Article Text
PDF
Abstract
Objective To clarify the necessity of acetylsalicylic acid (ASA) administration combined with intravenous immunoglobulin (IVIG) therapy in the treatment of acute Kawasaki disease.
Design Retrospective cohort study.
Setting Multicentre.
Participants This study included 735 patients with Kawasaki disease aged ≤10 years and hospitalised between 4 and 10 days of illness in eight Japanese hospitals from January 2016 to December 2020.
Exposures High-dose (HD) ASA was administered with initial IVIG to 333 patients in 6 hospitals (HD group). ASA was not administered routinely to 402 patients in the other two hospitals, and low-dose ASA was only administered when patients developed coronary artery lesions or pericardial effusion (non-HD group).
Primary and secondary outcome measures The primary outcome was the presence of coronary artery lesions, defined as a coronary artery diameter >+2.5 SD of body surface area within 1 month of onset. The secondary outcome was responsiveness to the initial IVIG therapy. Adjusted risk ratios for the outcomes were calculated using modified Poisson regression models. Bayesian analysis was conducted to estimate the posterior probability of the treatment effect of HD ASA under several prior distributions.
Results The incidence of coronary artery lesions was not significantly higher in the HD group than in the non-HD group (12/333 (3.6%) vs 15/402 (4.0%)). The proportion of non-responders to initial IVIG was similar between the two groups (HD group: 78/333 (23%); non-HD group: 83/402 (22%)). In the Bayesian analysis, considering a difference of ≤2% to be of no clinical importance, there was only a 9.3% chance of reduced risk of coronary artery lesions in the HD group compared with the non-HD group even with a strongly enthusiastic prior for HD treatment.
Conclusions Compared with HD ASA treatment, treatment without ASA in the acute phase of Kawasaki disease was not associated with increased complications from Kawasaki disease.
- cardiology
Data availability statement
No data are available. Due to the nature of this study, the participants did not agree for their data to be shared publicly; therefore, supporting data are not available.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
In the acute phase of Kawasaki disease, medium-dose to high-dose acetylsalicylic acid is concomitantly administered with intravenous immunoglobulin. However, the efficacy of acetylsalicylic acid administration in the treatment of Kawasaki disease complications is controversial.
WHAT THIS STUDY ADDS
Zero acetylsalicylic acid administration during the acute phase of Kawasaki disease did not significantly elevate the risk of complications compared with the administration of high-dose acetylsalicylic acid.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The study findings suggest the reconsideration of current guidelines recommending the administration of acetylsalicylic acid in acute Kawasaki disease.
Introduction
Kawasaki disease (KD) is a disease of unknown aetiology that causes vasculitis of small-sized to medium-sized arteries such as the coronary artery. Coronary artery lesions (CALs) develop in 15%–25% of children with untreated KD1; moreover, KD remains a leading cause of acquired heart disease in children. Kusakawa et al reported that acetylsalicylic acid (ASA), as a treatment for the acute phase of KD, might have a better effect on reducing CALs than flurbiprofen and prednisolone plus dipyridamole.2 In addition, Koren et al demonstrated the efficacy of high-dose (HD) ASA in shortening the febrile period of KD.3 Originally, Furusho et al showed that intravenous immunoglobulin (IVIG) therapy alone was superior to HD ASA therapy for KD4; however, HD ASA was concomitantly administered with IVIG therapy.
ASA has antiplatelet and anti-inflammatory effects. Low-dose ASA (3–5 mg/kg/day) inhibits cyclooxygenase (COX)-1 and exerts antiplatelet effects. HD ASA (80–100 mg/kg/day) is believed to be necessary for inhibiting COX-2 and inducing anti-inflammatory effects.
