Abstract
Background pediatric inflammatory multisystem syndrome (PIMS) is a complication of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in children that resembles Kawasaki syndrome and poses children at high risk of cardiorespiratory instability and/or cardiac damage. This study aims to describe the clinical presentation and outcomes of patients with PIMS in Mexico City.
Methods this was an observational study (May 1, 2020, to September 30, 2021) of children with PIMS according to Centers of Disease Control and Prevention case definition criteria, hospitalized in a single tertiary care pediatric center in Mexico City. Demographic characteristics, epidemiological data, medical history, laboratory tests, cardiology evaluations, treatment, and clinical outcomes were analyzed.
Results Seventy-five cases fulfilled case-definition criteria for PIMS (median age 10.9 years, IQR: 5.6-15.6). Fifteen (20%) had a severe underlying disease. Forty-eight cases (64%) were admitted to the intensive care unit, 33 (44%) patients required invasive mechanical ventilation, and 39 (52%) received vasopressor support. Two distinct groups of patients were identified: cluster 1 (n=60) who had rash or gastrointestinal symptoms and cluster 2 (n=15) with predominantly respiratory manifestations. Two cases (2.7%) died, both with severe underlying conditions. Five cases (6.7%) developed coronary aneurysms, all of them from cluster 1.
Conclusion clinical manifestations and outcomes are in general comparable what has been previously reported in international series. In our series, there was a high proportion of patients with severe respiratory involvement and positive RT-PCR SARS-CoV-2 and a low frequency of coronary aneurysms which suggests a possible higher proportion of children with severe acute COVID-19 in our included cases.
Introduction
Infection by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) usually has a mild clinical presentation in children. However, in rare cases, children can be severely affected and have clinical manifestations different from adults. In April 2020, some reports described a clinical syndrome similar to Kawasaki disease or toxic shock syndrome temporally associated with current or recent SARS-CoV-2 infection in the pediatric population(1). Since then, this syndrome has been recognized worldwide and named pediatric inflammatory multisystem syndrome (PIMS) or multisystem inflammatory syndrome in children (MIS-C). (2–6).
PIMS case definition varies slightly between different health agencies (7–9) and is likely to change over time. Criteria include fever, inflammatory marker elevation, multisystem organ involvement, evidence of current or recent SARS-CoV-2 infection, and exclusion of alternative diagnoses. Besides the need for hospitalization for life support, an outcome of concern is possible myocardial and coronary artery involvement similar to those observed in Kawasaki disease (10–12).
PIMS incidence has been estimated to be approximately 3-5 per 10,000 individuals younger than 21 years of age infected with SARS-CoV-2. Black and Hispanic ethnicities have been associated with a higher incidence (13).
Epidemiological description of PIMS has been difficult because of varying awareness of the condition in different clinical scenarios and because of the diversity of clinical manifestations. Besides, there is an overlap between the clinical manifestations of severe acute COVID-19 and PIMS (10). Current PIMS case definition is purposely broad and unspecific in the aim to collect the maximum information about this phenomenon. However, increasing evidence suggests the existence of different phenotypes of PIMS, the more clearly defined of them being a Kawasaki-like syndrome, a Toxic Shock-like syndrome, and a predominantly respiratory syndrome which might have different pathogenesis and clinical outcomes. (14–16).
This study aims to describe the clinical characteristics and outcomes of PIMS cases admitted to a tertiary care pediatric center in Mexico City. An attempt was made to define different subgroups of patients who otherwise fulfilled case definition criteria for PIMS.
Methods
Study design
This is a cross-sectional analysis of PIMS cases diagnosed in “Hospital Infantil de México Federico Gómez” (Federico Gómez Mexico Children’s Hospital), a tertiary care pediatric facility officially designated to attend patients with severe SARS-CoV-2 infection in the population younger than 18 without public or private health insurance in Mexico City.
