Background
Henoch-Schönlein purpura (HSP) is the most common systemic vasculitis in children [
1]. The incidence of HSP in children is approximately 6–22 per 100,000 person-years [
1‐
4], which is higher than that in adult (3.4–14.3 per 100,000 person years) [
5]. Most HSP symptoms, such as temporarily palpable purpura, gastrointestinal (GI) pain, and joint pain, are self-limited; however, intestinal obstruction, central nervous system involvement, and severe nephritis can also occur [
1,
6‐
10]. The prognosis of HSP is generally good, but recurrence is common among children (recurrence rate, 2.7%–66.2%) [
1,
11‐
16]. Various predictors for recurrence, including greater joint and gastrointestinal involvement at diagnosis, history of infection, elevated erythrocyte sedimentation rate, steroid treatment, and renal manifestations, have been identified but they are inconsistent [
1,
12,
13].
Glucocorticoids used for HSP treatment do not prevent renal disease; therefore their use is controversial [
17]. However, early treatment with glucocorticoids in HSP children may reduce the intensity or mean resolution time of joint or abdominal pain [
18,
19]. Additionally, glucocorticoids should be tapered slowly to prevent the relapse of symptoms. Although previous literature has mainly focused on the risk factors for renal involvement and long-term complications in patients with HSP, no standard protocol or long-term follow-up studies are available to clarify the impact of steroid on the clinical course or subsequent recurrence of HSP.
Furthermore, the few studies that focused on the average time to second episode of recurrence revealed discrepant results (range, 1–13.5 months) [
12,
13]. Little is known regarding the third episode of recurrent HSP, and the long-term disease-free rate. However, these data are important in the decision making for adequate follow-up time. To clarify these aspects, data from a nationwide, population-based claims database, the Taiwan National Health Insurance Research Database (NHIRD), were used to investigate the mean duration between first and second HSP episodes, risk factors for recurrence, and the real-world use of steroids in patients with HSP.
Discussion
To our knowledge, this is the first population-based cohort study investigating the incidence and risk factors of recurrent HSP. Our results revealed that renal involvement, underlying allergic rhinitis, and steroid treatment for > 10 days were risk factors for HSP recurrence in children, regardless of age, sex, and income levels.
The recurrence rate (16.3%) in our study was within the range described by a previous study (2.7%–66.2%) [
1,
12‐
16]. Reasons for discrepancies between our results and previous research may be partly related to different definitions of recurrence and patient selection. In this study, we defined our recurrent events as being re-diagnosed 3 months apart from the first HSP, which was also used by previous research, including one epidemiological HSP study in Taiwan [
12,
14,
24]. Patients with a second diagnosis of HSP from both outpatient and inpatient departments were included in our study, whereas in previous studies, recurrence rates may have been underestimated if the definition of recurrent HSP was limited to patients readmitted for treatment [
12,
16], because hospitalization is not always necessary for HSP patients unless complications (such as dehydration, severe abdominal pain, gastrointestinal bleeding, joint pain with ambulatory limitations, and renal insufficiency) occur. Conversely, recurrence may have been overestimated in studies that defined their recurrent episode as after at least 2–4 weeks of asymptomatic periods, because relapses of symptoms may occur after 4–6 weeks of spontaneous resolution [
26]. Additionally, studies that defined recurrent episodes as asymptomatic for at least 2–4 weeks had the highest recurrent rate (33%–66%) [
1,
7,
13,
15]. However, in children with HSP, relapses of symptoms occurred over 4–6 weeks before spontaneous resolution even in the absence of a complicated disease course [
27].Thus, remission of symptoms for 2–4 weeks may not be adequate for HSP patients to affirm recovery. Furthermore, steroid therapy use may be another explanation for different recurrence rates between studies. In our study, the proportion of steroid use in first HSP was 40.8%. There might be an association between more steroid use and lower rates of recurrence. Lee et al. described a low recurrence rate of 5.2%, and the majority of patients (88.7%) from their study were treated with steroids [
28]. Fretzayas et al. applied the strictest policy of using corticosteroids only in 18% of patients with a severe clinical course and reported the highest recurrence rate (66%) to date [
15]. The choice of using steroids may be confounded by the disease severity. However, there was no significant difference in the initial clinical severity between the steroid-treated and non-steroid treated groups in Lee’s report. Nevertheless, such observation requires further validation in different populations.
Our finding that allergic rhinitis was an independent risk factor for HSP was in line with studies reporting an increased risk of HSP in atopic children [
29,
30]. However, no study has addressed the correlation between recurrent HSP and allergic diseases. For instance, elevated serum Th2-related cytokine levels (such as interleukin [IL]-4 and IL-5, and IgE) were reported in children with HSP [
31‐
34]. Most HSP occurred after bacterial or viral infections, insect bites, or even food allergies, suggesting that allergic reactions predispose to HSP initiation [
26,
35]. HSP is characterized by elevated serum IgA levels and vascular deposition of IgA immune complex, causing vascular necrosis mediated by IgE-sensitized mast cells [
33]. For those with allergic rhinitis, the release of Th2-related cytokines results in chronic mucosal inflammation [
36]. HSP patients with underlying allergic rhinitis may be more susceptible to chronic inflammatory status leading to a more intense immune response to causative antigens, thereby predisposing the recurrence of HSP. Moreover, the nasal mucosa (other than the bronchiole epithelium) is the first line of defense against pathogens [
37]. The inability of the ailing mucosa to remove pathogens could forecast an infectious disease, which is a critical predisposing causative factor for HSP. This may partly explain why allergic rhinitis other than asthma or atopic dermatitis was a risk factor for recurrent HSP in this study. Further studies are needed to clarify the pathogenesis of this association.
