Statistics from Altmetric.com
Using epidemiological investigation to characterise a disease in space and time can be a powerful tool to generate hypotheses about aetiology. Reports from Japan1 and the UK2 underscore the distinct seasonality of Kawasaki disease (KD) in two geographically distant regions of the northern hemisphere. The reports also document an increasing incidence of KD over time. While the seasonality of KD both in Japan and globally has been previously well documented, further analyses of patient characteristics and outcomes by season are yielding a more granular picture of the interaction between the environment and clinical features.3–5
In the report by Shimizu and colleagues1, the authors analysed 744 patients admitted to six hospitals in a single metropolitan region in Japan. Although details of the completeness of the reporting are lacking, this type of regional analysis may be more complete than the nationwide KD surveys in Japan that have an ~75% response rate. In addition, the authors had access to more detailed clinical information than is provided in the national surveys. Using these data, the authors found a seasonal effect on response to intravenous immunoglobulin (IVIG) treatment with more patients manifesting resistance in the warm season from May through October. There was no difference in coronary artery outcome by season. The authors used the definition of recurrent fever after 24 hours to define IVIG resistance and the Japanese Ministry of Health and Welfare (JMHW) criteria to assess coronary artery outcomes. These definitions differ from definitions used in the UK and North America and could potentially affect the conclusions of the analysis. For example, it has been well documented that in children treated for KD in the USA, using 24 hours instead of 36 hours after the end of the IVIG infusion would result in unnecessary treatment of 11% of patients who became and remained afebrile during the 12 hours period from 24 to 36 hours post-IVIG infusion.6 The more conservative definition of 36 hours post-IVIG infusion has been adopted by the American Heart Association guidelines.7 Using JMHW criteria for coronary artery outcomes also underestimates the prevalence of aneurysms as defined by coronary artery Z scores ≥2.5 SD units.8 These limitations notwithstanding, the finding of a different response to IVIG in the ‘warm period’ may point to different triggers for KD operating at different times of year. Other studies showing variation in seasonality of KD incidence by age and sex also point to different exposures potentially occurring at different times of year.9 10
In the report of 553 KD cases from the UK by Tulloh and colleagues2, seasonality was also observed with a peak in January. Future analyses of their data to examine the effect of seasonality on coronary artery outcome and response to treatment could be instructive and should be stratified by age and sex. Geospatial analysis revealed increased cases from rural areas. Compared with a previous period of surveillance in the UK, incidence is rising and coronary artery outcomes have improved. The current analysis of coronary artery outcomes defined as Z score ≥2.5 revealed aneurysms in 19% of the study population. Although the authors cite this as an increased rate compared with reports from other regions, their use of the conservative Z score definition likely accounts for reported differences. In fact, in a recent report from a single centre in the USA, 180 of 788 (22.8%) patients with KD had a Z score ≥2.5%, and 90.2% of patients with aneurysms received treatment within the first 10 days after fever onset.11 It is perhaps worth emphasising here the arbitrary nature of the 10-day window for IVIG treatment. This window was chosen for enrolment of patients in the first randomised IVIG trial reported in 1986.12 It subsequently became dogma that patients with KD should be treated with IVIG within the first 10 days after fever onset. As Tulloh and colleagues appropriately highlight in their analysis, treatment at the earliest possible point in time is associated with better outcomes. The authors highlight infants as a particularly high-risk population, as has been noted previously, and a group that might benefit from intensification of initial therapy with other anti-inflammatory agents (eg, steroids or infliximab) based on their increased risk. Two deaths diagnosed at autopsy underscore the importance of a high index of suspicion for making a KD diagnosis in young infants. Expressing the deaths using a standardised mortality ratio would have allowed comparison of the death rate to other regions.13
In summary, the epidemiological investigations reported here suggest that different triggers for the syndrome may be operating at different times of year and in different age groups. This might explain why the last four decades of search for the trigger for KD have come up empty handed. Future epidemiological investigation should focus on temporal and geospatial clustering of cases to gather further evidence for the possibility of different triggers for KD in different populations and in different locations.
Funding None declared.
Competing interests None declared.
Provenance and peer review Commissioned; internally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.