Possible PIMS cases were identified in the emergency room, were transferred from other health care facilities directly to the Intensive Care Unit or were identified by treating physicians in the general hospitalization ward for COVID-19 patients. Cases were evaluated to determine if they met the CDC PIMS case definition(7). In brief, criteria included the presence of fever, the elevation of inflammatory markers, signs of involvement of at least two organic systems requiring hospitalization, and evidence of recent SARS-CoV-2 exposure or infection. Patients were excluded if they had another plausible explanation for the illness. When there was uncertainty about an alternative diagnosis, the complete case file was reviewed by an expert panel which included an infectious disease and a critical care specialist, and a consensus was reached. Detailed inclusion criteria are specified in Chart 1.
Case definition criteria for PIMS
Data collection
Clinical and laboratory data were extracted from the clinical file and included demographics, underlying medical conditions, clinical manifestations, laboratory values, management, and outcomes. For laboratory values with more than one measurement, the worst value measured within the first three days of hospitalization was registered. Vital support management (respiratory support and vasopressor utilization), intravenous immune globulin, systemic steroids, and anticoagulant therapy were documented as the main therapeutic interventions. Outcomes of interest were intensive care unit (ICU) admission, length of ICU and hospitalization stay, need for invasive mechanical ventilation, vasopressor support, myocardial depression (left ventricular ejection fraction < 55%), coronary aneurysms (coronary artery diameter z-score >= 2.5), or dilatation (z-score 2> - <2.5) on the echocardiographic evaluation performed during hospitalization, and in-hospital mortality. Markers with more than 20% missing data were not analyzed.
For this study, severe respiratory involvement was defined as infiltrates on chest X-ray or computed tomography plus the need for oxygen administration with a non-re-breathable mask or a higher oxygen concentration device. The variable “gastrointestinal symptoms” was defined as the presence of at least one of the following: abdominal pain, diarrhea or emesis.
The study center has an important population of patients with severe underlying diseases. To have a picture more alike general population, we performed a sub-analyisis with the exclusion of cases with severe underlying diseases (i.e., cancer and other forms of immunosuppression, neuromuscular disability, chronic respiratory disease except for asthma, congenital cardiopathies, and chronic kidney failure).
Statistical analysis
Descriptive analysis was undertaken in STATAv.14.0 (StataCorp, Tx). Categorical variables are reported as frequencies and continuous variables as medians and interquartile ranges (IQR). Analysis was performed in the whole sample and three subgroups. In the first subgroup, cases with severe underlying conditions were excluded. The other two groups were derived using latent class analysis (LCA). Two class LCA were conducted using the R software package “poLCA” (17) with 100 iterations to identify the clusters. The indicator variables used in LCA were the presence or absence of severe respiratory involvement, rash, and gastrointestinal symptoms. The fit of each model was assessed using a Bayesian Information Criterion score.
The ethics review board of Federico Gómez Mexico Children’s Hospital authorized the analysis and publication of de-identified data (Reg: HIM2020-031).
Results
Clinical departments sent 167 reports of cases with probable PIMS admitted to the Hospital from May 1, 2020, through September 30, 2021. For reference, during this period, 8193 real time polymerase chain reaction (RT-PCR) tests for SARS-CoV-2 tests were performed in the study center, from which 837 had a positive result.
From the 167 received reports, 42 were from the emergency department, 34 from the intensive care unit, and 91 from the infectious disease department. Duplicated reports were eliminated (n=23), 23 reports failed to meet inclusion criteria and 46 were excluded because an alternative etiology was found to explain the clinical manifestations after the comprehensive review of the case’s file and consensus among researchers. All subsequent analyses were performed in the remaining 75 cases unless otherwise specified.
The median (interquartile range, IQR) age was 10.9 (5.6-15.6) years old and 52% were female. At least one underlying condition was found in 45.3% of the population, including obesity (25%), cancer (9%), neuromuscular (8%), and respiratory (7%) diseases among others (Table 1).
