Since the identification of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in late 2019, coronavirus disease (COVID-19), caused by this virus, has reached pandemic proportions [1], severely impacting our health systems, health care workers and the general community [2-4]. Clinical symptoms in children with COVID-19 are generally mild and non-specific [5, 6]. However, severe cases of COVID-19 affecting previously healthy children and adolescents, akin to toxic shock syndrome (TSS) or incomplete or complete Kawasaki disease (KD), have been reported from around the world, including but not limited to the United Kingdom, Italy, the United States and Turkey [6-12].

Severe presentations in children affected by, or with exposure to the SARS-CoV-2 virus, have since been referred to as multisystem inflammatory syndrome in children (MIS-C), paediatric multisystem inflammatory syndrome temporally associated with SARS-CoV-2 (PIMS-TS), paediatric inflammatory syndrome or paediatric inflammatory shock [13]. This is in contrast to classic KD, which typically affects infants and young children and higher reported incidence in children of Asian descent. MIS-C, on the contrary, typically has far more gastrointestinal symptoms, grossly elevated inflammatory markers and marked lymphopenia and thrombocytopenia [13, 14], with a disproportionately higher incidence in Black (25–40%) and Hispanic (30–40%) children, and a small number of cases in Asian (3–28%) children [8, 15, 16]. However, clinical features of MIS-C, PIMS-TS or KD are overlapping and distinguishing between them is often very difficult, with many patients fulfilling criteria for two or all of the conditions, and vastly similar treatment modalities being employed [14, 17]. Additionally, epidemiological studies have reported that MIS-C can present with KD-like features in younger children [18]. To our knowledge, summative studies reporting on the prevalence of all three of these paediatric hyperinflammatory syndromes in hospitalized patients specifically are limited.

This objective of this study was to perform a meta-analysis to determine a pooled prevalence estimate of these paediatric hyperinflammatory conditions (PHICs), namely MIS-C, PIMS-TS or KD, in paediatric and adolescent hospitalized patients admitted for treatment due to COVID-19.

Our underlying question was: what is the estimated pooled prevalence of PHICs in paediatric and adolescent hospitalized COVID-19 patients?

Material and Methods

The study was performed as per the Standards for Reporting Diagnostic Accuracy (STARD) 2015 guidelines [19] (Table S3) and the Meta-analysis of Observational Studies in Epidemiology (MOOSE) Checklist (Table S4) [20]. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [21] diagram showing the pathway for screening and selecting studies for inclusion in the systematic review and meta-analysis is provided (Fig. 1).

PRISMA flow chart showing the process of screening and selecting studies for the meta-analysis. The PRISMA flow chart shows the main characteristics of the included studies. n = number of studies.

Literature search: identification and selection of studies

The following databases were searched: Embase, PubMed/Medline and Cochrane Library, until 2 June 2021. Additional articles published till 12 November 2021 were added during the revision stage of the peer review. Keywords used included a combination of terms including: ‘Kawasaki disease’, ‘COVID-19’, ‘severe acute respiratory syndrome coronavirus 2’ and variations thereof. Full search strategies and a complete list of keywords are provided in the Supporting Information (Search Strategy). In addition, the reference list, and related bibliography thereof, of articles selected was also examined to retrieve studies relevant to our analysis.

Inclusion and exclusion criteria

Inclusion criteria for incorporation into this study included: (1) patients were aged less than 21 years; (2) patients with COVID-19 infection or exposure, reporting on the prevalence of hospitalized patients diagnosed with MIS-C, PIMS-TS or KD; (3) studies with good methodological design (including sufficient sample size, defined as >20 patients) and (4) relevant data on associated outcomes were available. The exclusion criteria were: (1) animal/preclinical studies; (2) duplicated publications; (3) where multiple studies from overlapping centres with varying study periods reporting similar outcomes were present, studies with smaller sample size or shorter study period were excluded; (4) full-text article not available; (5) systematic reviews or meta-analyses, conference abstracts, letters and case reports or series; (6) studies presented as abstract, with relevant KD, PIMS-TS or MIS-C or outcome data not reported and (7) publications not in the English language. Case definitions for MIS-C and PIMS-TS have been summarized succinctly previously [13].

Data extraction

Title and abstracts were first reviewed on Endnote to exclude articles mismatched to eligibility criteria. The remaining articles underwent a thorough full-text examination to determine whether they were to be included in the systematic review or meta-analysis as per the eligibility criteria. Reviews, former systematic reviews and meta-analyses and opinion articles were kept separately for discussion in the manuscript. Two authors conducted the screening independently, and any disagreements were discussed until a consensus was made. Data from each study/trial were extracted independently using a standardized data extraction sheet to obtain the following information: (1) baseline demographics: author, country and year of publication; (2) study population: age of patients, sample size and comorbidities; (3) treatment details: intravenous immunoglobulins (IVIG), steroids, aspirin and interleukin blocker use; and (4) outcome measures: in-hospital mortality, intensive care unit (ICU) admission, hospital length of stay, ICU length of stay and mechanical ventilation.

