SARS‐COV‐2–associated coagulopathy and thromboembolism prophylaxis in children: A single‐center observational study

Abstract

Background

Multiple investigators have described an increased incidence of thromboembolic events in SARS‐CoV‐2–infected individuals. Data concerning hemostatic complications in children hospitalized for COVID‐19/multisystem inflammatory syndrome in children (MIS‐C) are scant.

Objectives

To share our experience in managing SARS‐CoV‐2–associated pro‐coagulant state in hospitalized children.

Methods

D‐dimer values were recorded at diagnosis in children hospitalized for SARS‐CoV‐2–related manifestations. In moderately to critically ill patients and MIS‐C cases, coagulation and inflammatory markers were checked at multiple time points and median results were compared. Pro‐thrombotic risk factors were appraised for each child and thromboprophylaxis was started in selected cases.

Results

Thirty‐five patients were prospectively enrolled. D‐dimer values did not discriminate COVID‐19 of differing severity, whereas were markedly different between the COVID‐19 and the MIS‐C cohorts. In both cohorts, D‐dimer and C‐reactive protein levels increased upon clinical worsening but were not accompanied by decreased fibrinogen or platelet values, with all parameters returning to normal upon disease resolution. Six patients had multiple thrombotic risk factors and were started on pharmacological thromboprophylaxis. No deaths or thrombotic or bleeding complications occurred.

Conclusions

COVID‐19 pediatric patients show mildly altered coagulation and inflammatory parameters; on the other hand, MIS‐C cases showed laboratory signs of an inflammatory driven pro‐coagulant status. Universal anticoagulant prophylaxis in hospitalized children with SARS‐CoV‐2–related manifestations is not warranted, but may be offered to patients with other pro‐thrombotic risk factors in the context of a multi‐modal therapeutic approach.

Essentials

• •Hemostatic complications in children hospitalized for COVID‐19/MIS‐C are not well characterised.
• •D‐dimer values may not parallel disease severity in pediatric COVID‐19 but increase in MIS‐C.
• •Repeated laboratory assessments revealed no signs of consumptive coagulopathy in either condition.
• •COVID‐19/MIS‐C pediatric patients might benefit from a tailored anticoagulant prophylaxis regimen.

Alt-text: Unlabelled Box

1. INTRODUCTION

With approximately 50 million people infected, over 1 million deaths (as of mid‐October 2020),¹ and innumerable economic losses, SARS‐CoV‐2–related disease (COVID‐19) is among the most severe pandemics to seize the world since the 1918 “Spanish flu.” As this infection spread across the globe, the medical literature has been flooded by clinical observations and tentative pharmacologic trials, all striving to better inform bedside practice. Nonetheless, as this pandemic reaches its twelfth month, there are still many uncertainties regarding the best therapeutic approach to this elusive disease.

A pattern of increased thrombotic risk among adult patients (especially those most greatly affected) has so far emerged,² resulting from an inflammatory‐driven endothelial dysfunction and hypercoagulable state,3., 4. and translating into a high rate of multi‐organ macro‐ and micro‐vascular injury, and death.⁵ Thus, international societies and multiple national health‐care institutions have offered guidance on how to evaluate for coagulopathy and implement an anticoagulation prophylaxis protocol in patients admitted with a diagnosis of COVID‐19.6., 7., 8.

The available evidence concerning COVID‐19 pediatric patients draws a reassuring picture in terms of morbidity and mortality, with a low incidence of thrombotic complications (widely unreported in even the most severely affected patients9., 10.). Nonetheless, the International Society on Thrombosis and Haemostasis (ISTH)‐endorsed Consensus‐Based Clinical Recommendations for Anticoagulant Thromboprophylaxis in Children Hospitalized for COVID‐19‐‐Related Illness suggest a low threshold of clinical suspicion (ie, a single additional pro‐thrombotic risk factor) and an explicit reliance on D‐dimer values to guide the choice of pharmacologic prophylaxis.¹¹