Recently, some studies have reported negative results regarding the anti-inflammatory effects of HD ASA. Dallaire et al reported that low-dose ASA was not inferior to HD ASA in reducing the incidence of CALs.5 Four meta-analyses revealed that low-dose ASA had treatment outcomes similar to those of HD ASA, particularly on CAL-related morbidity.6–9 In addition, ASA results in rare but possibly severe side effects such as bleeding, Reye’s syndrome,10 liver dysfunction and hearing loss.11 The Japanese KD guidelines recommend the administration of medium-dose ASA (30–50 mg/kg/day), considering the hepatic vulnerability of Japanese people to ASA.12 Hence, the efficacy of ASA administration during IVIG therapy for patients with KD, especially in the prevention of CALs, remains controversial.
ASA may not reduce KD complications and may further increase its adverse effects. Few studies have compared HD ASA to no antiplatelet treatment as the initial treatment for KD; these studies showed non-superiority of HD ASA to no antiplatelet treatment.13–16 A systematic review confirming the efficacy of ASA in the treatment of KD did not provide sufficient evidence to determine whether the omission of antiplatelets from KD treatment provided equivalent treatment outcomes.17 Therefore, this multicentre retrospective cohort study aimed to investigate whether a no-ASA treatment modality could affect the treatment outcomes of KD such as the incidence of CALs or unresponsiveness to initial IVIG.
Methods
Study design and setting
This multicentre retrospective cohort study of patients with KD explored the effects of ASA in the acute phase of KD. Eight hospitals with different treatment policies for ASA administration were included. In six hospitals, HD ASA (targeted 30–50 mg/kg/day) was concomitantly administered with initial IVIG. In the remaining two hospitals, ASA was not routinely administered and low-dose ASA (3–5 mg/kg/day) was administered only when patients developed CALs or pericardial effusion. This study adhered to the reporting guidelines of Strengthening the Reporting of Observational Studies in Epidemiology. There was no patient and/or public involvement in this research.
Participants
Medical records of the participating hospitals were reviewed, and patients diagnosed with acute KD and hospitalised between January 2016 and December 2020 were identified. Patients diagnosed after 2021 were excluded because there many paediatric patients were infected with SARS-CoV-2 in Japan after ‘the fourth wave’ (July to September 2021).
This study included patients if (1) they were aged ≤10 years, (2) they presented with at least four of the six principal features (fever, change in extremities, polymorphous exanthema, conjunctival injection, changes in the lips and oral cavity and cervical lymphadenopathy) and (3) the initial IVIG treatment started between days 4 and 10. Patients were excluded if the size of their coronary arteries was >+2.0 SD18 at the time of diagnosis or if a corticosteroid was administered alongside the initial IVIG. Patients with a history of haemodynamically significant structural heart disease or KD were excluded.
Exposure
The primary exposure status was defined according to the administration of ASA at the time of initial IVIG treatment. Patients who received ASA ˃10 mg/kg/day were categorised as the HD group; those who did not receive ASA at all or received ASA ˂10 mg/kg/day were categorised as the non-HD group. Patients who were administered antiplatelet drugs other than ASA (eg, flurbiprofen) were excluded.
Outcomes and covariates
This study’s primary outcome was the presence of CALs, including transient dilatation of the coronary arteries, within 1 month of disease onset. Patients were considered to present with CALs when two-dimensional echocardiography showed that internal lumen diameters were >2.5 SD for body surface area, or there were apparent aneurysmal changes in the coronary artery in the medical records. Generally, the echocardiogram was performed at least at the time of diagnosis, within 48 hours after diagnosis and initial treatment, and approximately 10 days after disease onset, although it should be noted that these intervals were not predetermined. The secondary outcome was responsiveness to the initial IVIG treatment. Patients were defined as refractory to the initial IVIG treatment if they received a second-line treatment such as additional IVIG with or without systemic corticosteroid administration or other immunosuppressive therapies.
The following data were collected immediately before the initial IVIG treatment: white cell count, percentage of neutrophils, haematocrit, platelet (PLT) count and serum concentrations of C reactive protein (CRP), sodium, aspartate aminotransferase (AST), alanine aminotransferase, total bilirubin and albumin.
