolled in the first half of fiscal year 2017, see online supplemental figure S2), whereas the nationwide studies reported use during later years (April 2019 to March 2020).²³ Many of these patients likely initiated SLE treatment before hydroxychloroquine was approved for use in Japan, which may have contributed to the lower proportion of hydroxychloroquine use observed at the index date in our study. Notably, hydroxychloroquine use increased during follow-up in this study (to approximately 24%; data not shown), which is closer to the previously reported levels. In contrast, glucocorticoid use (93.0%) was similar to that reported in nationwide studies (89.1%).^{23 24} The proportions of mycophenolate mofetil, tacrolimus and cyclophosphamide use at the index date may appear lower than expected, given the frequency of lupus nephritis in SLE.²⁵ However, this likely reflects the study design and cohort composition, rather than an undertreatment of active lupus nephritis. The index date was defined as initiation of any SLE-related treatment during the study period and was not intended to represent the initiation of therapy for lupus nephritis. In addition, the cohort included a substantial proportion of patients already receiving treatment for SLE at the index date, which may influence the proportion of medications received at cohort entry. Although the prescription rate of mycophenolate mofetil and tacrolimus in this study was lower at baseline (7.2% and 16.2%, respectively) than that reported in recent Japanese nationwide database studies (12.0% and 22.9%, respectively),²⁴ mycophenolate mofetil use increased during follow-up in this study (to approximately 12.4%; data not shown), becoming more consistent with the levels previously reported.²⁴ The relatively older age distribution of our cohort (median age of 53 years, with approximately 22% aged ≥70 years) may also have contributed to a lower than expected use of intensive immunosuppressive therapy at cohort entry, as older patients with SLE are less likely to receive immunosuppressants such as mycophenolate mofetil or tacrolimus in routine clinical practice.²³ These findings support the interpretation that the lower proportion of these immunosuppressants observed at baseline reflects cohort timing rather than a lack of treatment. The proportions of moderate and severe SLE disease severity (57.8% and 29.7%, respectively) were comparable to a previous study using the Japan Medical Data Center claims database, although the definition of SLE severity was somewhat different.²⁶ Overall, the patient population in this study may be representative of the broader SLE population in Japan, supporting the generalisability of the results. However, the early (year 1) cumulative incidence may be attenuated by the inclusion of prevalent, stable cases that were enrolled at the start of the indexing period.

In this study, the overall incidence of SI was 4.80/100 P-Y among patients with SLE in Japan. Reported incidence rates in other studies include 4.34/100 P-Y in a nationwide study in Taiwan²⁷ and 6.16/100 P-Y in a single-centre cohort study in India.²⁸ Among patients with SLE with lupus nephritis, an incidence of 21.53/100 P-Y was reported in a US-based healthcare database study.²⁷ Regarding the incidence in patients receiving different SLE medications, the incidence rates for SI include 7.49/100 P-Y for patients receiving hydroxychloroquine in a health claims database study in Japan,²⁹ and 2.6/100 P-Y for patients receiving both glucocorticoids and antimalarials in a US integrated healthcare delivery system (Kaiser Permanente Northern California).³⁰ In the same US cohort, prednisolone equivalent dose ≤15 mg/day and >15 mg/day of glucocorticoid-treated patients without antimalarials had an incidence of 5.6/100 P-Y and 8.9/100 P-Y, respectively (online supplemental figure S6).³⁰ Overall, these incidence rates by treatment type are generally consistent with our findings.

Glucocorticoid treatment and immunosuppressive medications used as part of standard care to control SLE disease activity can increase the risk of infections, which are a major cause of both morbidity and early mortality in these patients.729 31,33 In SLE management, SI may trigger SLE disease flares, and these flares may be as severe as, or greater than, flares instigated by other causes.³⁴ Therefore, the findings from this study may inform clinicians about the occurrence of infections and highlight the importance of both treatment to control SLE disease activity and appropriate safety management.