Quality assessment of included studies

The methodological quality of each study was assessed using the modified Jadad scale by two researchers independently [22]. The scale evaluates study quality based on the following evaluation criteria: randomization, blinding, withdrawals, dropouts, inclusion/exclusion criteria, adverse effects and statistical analysis. A double-blind got a score of 1, whilst a single-blind got 0.5. The total score for each study ranged from 0 to 8 points, and using this, trials were divided into low-quality (0–3 points) and high-quality (4–8 points) levels.

The risk of funding bias included in studies was also evaluated independently by using the scoring test developed by Saunders et al. (2017), which analyses the declaration of funding sources and conflicts of interest [23]. A score of 1–2 was considered to indicate a moderate potential for bias. Whilst the absence of industry funding was not considered to signify an absence of bias, the presence of industry funding or conflicts of interest was assumed to indicate bias.

Statistical analysis

All statistical analyses were performed using STATA (version 13.0; StataCorp LLC, College Station, TX, USA). Baseline characteristics of patient populations were synthesized from all included studies (Table 1). Where applicable, median and interquartile ranges were converted to mean and standard deviation, and median and ranges to mean and standard deviation using the methods described by Luo et al. [24] and Wan et al. [25]. For studies where standard deviation (SD) was not available, we used the method outlined by Walter and Yao [26] to calculate the SD, assuming a normal distribution of data. Combined means were calculated where applicable.

Table 1. Overall summary of baseline demographics and clinical characteristics Characteristics Overall COVID-19 (including PHICs) PHICs1 Baseline data2 Patients 2202 (100) 780/2202 (35.42) Female 657/1520 (43.22) 286/679 (42.12) Age3 9.10 ± 8.67 (1478) 8.95 ± 6.14 (722) [Excluding Feldstein et al. [8]] [8.17 ± 6.10 (362)] [9.00 ± 5.59 (183)] Shock 42/163 (25.77) 52/87 (59.77) African or African American Ethnicity 371/1329 (27.92) 217/636 (34.12) Asian Ethnicity 24/193 (12.44) 12/87 (13.79) Caucasian Ethnicity 226/1329 (17.01) 86/636 (13.52) Comorbidities2 Obesity 354/1168 (30.31)4 198/643 (30.79) Prior CVD 85/1270 (6.69) 18/570 (3.16) Immunocompromised 6/174 (3.45) 0/10 (0) Lung diseases (including asthma) 241/1369 (17.60) 81/610 (13.26) Clinical symptoms2 Fever 196/247 (79.35) 100/100 (100) Fever ≥ 5 days 23/83 (27.71) 22/49 (44.90) Skin manifestations 439/1209 (36.31) 384/596 (64.43) Conjunctivitis 32/83 (38.55) 35/56 (62.50) Respiratory distress 63/135 (46.67) 21/30 (70) Gastrointestinal symptoms 889/1238 (71.81) 515/574 (89.72) Vomiting 45/135 (33.33) 28/37 (75.68) Diarrhoea 31/135 (22.96) 14/37 (37.84) Abdominal pain 46/135 (34.07) 30/37 (81.08) Neurological manifestations 448/1292 (34.67) 253/620 (40.81) Coronary artery abnormalities 69/1147 (6.02)4 80/614 (13.03) Laboratory characteristics3 Haemoglobin (g/dL) 17.52 ± 23.38 (1140) 11.20 ± 1.67 (550) Ferritin (ng/mL) 559.34 ± 932.93 (137) 858.09 ± 958.92 (104) C-reactive protein (mg/L) 112.13 ± 116.48 (985) 162.34 ± 119.66 (619) D-dimer (mg/dL) 1.47 ± 1.76 (103) 1.45 ± 1.85 (89)5 Neutrophils (109/L)6 7.20 ± 5.69 (1085) 8.84 ± 6.14 (571) Lymphocytes (109/L)7 2.05 ± 2.19 (1244) 1.68 ± 1.74 (588) Platelets (109/L)8 232.42 ± 136.74 (228) 203.10 ± 184.36 (53) Albumin (g/dL) 5.41 ± 7.52 (137) 7.10 ± 10.97 (62) Troponin (ng/L)5,9, 5,9 30.95 ± 54.40 (31) 109.77 ± 204.12 (52) Treatment2 IVIG 505/1219 (41.43) 521/654 (79.66) Steroids 559/1188 (47.05) 459/649 (70.72) Il-Blockers10 72/1168 (6.16)4 74/625 (11.84) Vasoactives11 329/1290 (25.50) 297/669 (44.39) Outcomes2 In-hospital mortality 40/1414 (2.83) 27/686 (3.94) Any mechanical ventilation 595/1310 (45.42) 329/669 (49.18) ICU admission12 744/1394 (53.37) 458/641 (71.45) ICU LOS3 5.42 ± 7.22 (782) 4.78 ± 5.00 (482) [Excluding Feldstein et al. [8]] [9.69 ± 13.84 (143)] [6.60 ± 8.23 (94)] Hospital LOS3 6.05 ± 5.08 (1236) 7.73 ± 4.74 (627) [Excluding Feldstein et al. [8]] [6.79 ± 6.92 (153 )] [8.04 ± 5.97 (104)] Data provided to 2 decimal places where required. This table depicts crude prevalence rates only calculated from descriptive statistics, not pooled estimates of prevalence. COVID-19, coronavirus disease 2019; CVD, cardiovascular disease; dL, decilitres; g, grams; ICU, intensive care unit; Il, interleukin; IVIG, intravenous immunoglobulins; L, litres; LOS, length of stay; mg, milligrams; ng, nanograms. Italic values are just the crude values, no significance.