Sample size
This study was considered an exploratory observational study; however, the sample size was calculated assuming it was a non-inferior trial. Based on previous studies,19 the incidence probability of the primary outcome was expected to be 15%; the inferior margin was set to 7.5%. With a one-sided type I error of 10% and a power of 80%, a sample size of 205 patients per group was required to detect non-inferiority. A target sample size of 230 patients per group was determined assuming a 10% drop-out rate.
Statistical analysis
Patient characteristics are summarised as numbers and percentages for categorical variables and means and SD for continuous variables. The adjusted risk ratios (aRRs) for the outcomes were calculated using modified Poisson regression models with robust variance estimator including the following covariates: age, sex, clinical day of initial IVIG treatment, percentage neutrophil count, and PLT, CRP, sodium, and AST levels. Variables, except for sex, were selected on the Kobayashi score,20 which is widely recognised as a high-risk indicator for the development of CALs in KD. As patients at high risk of developing CALs could be more likely prescribed with ASA, these variables were included as potential cofounders. In the regression analysis, the Wald test was used to obtain two-tailed p values, and a p<0.05 was considered statistically significant.
We also implemented propensity score matching and inverse probability of treatment weighting (IPTW). The propensity score was estimated using a logistic regression model including the above-mentioned covariates. We conducted one-to-one matching without replacement using a calliper width of 0.2 of SD of the logit of the propensity score. Using the matched cohort, the risk ratio was calculated using the Mantel-Haenszel method. In IPTW approach, the weights were stabilised, and the risk ratios were obtained using the modified Poisson regression models.
As a sensitivity analysis, we conducted a Bayesian analysis21 in which prior information based on our knowledge was combined with the present data to update the belief on the treatment effect of ASA. The aim of this analysis was to assess the clinical impact of HD ASA, particularly in populations believed to have a lower risk of CALs that included children aged between 3 months and 10 years with a low Kobayashi score (4 points). This targeted approach was also intended to mitigate the potential variability in criteria for initiating IVIG treatment across different institutions. By varying the degree of clinician belief in the efficacy of ASA, we aimed to validate its potential therapeutic significance. We assumed the following model:
n_HD~Binomial (θ_HD, N_HD).
n_non-HD~Binomial (θ_non-HD, N_non-HD).
θ_HD~Beta (80 920).
θ_non-HD~Beta (α,β).
where n is the number of patients who developed CALs within 1 month among N patients in each group. The parameter θ represents the probability of developing CALs, which was assumed to have a beta distribution.
Based on a national survey of patients with KD in Japan,22 the reference prior to the distribution of the HD group was assumed to be a beta distribution with parameters of 80 and 920, which was equivalent to the evidence that 80 of 1000 patients developed CALs. For the priors of the non-HD group (θ_non-HD), four different distributions were set to represent the researchers’ scepticism or enthusiasm for using HD ASA treatment: (1) a minimally informative prior (α=1/2, β=1/2), which represents no prior belief about how much likely CALs develop in the non-HD group, (2) a sceptical prior (α=80, β=920), which corresponds to a belief that the risk of CALs in the non-HD group is 8% in a trial of 1000 patients similar to the reference prior in the HD group, (3) moderately enthusiastic (α=88, β=912) and (4) strongly enthusiastic (α=96, β=904) priors, which were equivalent to a trial that 1000 patients in the non-HD group showed 1.1 and 1.2 higher risk of CALs than did the HD group, respectively. These prior beliefs were updated using data from patients aged between 3 months and 10 years with a low Kobayashi score (≤4 points) and were not considered to be at a high risk of developing CALs.
Four separate sampling sequences were set, each consisting of 1000 random samples (including 500 samples discarded for convergence). Sampling convergence was evaluated using Gelman-Rubin statistics and by visually inspecting a trace plot. Among 2000 samples (4 chains with 500 samples in each), we counted samples in which the posterior probability of developing CALs in the non-HD group (θ_non-HD) was higher than that in the HD group (θ_HD) by several degrees of clinical importance φ (0%, 1%, 2% and 3%), which was analogous to the risk difference with various inferior margins φ. R software V.4.1.1 (R Foundation for Statistical Computing, Vienna, Austria) was used for all analyses. The probabilistic programming language Stan (Stan Development Team) was used for the Bayesian analysis.
