This is the first study on KD carried out in Cyprus using National multicenter data. The epidemiology of the disease in our country, has similar characteristics with the previous studies in other countries especially those sharing the same climatic conditions such as Spain and Greece [9, 16]. There is a slight preponderance of males (male/female ratio 1.25:1). Children younger than 5 years old are the most frequently affected (83% of cases) with median age of cases 2.23 years, similar to other European countries and different than Japan where the incidence rate was the highest among children aged 6–11 months [17, 18]. The onset of KD was more frequent in winter and spring (65%), which is consistent with findings in European reports and in neighboring countries . However, in reports from some Asian countries, the highest incidence rates were reported during summer and spring .
The majority of children with KD in Cyprus presented with changes of the lips/oral cavity (89%), which is in accordance with reports in other countries [9, 16]. Followed by the skin rash, which considerably varied in nature. More precisely, maculopapular was the most common form, while in two cases a petechial rash was present. Petechial rash is not described as a frequent finding in other studies. Whilst bilateral bulbar conjunctival injection without exudates was the first or second most frequently seen manifestation in some studies  in our study it was the third most frequent manifestation noted (80.4%). Unilateral cervical lymphadenopathy was the least common finding (57%), which was in accordance with other studies .
Diagnosis of complete and incomplete KD depends on clinical and/or laboratory criteria. The rate for incomplete cases in our population was 31%, which is consistent with the rate found in a study in the Spanish population but differs from the rate in a study in Japan where the rate of incomplete disease was only 10% [9, 20]. In general, the rate of incomplete KD is underestimated and is higher in infants than in older age groups [20, 21]. As expected, all clinical criteria were detected at a significantly lower rate in incomplete vs complete cases. However, similar to complete cases, changes of lips and oral cavity was the most frequent clinical finding in incomplete cases. This was similar to the findings in the Spanish population . Hemoglobin levels on admission were found to be significantly lower in incomplete vs complete cases and AST and ALT were found to be significantly higher. Platelets before administration of IVIG were found higher in incomplete vs complete cases.
The rate of responsiveness to the first dose of IVIG administered is high (84%) similar to other countries [5, 16, 22, 23]. Multivariable regression analysis indicated that four laboratory parameters on admission i.e. sodium ≤ 133mmol/L, albumin ≤3.2g/dl, ALT≥80 U/L and the percentage of neutrophils ≥80% were positively associated with non-responsiveness to IVIG. There is great variation in risk factors for IVIG non-responsiveness identified in different studies. A large study in the Spanish population also revealed hypoalbuminemia and hyponatremia as risk factors for non-responsiveness. In addition to these it also revealed anemia and high procalcitonin values as risk factors . In a study in San Diego USA, multivariable analysis revealed early administration of IVIG (before day 5), higher concentrations of γGT, higher % bands and lower hemoglobin, as risk factors associated with resistance to IVIG .
In Japanese children, Kobayashi, the Egami and the Sano scoring systems, have been shown to successfully predict the risk of IVIG resistance [22, 25, 26]. The KS includes some of the risk factors identified in our study such as the sodium ≤133mmol/L, neutrophils≥80% in the white blood cell count, but also uses some other risk factors such as AST≥100 U/L, low platelet count≤300x10^9/L, early administration of IVIG before day 5 of illness and high CRP. Egami scoring system gives scores for age <6 months, days of illness <4 when administering IVIG, low platelet count, increased CRP and raised ALT. More precisely, the KS has been shown to predict IVIG non-responsiveness with 78% sensitivity and 76% specificity, while the Egami score showed 78% sensitivity and 76% specificity in Japanese children. In our population both scoring systems showed low sensitivity and medium specificity with low positive predictive value (PPV). This was comparable with other non-Japanese populations [9, 27, 28].
Risk factors for the development of CAAs were evaluated in our study and only young age less than 12 months was identified as a risk factor. Young age<12 months was also identified in other studies . However, some other studies have also identified other factors as risk factors for the development of CAAs such as gender, delayed treatment, the total duration of fever and blood parameters . The Kobayashi and Egami scoring systems also showed poor sensitivity in predicting cases with CAAs in our population as was the case in other studies [9, 31].
Our study has several strengths and limitations. The retrospective nature of study means it was not possible to collect additional data relevant to Kawasaki disease and that some recorded data may have been missed. However, the records kept were very detailed, in nearly all cases, and contained most data required for the study. The study covers a 19 Year period, providing one of the longest Kawasaki datasets described. The study includes children with KD who have been diagnosed since 2000, therefore the treatment regimen maybe variable in some cases, according to the existing guidelines at the time. However, the total amount of immunoglobulin administered was the same in all cases.