In addition to fever, the most common clinical manifestations were cough (52%), conjunctivitis (50.6%), abdominal pain (51%), rash (48%), headache (45.3%), and diarrhea (44%). The median (IQR) duration of symptoms before admission was 7 (4-9) days. About 51% had lower respiratory symptoms, 85% had pulmonary infiltrates in the radiograph or computed tomography, and 65.3% required oxygen administration with a non-rebreathable mask, high flow nasal cannula, or mechanical ventilation. Vasopressor support was used in 52% of cases, and about 40% had a left ventricular ejection fraction < 55% on echocardiography. Median values for markers of inflammation, coagulation, and system damage are summarized in table 1.
Three subgroups were analyzed independently. In the first subgroup (n=60), 15 cases with a severe underlying disease were excluded. Clinical manifestations and outcomes did not differ significantly from those observed in the total sample (Table 1 and 2)
The second (n=60) and third (n=15) subgroups were derived from LCA which identified two classes of patients: cluster 1 (n=60) composed of patients with rash and/or gastrointestinal symptoms, and cluster 2 (n=15) with predominant cardiorespiratory involvement without a rash or gastrointestinal symptoms. Despite the reduced number of cases in cluster 2, some differences are evident: cluster 2 cases were more prone to have a positive RT-PCR test (76% vs 100%, p= 0.04) and less so to have a positive serology test (0 vs 54%, p<0.001). An underlying medical condition was also more frequent in cluster 2 (80 vs 36.7%, p=0.003), especially a severe one (47 vs 13%, p=0.008) such as cancer, neuromuscular disability, or a chronic respiratory condition. (Table 1). Cluster 1 patients tended to have higher levels of inflammation markers, though the difference was statistically significant only for the neutrophil count and D-Dimer concentration.
About treatment, 45.3%, 64%, and 63% of cases received intravenous immune globulin, systemic steroids, and anticoagulation therapy respectively, much more frequently in the cluster 1 group. (Table 2)
Regarding outcomes, two (2.7%) patients died. Both had severe underlying diseases; the first one had a medulloblastoma and died two weeks after the antineoplastic chemotherapy dose and the second one had chronic kidney failure and was on dialytic replacement therapy. One death occurred in each cluster group. About 44% of cases needed mechanical ventilation, 52% of cases required vasopressor support while 40% had a depressed ventricular function. (Table 2)
All cases had an echocardiographic evaluation during hospitalization. Five cases (6.7%) had coronary aneurysms. Notably, all cases with aneurysms were in Cluster 1. No patient had a coronary diameter with a z-score between 2.0 and 2.5. Myocardial involvement (left ventricular ejection fraction < 55%) was observed in 30(40%) of cases. Intensive care unit (ICU) admission occurred in 64% of cases with a median (IQR) length of stay there of 5 (3-8) days and a length of hospital stay of 7(4-13). There was a tendency of cluster 2 patients to remain longer in ICU and in hospital. (Table 2)
Discussion
The present study describes the largest cohort of patients with PIMS in the Mexican population to date, as long as we are aware. Most clinical characteristics and outcomes of our series were like those described in large international reports. The median age of our sample was some years older than most series. Prevalence of underlying medical conditions, including obesity was similar to the reported in the United States and Latin America (10,16,18–20), except for cancer which is relatively frequent in our total sample, reflecting the characteristics of the population regularly attending the study center. Cardiac dysfunction, vasopressor requirements, rash, and gastrointestinal symptoms were comparable to those reported elsewhere, so as ICU admissions, length of hospital stay, and length of ICU stay. The death rate of 2.7% (binomial 95% confidence interval: 0.3-9.3%) was similar to the previously reported (3,5,10,14,18,21).
Our population had a relatively high frequency of lower and serious respiratory involvement in comparison with most of the other series. Besides, in our sample, serology was performed in only 55% of cases, so the evidence of SARS-CoV-2 infection in our population is mostly by RT-PCR (80% positivity rate). These findings might point toward a higher proportion of acute severe SARS-CoV-2 infection vs a theoretical late-onset post-infectious syndrome in our sample compared to other case series where respiratory involvement and RT-PCR positive rate was lower (3,5,10,14,18). Current PIMS diagnostic criteria does not exclude acute severe COVID-19; this overlap between both clinical presentations has been previously discussed (10,16).