Results
Participant characteristics
Of the 1233 eligible patients, 806 without missing data necessary for inclusion were registered (figure 1). Sixty-one patients with CALs on admission were excluded. Additionally, 71 patients who were administered antiplatelet medications other than ASA were excluded. The registration forms for some patients who did not satisfy the inclusion criteria were incomplete; therefore, the precise reason for their exclusion was unknown. Finally, 735 patients were included in this analysis.
Patient flow. ASA, acetylsalicylic; CALs, caronary artery lesions; HD, high dose; IVIG, intravenous immunoglobulim.
Of the 735 patients, 333 were administered HD ASA. Among the 402 patients not administered with HD ASA, 324 did not receive antiplatelet agents and 78 received low-dose ASA. Table 1 shows the baseline characteristics of the patients and their outcomes. Additionally, online supplemental table S1 presents a detailed comparison of patient characteristics based on the use of low-dose ASA within the non-HD group.
Supplemental material
Supplemental material
Patient characteristics
Multivariable regression and propensity score analysis
Table 2 summarises the results of the multivariable analysis for the risk of developing CALs. Compared with the HD group, the aRR in the non-HD group was 1.12 (95% CI 0.83 to 1.51). In the propensity score analyses, neither the matching nor the IPTW approach revealed a statistically significant difference between the HD and non-HD groups despite a higher point estimate being observed. The distribution of propensity score and the balanced of covariates before and after matching are shown in online supplemental figure S1.
Results of multivariable analysis for the risk of developing coronary artery lesions within 1 month
Table 3 shows that there was no significant difference in unresponsiveness to the initial IVIG treatment between the non-HD and HD groups (aRR 1.09; 95% CI 0.92 to 1.29).
Results of multivariable analysis for unresponsiveness to IVIG
Bayesian analysis
In the Bayesian analysis, we used the data of patients at low risk of developing CALs (age, 3 months to 10 years; Kobayashi score ≤4 points); 8 of the 204 (3.9%) patients in the HD group and 7 of the 307 (2.3%) patients in the non-HD group developed CALs. The posterior probability distributions with four different priors are graphically shown in online supplemental figure S2. The posterior probability distribution with a minimally informative prior (online supplemental figure S2A) showed a higher incidence of CALs in the HD group, demonstrating that the incidence of CALs in the HD group was higher than that in the non-HD group based on the data used. Even with a moderately enthusiastic prior belief for HD treatment, the risk of developing CALs in the non-HD group was 10% higher, and the posterior probability of both groups was similarly distributed (online supplemental figure S2C).
Table 4 summarises the posterior probability that the incidence of CALs in the non-HD group was higher than that in the HD group by a certain degree φ (0%, 1%, 2% and 3%), which demonstrates clinical importance. For example, a probability of (
− 
) >2% implies the likelihood that non-HD treatment, compared with HD treatment, increases the risk of developing CALs by ˃2%. If we consider that a difference of ≤2% is not of clinical importance (ie, an inferior margin of 2%), there was only a 9.3% chance that HD treatment reduced the risk of CALs than did non-HD treatment, even with a strongly enthusiastic belief for HD treatment. The results using the minimally informative prior were all 0% because it is essentially equivalent to judgement based solely on the data.
Difference between posterior probability in the HD and non-HD groups
Discussion
HD ASA exerts anti-inflammatory effects by inhibiting COX-2. In contrast, low-dose ASA inhibits COX-1 and exerts an antiplatelet effect. ASA is 170-fold less effective at inhibiting COX-2 than it is at inhibiting COX-1.23 Therefore, there is a difference in the dosage required to obtain each effect. Before the administration of IVIG for KD, HD ASA was used for KD treatment because of its anti-inflammatory effects. However, some studies have not shown the superiority of HD ASA over low-dose ASA in terms of the incidence of CALs.6–9 24 In this study, the majority of patients in the non-HD group (80.5%; 324/402) did not receive ASA, while 54 patients received low-dose ASA after the initial IVIG treatment owing to the development of CALs or pericardial effusion. The no-ASA modality showed non-inferiority to routinely administered HD ASA with respect to the incidence of CALs.