A pooled estimate of the prevalence of PHICs (comprising MIS-C, KD or PIMS-TS) among hospitalized paediatric and adolescent patients treated due to COVID-19 was obtained. The ‘metaprop’ STATA command was used to pool proportions by performing a random effects meta-analysis of proportions obtained from individual case series. Random effects modelling was performed using DerSimonian and Laird method. To stabilize variances, the Freeman–Tukey double arcsine transformation was also applied. The overall effects were presented using forest plots. A fixed effects model was used for heterogeneity <50% and a random effects model for heterogeneity >50%. The heterogeneity was estimated from the inverse-variance fixed effects model. I2 statistics and p-values were used to assess heterogeneity between studies, with <40%, 30–60%, 50–90% and 75–100% representing low, moderate, substantial and considerable heterogeneity, respectively [27]. p-Values <0.05 were considered statistically significant.

Results Description of included studies

Overall, 14 studies comprising of 2202 paediatric and adolescent hospitalized patients admitted for treatment due to COVID-19 were included in the meta-analysis, out of which 780 (35.42%) were diagnosed with PHIC [MIS-C, including one study where MIS-C also included ‘Kawasaki-like disease’ (n = 722), KD (n = 2) and PIMS-TS (n = 56)]. Patient demographics for overall and PHICs groups were 657/1520 (43.22%) and 286/679 (42.12%) female patients, respectively, and mean age 9.10 ± 8.67 (1478 patients) and 8.95 ± 6.14 (722 patients), respectively. However, on the exclusion of the largest study, Feldstein et al. [8], mean age was 8.17 ± 6.10 (362 patients) and 9.00 ± 5.59 (183 patients), respectively. The overall summary characteristics of all studies included in the meta-analysis are shown in Table 1 and details pertaining to individual studies in Table 2. The results of methodological quality and funding bias assessment of included studies are provided in Table S2.

Table 2. Clinical characteristics and outcomes of included studies Study ID Author (country) Year R/P Diagnosis Severity of COVID patients included Group Treatment Outcomes IVIG Steroids Il-Blockers VAs IHM Any MV ICU Ad. ICU LOS Hospital LOS 1 Bari et al. [43] (Pakistan) 2021 P MIS-C/Kawasaki-like illness All Overall 5 PHICs 1 2 Gupta et al. [44] (India) 2021 R MIS-C All Overall 13 14 PHICs 16 13 12 13 3 Javalkar et al. [45] (USA) 2021 R MIS-C All Overall PHICs 34 30 9 8 9 5.17 ± 5.06 6.42 ± 5.17 4 Ozsurekci et al. [46] (Turkey) 2021 R MIS-C Severe/critical Overall 37 39 27 19 3 15 32 6.20 ± 5.31 9.73 ± 6.76 PHICs 30 28 26 13 0 9 14 5.43 ± 4.41 9.14 ± 6.37 5 Rektman et al. [38] (USA) 2021 R MIS-C All Overall 19 6 15 6.87 ± 7.57 PHICs 19 5 12 7.18 ± 3.84 6 Rostad et al. [47] (USA) 2020 R MIS-C Symptomatic Overall 10 5 15 6.52 ± 9.13 PHICs 10 5 10 7.03 ± 3.01 7 Alharbi et al. [48] (Saudi Arabia) 2021 R

MIS-C (n = 5)KD (n = 1)

All Overall 12 13.14 ± 17.92 PHICs 6 5 2 4 6 20.83 ± 25.10 8 Consioglo et al. [49] (Rome, Sweden) 2020 R MIS-C Mild Overall PHICs 9 Kim et al. [50] (USA) 2020 P MIS-C All Overall PHICs 10 Ray et al. [