The overall incidence of HZ in this study, including patients with a history of HZ, was 0.52/100 P-Y. In contrast, previous studies in Japan reported incidence rates of new-onset HZ only, with 1.59/100 P-Y in a database study¹³ and 5.37/100 P-Y in a single-centre cohort study.³⁵ In this study, we applied the outcome definition for HZ by both diagnostic codes and at least one antiviral prescription in order to increase precision compared with earlier studies. Because HZ is diagnosed and treated under insurance coverage in Japan, we believe that the incidence of HZ requiring treatment, as described in this database study, accurately reflects routine clinical practice. In other countries, variations in the definition of HZ have resulted in a broad range of reported incidence rates (0.64–3.77/100 P-Y) (online supplemental figure S7).36,41

A history of HZ was observed as a clinically relevant patient characteristic for non-serious HZ in the exploratory Cox analysis, in line with the incidence rate of HZ among patients with a history of HZ. However, further subgroup analyses for the follow-up characteristics did not reveal any other characteristics showing HRs with lower 95% CI limits exceeding 1, including SLE disease severity or medication use. These exploratory HR estimates should be interpreted with caution, as residual and unmeasured confounding cannot be excluded in claims-based data, and the analyses were not intended to establish independent or causal effects. Nevertheless, our descriptive findings are consistent with previous studies in rheumatoid arthritis patients with a history of HZ after the initiation of biologics⁴² and support consideration of recombinant zoster vaccination to reduce HZ risk in patients with SLE.⁴³

The overall incidence of HBV reactivation in this study including patients with a history of HBV reactivation was 0.37/100 P-Y. To our knowledge, there are no previous reports using a Japanese health insurance claims database. In a non-claims-based clinical cohort study from Taiwan, the incidence rate has been reported as 1.76/100 P-Y among patients with SLE.⁹ In patients with rheumatoid arthritis, the incidence rates have been reported as 1.93/100 P-Y in a Japanese prospective multicentre cohort⁴⁴ and 10.3/100 person-years in a Taiwanese cohort.⁴⁵ Descriptively, notable patterns of HBV reactivation were observed among patients receiving intensive immunosuppressive treatments, including higher-dose or pulse-dose glucocorticoids, during follow-up. These observations are consistent with previous reports highlighting the clinical importance of careful HBV monitoring in patients undergoing intensive immunosuppressive treatments.⁴⁶ In Japan, all patients receiving immunosuppressants are closely monitored and antiviral therapy is rapidly initiated when HBV DNA is detected, according to the relevant guideline.⁴⁷ The prevalence of previous HBV infection decreases sharply in younger cohorts in Japan,⁴⁸ which may help explain the observed rates in this study. In addition, previous studies have defined HBV reactivation by the detection of HBV DNA in blood samples,^{9 45} whereas this study used a claims-based definition.

Limitations

This study has several limitations. The claims database did not include information on SLE disease activity (eg, Systemic Lupus Erythematosus Disease Activity Index 2000), so disease activity in this study was inferred from glucocorticoid dosage and SLE disease severity based on a previous study.¹⁸ In addition, the temporal relationship between treatment exposure during follow-up and infection events could not be fully clarified. Treatments such as high-dose glucocorticoids, antiviral prophylaxis or rituximab may reflect changes in underlying disease activity or clinical status that preceded the infection events. Vaccination history was unavailable, limiting assessment of immunisation effects on infection risk. Potential confounding and missing data may result from the absence of some patient-reported outcomes and laboratory results in claims records. The database included a higher proportion of patients aged >75 years (approximately 20%–25%) compared with those aged <75 years (10%–15%), potentially leading to an under-representation of younger patients; incidence rates were therefore calculated within age subgroups. Due to the study design, the cohort included both newly diagnosed patients and those already treated for SLE. As over half of the cohort were enrolled in early 2017, this likely introduced many stable prevalent cases into year 1, which may have made the early cumulative incidence appear lower than in a cohort that only included newly diagnosed patients. Additionally, the 6-month baseline window could have missed prior events, further reinforcing this potential attenuation. The lack of mortality data precluded adjustment for competing risk of death, which may result in overestimation of cumulative incidence rates. Information on investigational drug use was also unavailable. Censoring due to withdrawal from health insurance (eg, employment change or relocation) may have led to under- or overestimation of cumulative incidence. In addition, approximately 30% of the patients were followed for less than 1 year, which may have resulted in an underestimation of the incidence rates. However, these patients were retained to reflect real-world clinical practice. The history of comorbidities was assessed only during the 6-month baseline period; therefore, any conditions or events occurring before the baseline period were not evaluated.

Conclusions

This study describes the incidences of SI, HZ and HBV reactivation among patients with SLE. The results of this study provide clinically relevant information that may support clinicians in recognising infection occurrence and reinforce the importance of appropriate safety management alongside disease control for SLE.

Data availability statement

Data may be obtained from a third party and are not publicly available.