In addition, this study conducted a Bayesian analysis to estimate the posterior probability of the treatment effect of HD ASA under various reference priors. Even with a strongly enthusiastic prior for HD treatment that the risk ratio of CALs was 1.2 in the non-HD group compared with the HD group, this study showed the posterior probability that HD superiority was only 69.0%. This analysis showed that this study’s results are considered robust, even though clinicians strongly prefer HD ASA.
Regarding the proportion of unresponsiveness to initial IVIG therapy, there was no inferiority in the non-HD group compared with the HD group, which is consistent with the results of previous studies and meta-analyses.5 7–9 14 16 The duration from treatment to the absence of fever was significantly shorter in the non-HD group than in the HD group; this result is contrary to that of two previous studies.14 25 ASA may cause various adverse events. Regarding the adverse events associated with ASA, abdominal discomfort is the most common, and the incidence of liver dysfunction was 0.9%.26 Although this study found no significant difference in the frequency of adverse events between both groups, the sample size might not have been large enough. It is uncertain whether low-dose to medium-dose ASA causes severe adverse effects related to ASA; hence, the risks of administering ASA should be considered.
This study had several limitations. First, this was a retrospective observational study, which was not the optimal design for evaluating non-inferiority. Additionally, there was no pre-established protocol for treatment and evaluation methods within and between facilities, potentially introducing bias or residual confounders related to this lack of standardisation. To address these limitations, we employed several analytical methods, including Bayesian analysis, to validate the conclusions drawn. While the subjectivity of a Bayesian approach provides readers with greater interpretative flexibility, it might also leave room for interpretative arbitrariness. Second, this study included only patients with KD who presented with at least four principal features and began treatment after 4 days of illness. Patients who started receiving treatment before meeting the KD criteria and those who might be at a high risk of developing CALs were excluded; this may explain why CAL-related morbidity in this study was lower than that reported previously.21 However, comparability across different institutions was prioritised, making lower generalisability inevitable. Third, long-term outcomes were not assessed. While the effects of ASA therapy in KD might warrant evaluation from a medium-term to long-term perspective, our study lacked sufficient data on long-term prognoses. Although Kato et al reported a 49.3% regression rate for coronary aneurysms within 1–2 years,27 the decreased incidence of cases presenting with coronary aneurysms in the modern era means that our study’s sample size was inadequate for outcomes related to aneurysm regression or long-term coronary aneurysm persistence. Fourth, 24 patients in the non-HD group were administered low-dose ASA before initial IVIG. Compared with patients who were not administered ASA throughout the acute phase, those patients tended to be younger and had a higher Kobayashi score; this could lead to an indication bias, although adjusting for possible confounders yielded the same result.
In conclusion, this study revealed that the no-ASA modality in the acute phase of KD was non-inferior to HD ASA administration with respect to the incidence of CALs. This study suggests that the KD guidelines recommending ASA administration in the acute phase of KD should be revised.
Data availability statement
No data are available. Due to the nature of this study, the participants did not agree for their data to be shared publicly; therefore, supporting data are not available.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and the study protocol was approved by the Ethics Committee of Kobe City Medical Center General Hospital (no. zh220913). Participants gave informed consent to participate in the study before taking part.
Acknowledgments
We would like to extend our sincere gratitude to Editage for their professional assistance in editing and proofreading this manuscript. We are also deeply grateful to Prof. Hisashi Noma for his exceptional statistical guidance.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors CM had full access to all the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis. Additionally, CM acts as the guarantor of this study, accepting full responsibility for the overall content. Concept and design: KH, CM, SH, NK, CA, MY and ET. Acquisition, analysis or interpretation of data: all authors. Drafting of the manuscript: KH and CM. Critical revision of the manuscript for important intellectual content: SH, NK and ET. Statistical analysis: CM. Supervision: SH, NK and ET.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.