Studies that addressed the relationship between COVID-19 and thromboembolic events in the pediatric population (< 21 years) with data on D-dimer value were included.

The designs eligible for this review were randomized controlled trials, cohorts, case-control studies, case series, and observational studies. All articles in Portuguese, English, and Spanish were included. In addition, documents accessible in full text via the Coordination for the Improvement of Higher Education Personnel (CAPES), Google, or Google Scholar were also considered.

Articles with focus on patients older than 21 years, pregnant women, and populations with comorbidities, such as coagulopathies or pulmonary diseases were excluded.

Systematic reviews, books, non-academic researches, reviews, abstracts, commentaries, policy statements, case reports, and contents in languages other than those listed in the inclusion criteria were also excluded, besides articles published before 2018.

The database search was conducted on December 14, 2021 by two authors (JC and PC). Studies were chosen from five databases: PubMed (all fields; National Center for Biotechnology, Information, National Institutes of Health; Bethesda, Maryland, USA), Scopus (article title, abstract, keywords; Elsevier; Amsterdam, Netherlands), MEDLINE / Bireme (Virtual Health Library [VLH] title, abstract, subject; US National Library of Medicine, National Institutes of Health; Bethesda, Maryland, USA), Web of Science (article title, abstract, keywords; Thomson Reuters; New York, New York, USA), and Embase (article title, abstract, keywords; Elsevier; Amsterdam, Netherlands).

A total of 79 articles were found on the five chosen platforms using the search key detailed in Figure 1. The first step, in the process of selecting the study, was to organize the articles in a PRISMA flow diagram, shown in Figure 2. Secondly, two authors (JC and PC) initially checked for duplicate documents using the “find duplicates” tool of EndNote X9 software (Clarivate Analytics; Philadelphia, Pennsylvania USA; version for Windows). In this step, 20 texts were excluded. Thereafter, the duplicates that were not excluded by the platform tool were manually excluded, ruling out another 18 articles, and leaving 41 after these two steps.

In the third step, the authors read the titles and excluded those documents that did not fit the inclusion criteria cited above, totaling the exclusion of 16 more articles, being one with a focus on imaging tests, seven focused on other diseases, two were systematic reviews, one was not related to COVID-19, one had a focus on vaccination against COVID-19, and the others were case reports, reducing the number of documents to 25.

In the fourth step, the abstracts of the remaining articles were read and seven more, which did not meet the inclusion criteria, were eliminated, leaving 18 articles. Five were excluded because they focused on patients over 21 years of age, one was not on COVID-19, and one was a comparison between COVID-19 and Influenza patients regarding the need for extended extracorporeal membrane oxygenation (ECMO).

In the final procedure, these 18 texts were read in full and 12 more were eliminated: four because the focus was on the establishment of protocols, five for not focusing on the pediatric population, one was duplicated, and the last one, because it was not possible to find the complete content.

At the end of this process, we were left with six articles considered relevant to our research. Thereafter, a “citation tracking” was done in the bibliographical references of the selected papers, where one more pertinent article was found, totaling seven to be studied in the review.

The relevant data were added to the knowledge matrix, which was used to organize and systematize the relevant information, in a Microsoft Excel spreadsheet (Microsoft Corporation; Redmond, Washington, USA), version 7.0.25, for Windows 10.

Articles were independently assessed by all authors (JC, PC, FC, and RM) for risk of bias. Any disagreement was resolved through discussion among the group. The Joanna Briggs Institute (JBI) checklists for cohort studies, case series, and case-control studies were used.

These checklists have four possible answers: yes, no, unclear, and not applicable. When the answer was not applicable, the criterion was not considered in the evaluation, as recommended by the JBI guidelines. The bias assessment was calculated based on the amount of “yes” answers in the assessment instrument. Thus, it was possible to divide the articles into three groups: high (up to 49%), moderate (50 to 70%), and low risk (above 70%) of bias.

Data considered relevant by the authors were extracted from the articles and added to the knowledge matrix elaborated using a spreadsheet from Microsoft Excel (Microsoft Corporation; Redmond, Washington, USA), version 7.0.25, for Windows 10.

The following information was extracted from each article: title, author(s), institution and country of the first author, year and journal of publication, the objective of the work, method, difficulties and limitations, age of the children in the study, whether or not there was a change in D-dimer and if this influenced the prognosis, which were the pre-thrombotic events, and if the patients had symptoms before the development of PTE or DVT.