Regarding the clinical manifestations of HSP, renal involvement remains a major concern because it may lead to permanent renal function impairment. In conformity with Jauhola et al. and Alfredo et al., who both reported higher recurrences in patients with nephritis but not in those with joint symptoms and GI manifestations [
14,
38], our result revealed a higher risk of recurrence in patients with renal involvement (HR, 2.41). Additionally, the average time to recurrent episode in our patients with HSP nephritis was also shorter. Nevertheless, HSP nephritis affected 6.8% of patients in the current study, which is lower than the 10%–50% reported in previous literature [
39] Moreover, our result that HSP nephritis occurred in 22% of patients with a recurrent episode was also lower than the 58.3% reported by Alfredo et al. [
14]. Although the pathogenic mechanisms of HSP nephritis remain unclear, studies have described that galactose-deficient IgA1, recognized by antiglycan antibodies, might lead to the formation of circulating immune complexes and their mesangial deposition, resulting in renal injury among HSP patients [
7].
Our study showed a significantly increased association between steroid use for > 10 days regardless of early initiation and the risk of recurrence. Steroids might have some treatment effects on HSP through their anti-inflammatory ability. Although early steroid therapy can ameliorate acute abdominal symptoms and mitigate short-term morbidity [
18,
40‐
43], previous literature revealed conflicting results on the association between corticosteroid treatment and recurrence. Pamela et al. reported a non-significant protective effect of corticosteroids on recurrence rate [
19]. However, Trapani et al. and Calvo-Rio et al. observed an opposite finding that emphasized an increased association between early corticosteroid use and the risk of relapse [
1,
13]. In Trapani’s case series of 150 patients, steroids were prescribed only in patients with a severe presentation, including 16 patients with severe abdominal and 3 patients with severe nephropathy. Their results of increased association might be explained by the fact that patients prescribed with corticosteroids had a more severe disease manifestation [
1,
44], which is a risk factor for recurrence [
44]. Currently, steroid therapy is suggested only in patients with nephritis and severe GI symptoms (dosage, 1–2 mg/kg/day). One might argue that our results of elevated recurrence rates in those with steroid use for > 10 days were associated with their disease severity. However, this may only partly explain our result because not all recurrences were associated with more severe clinical courses, such as renal involvement. Shin et al. also pointed out that there is a distinct group of multiple recurrent non-renal HSP [
45]. Our findings indicate that the risk of recurrence might be associated with a longer duration of steroid use in certain patients. Thus, we hypothesized the mechanism that the longer use of steroid might indicate an “over-prescribing” of steroid, which might interfere with the clinical course, including the chance of recurrence. To date, there is no optimal recommendation for the duration of treatment. Expert opinions have suggested that steroids should be tapered slowly by < 25% per week to prevent relapse, thus, the overall duration of steroid treatment may easily exceeds 10 days and generally requires 4 weeks. Since the optimal duration and timing of steroid administration remains unanswered, further prospective randomized trials are needed to elucidate this possible underlying mechanism between longer-term steroid use and HSP recurrences.
The duration of follow-up in children with HSP for the early identification of possible recurrence is an important issue. In our study, the average duration between the first and second HSP episodes was 9.2 months (range, 4 months to 1 year in previous literature) [
1,
8,
46]. Prais et al. described a longer duration of 13.5 months between episodes [
12]. The reason for the longer period in Prais’s study may be because they only considered recurrences that required hospitalization in their analysis. We were the first study to report a 6.4-month average duration between the second and the third recurrences. Although there were only 6% of children with > 3 recurrences during our follow-up period, nearly half of the patients with second HSP had a third HSP event. Thus, for the early detection of recurrence, it may be worth considering following up patients with the first HSP episode for at least 9.2 months and those with a second HSP episode for another 6.4 months.
Strengths and limitations
A strength of the current study is the use of a nationwide, population-based dataset large enough to provide sufficient statistical power for detecting the association of interest. The distinctive feature of the data source strengthens the validity of the estimated disease incidences [
47]. However, there were still several limitations in this study. First, although we have adjusted for covariates such as age, sex, income levels, and comorbidities, we were unable to control for information not recorded in the NHIRD, such as serum IgE and specific IgE levels, eosinophil counts, levels of albumin, triglyceride, urine protein, urine creatinine, infection pathogens, or genetic factors. Second, this population- based study mostly assessed the Han Chinese ethnic group, and the findings may not be generalized to other populations. Third, a proper definition of ‘recurrence’ was supposed to be defined as the occurrences of new symptoms after an initial remission, requiring the resumption of immunosuppressive therapy or an increased dose. Unfortunately, details of symptoms were not recorded in the NHIRD database. Besides, not every patient with HSP needs hospitalization or was prescribed with medications. As a consequence, we were unable to use symptoms or steroid prescription to define the disease episode or disease free intervals. However, the definition of our recurrent events as at least 3-months apart from the first HSP diagnosis was in accordance to previous research, and also considered the duration of tapering steroid. Although patients that never remitted from their first episode and not treated in the NHI system between the two diagnoses 3- months apart might be included as recurrent subject, such chances were small because over 99% of the citizens utilizes health care through the NHI program.