On the other hand, the frequency of coronary aneurysms in our study was relatively low (6.7%). Aneurysm frequency is heterogeneous among reports, varying from 6.7 to 23.3% (14,18,19,21). This probably reflects the lack of specificity on the PIMS case-definition criteria and the variability of the cut-point for some inflammatory markers used in different reports. Most case series are derived from active or passive surveillance, either retrospectively or prospectively. Many of them do not inform a cut-point for inflammatory markers concentration, or clinical symptoms and it is assumed that the higher limit of the normal reference range was used. This probably explains the high frequency of coronary aneurysms (23.3%) found by Flood et al., in the United Kingdom, their study having a very high cut-point for CRP (100mg/L) and requiring “acute abdomen” and very specific dermatological findings, not only “abdominal pain” or “rash” as criteria for gastrointestinal or mucocutaneous involvement respectively (14). Stringency in diagnostic criteria automatically excludes a high proportion of potential cases keeping a more homogeneous and severely affected population.
Notably, none of the five aneurysm cases of our series had a serious underlying disease and all of them belonged to cluster 1 (Mucocutaneous/gastrointestinal predominant symptoms). Although ours was a small sample to make any inference, it supports the increasing awareness about the existence of different clinical phenotypes. Godfred-Cato et al.(15), and later Geva et. al(16), identified three classes of patients: 1) a seriously ill group with significant cardiovascular involvement, multiple organ dysfunction, a pronounced elevation of inflammatory markers who tended to have positive serology for SARS-CoV-2, a second group with predominant respiratory involvement and high RT-PCR positivity rate that resembles our cluster 2 group, and the third group with predominant Kawasaki-like mucocutaneous findings. Meanwhile, Flood et al. also identified three PIMS phenotypes using cluster analysis: PIMS with Kawasaki disease-like presentation, PIMS with findings of Kawasaki disease and shock, and PIMS without any of these features (14) and consensus derived clinical recommendations have been generated accordingly (22). In all three cluster analyses, coronary artery aneurysms are more frequent in groups with Kawasaki-like features and higher levels of inflammatory markers, which in our cluster analysis corresponds to Cluster 1. It might be important to differentiate the clinical subtypes diagnosed as PIMS under the umbrella of CDC and WHO case definition criteria, since probably not all of them benefit from the same therapeutic interventions. The role of inflammation along all spectrum of severe illness associated with SARS-CoV-2 infection should be deeply studied, including its role in myocardial disfunction since it is a frequent complication both in adults with COVID-19 and children with PIMS. (23)
A limitation of this study, as in others, is the method for identification of included cases for analysis depending on spontaneous reports by treating clinicians. In the future, a prospective multicentric cohort design with clear-cut criteria for subclassification of every patient fulfilling the broad clinical criteria after SARS-CoV-2 infection would provide more information on the outcomes and possible associated factors.
Conclusions
Epidemiology of PIMS cases observed in our center is comparable to the reported elsewhere, although our series included a high proportion of patients with an acute-like respiratory phenotype. Coronary aneurysms are infrequent and usually present in previously healthy patients with rash or gastrointestinal symptoms. The death rate is alike the reported elsewhere and occurred in patients with severe underlying diseases.
We agree with other researchers that underneath the umbrella of case definition for criteria for COVID-19 at least two groups of patients can be identified according to their clinical presentation. Standardized criteria for sub-classification of PIMS cases need to be developed, since different phenotypes might have different pathophysiology, different outcomes, and possibly require different management. Clinical trials are needed to evaluate therapeutic interventions in different PIMS phenotypes.
Data Availability
All data produced in the present study are available upon reasonable request to the authors
Conflict of Interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Data Availability
All data produced in the present study are available upon reasonable request to the authors
Author Contributions
All authors contributed with study planning, design and review of the final manuscript. RGG, MLMY, RNJJ, VBOL contributed with data collection. HMG coordinated the research team. MFCP and NGG performed data analysis and the writing of the manuscript.
Funding
This study was funded by Hospital Infantil de México Federico Gómez
Data Availability Statement
The datasets generated for this study can be requested to corresponding authors.