A narrative synthesis was chosen as an appropriate method to discuss the results of this review since multiple types of studies were evaluated. The data were assessed through the knowledge matrix (Table 1) which was used to discuss the key information drawn from the documents. A discussion on the impact of the methodological quality of the included articles, in addition to the main topics of the studies, was input into the matrix.

These data were used to interpret and describe the main information about the relationship between the D-dimer level and thrombotic events in COVID-19 pediatric population.

A total of seven articles were included since they met the inclusion criteria. All were evaluated with the JBI methodological assessment tools for cohorts, case series, and case-control studies, of which six demonstrated a low risk of bias and one demonstrated a moderate risk. Therefore, none of them were excluded for methodological failure.

Six topics were categorized for the presentation of this study results: 1) Characteristics of included studies, number of patients, and age of the children; 2) Occurrence of pre-thrombotic events; 3) Symptoms of PTE or DVT; 4) The use or not of prophylactic anticoagulation; 5) Whether the change in D-dimer worsened, improved, or had no influence on prognosis; and 6) Alleged difficulties and limitations.

Three of the seven studies were conducted in the United States, one in Italy, one in Pakistan, one in Saudi Arabia, and one in China. Two of them were case series, with a mean number of eight patients, four were cohorts with a mean number of 43 patients, and one was a case-control study with 43 patients. The age ranged from less than one year to 21 years.

The number of patients in the studies was below 100 in all the papers evaluated, and below 10 for Al-Ghafry et al.²⁰ and Sadiq et al.,²¹ as both were case series studies.

The maximum age in the studies was variable with Al-Ghafry et al.,²⁰ Al-Ghafry et al.,²² Del Borrello et al.,²³ and Mitchell et al.²⁴ considering the population under 21 years old. Sadiq et al.²¹ considered the age under 15 years for the population of their study, Saleh et al.,²⁵ under 14 years, and Wang et al.,²⁶ under 18 years.

In most articles there were no reports of thromboembolic events, except for that of Mitchell et al.²⁴ Del Borrello et al.²³ stated that D-dimer is still controversial in predicting VTE because it tends to increase in all inflammatory conditions that are associated with an increase in thrombotic risk, with no clear association between the intensity of the change in D-dimer and the magnitude of thrombotic risk. However, in the cohort study by Mitchell et al.,²⁴ seven patients developed thrombotic events, two DVT and four PTE, one developed both DVT and PTE, and one of the later evolved to death.

Prophylactic anticoagulation was widely used in the selected studies, which may have generated some degree of confusion about the correlation between increased D-dimer and thrombotic events.

Only the study by Wang et al.²⁶ did not use therapy in any of the patients. In the cases where it was used, prophylactic anticoagulation was applied according to the protocol for adults, due to the lack of a specific protocol for children.

For Al-Ghafry et al.,²⁰ all patients in the case series admitted to the intensive care unit (ICU) had a three- to tenfold increase in D-dimer levels, but had no increase in mortality or thrombotic events. In the cohort of Al-Ghafry et al.,²² there was a D-dimer value higher than 2,144 ng/mL, considered sensitive (82%) and specific (75%) to predict pediatric ICU (PICU) admission. This value was also used as a cutoff to correlate with MIS-C severity, based on higher rates of PICU admission and the need for supplemental oxygen. In the study by Wang et al.,²⁶ D-dimer was increased in 37.5% of patients with severe COVID-19.

For Del Borrello et al.,²³ the D-dimer values were not statistically different between COVID-19 patients and categories of disease severity. In the cohort, D-dimer levels increased slightly at the peak of clinical manifestations, returning to normal after disease resolution.

In the study by Saleh et al.,²⁵ D-dimer values were higher (above 0.5 μg/ml) in 55.8% of patients on admission. An association between high values and PICU admission has been noticed, as in the paper by Del Borrello.²³ Sadiq et al.²¹ and Mitchell et al.²⁴ did not have D-dimer as the focus of their work but when measured in pediatric patients with COVID-19 it was also elevated.

All authors in their case series claimed the low number of patients as one of the main difficulties in conducting the study, relating this to a possible lack of statistical relevance, except Al-Ghafry et al.²⁰

Furthermore, Dal Borrello et al. stated that D-dimer assays in pediatrics are prone to pre-analytical confounding factors, such as blood drawing quality, which may limit their reliability. Al-Ghafry et al.²² also claimed in their cutoff study that therapies administered for MIS-C simultaneously to the study may have confounded some results, and administration of aspirin or anticoagulation may have hindered the analysis of absolute pro-thrombotic status. Finally, platelet mapping was not possible due to the absence of some antithrombin III data and fibrinolysis parameters, rendering the study not as complete as the researchers would have hoped.