To further complicate the issue, children may develop a SARS‐CoV‐2–related inflammatory syndrome (ie, MIS‐C, multisystem inflammatory syndrome in children) that usually arises weeks after an infection. This condition may present with a wide range of cardiovascular complications, ranging from arrhythmias to coronary artery aneurysm, from myocarditis to sudden cariogenic shock; tends to evolve rapidly and may come to the attention of the practicing pediatrician at its early stages; characterized only by persistent fever, various cutaneous manifestations, and mild‐to‐moderate gastrointestinal disorders (ie, mimicking COVID‐19‐‐related symptoms in children).¹²

As the hemostatic impact of these two related conditions is still not clearly defined in children, we hereby share our preliminary experience in evaluating and managing the SARS‐CoV‐2‐‐related pro‐coagulant state and thromboembolism risk in hospital‐admitted children.

2. MATERIALS AND METHODS

2.1. Patient population

We performed a single‐center observational cohort study at a tertiary care children hospital (Regina Margherita Children Hospital) in Turin, Italy. We prospectively enrolled all pediatric patients (birth to 21 years old) with SARS‐CoV‐2‐‐related acute clinical manifestations requiring hospitalization. Current or previous SARS‐CoV‐2 infection was confirmed by either real‐time polymerization chain reaction (rt‐PCR) performed on nasal and pharyngeal swabs (Simplexa^TM COVID‐19 Direct Reaction Mix), and/or by anti‐S specific antibodies (In3diagnostic Eradikit COVID19), as previously described.¹³

Patients were labeled in terms of severity according to the following pragmatic parameters (largely in agreement with internationally recognized indications¹²): mild disease was defined as acute upper respiratory or gastrointestinal symptoms, without systemic involvement or abnormal findings on chest radiographs or lung ultrasounds; moderate disease was defined as imaging‐confirmed pneumonia with mild respiratory distress and no oxygen requirement, or as gastroenteritis with dehydration requiring intravenous fluids; severe disease was defined as oxygen‐dependent imaging‐confirmed pneumonia, severe gastroenteritis, or overt sepsis; critical illness was defined as single‐ or multi‐organ failure requiring pediatric intensive care unit (PICU) admission and monitoring and/or mechanical ventilation. MIS‐C cases were defined according to Centers for Disease Control (CDC) criteria.¹⁴ Informed consent was obtained from the parents of all minors involved in this study, as well as from the patients themselves, if older than 14 years of age. The study protocol was approved by the local ethics committee.

2.2. Laboratory analysis

A D‐dimer assay was performed at diagnosis in all hospitalized patients. In order to uncover the evolution of coagulation and inflammatory parameters, a complete blood count, prothrombin time (PT), fibrinogen, D‐dimer, and C‐reactive protein (CRP) quantification were recorded at three time points (ie, hospital admission, day of worst clinical manifestations, symptoms’ resolution) only in a subset of hospitalized patients (namely, the moderately to critically ill COVID‐19 patients and the MIS‐C cases). Serial measurements were not performed in patients with persistently mild clinical disease, as we did not want to burden these children with repeated blood draws that would have likely not impacted their clinical management. We defined the day of worst clinical manifestation retrospectively, taking into consideration all the usual clinical parameters and vital signs (eg, temperature, respiratory rate, work of breathing, number of bowel movements, and so on), and we used the laboratory values collected within 24 hours of that day. In mechanical ventilation or extracorporeal circulation were instituted, we made sure to include only laboratory values collected before such measures were started, given their impact on coagulation and inflammatory markers.

D‐dimer values were tested with an automated, latex‐enhanced turbidimetric immunoassay (HemosIL® D‐Dimer HS 500, Instrumentation Laboratory [IL]), upper limit of normal 500 ng/mL expressed as fibrinogen equivalent units), fibrinogen values were determined with an automated assay based on the Clauss method (HemosIL® QFA Thrombin [Bovine], IL) and PT was quantified by means of a recombinant tissue factor‐‐based reagent (HemosIL® RecombiPlasTin 2G, IL‐‐‐local average value is 11 s). All coagulation testing was performed on ACL TOP systems and underwent a two‐stage quality control.

2.3. Thrombotic risk appraisal and anticoagulation management

Our institutional risk assessment model (RAM) was applied as previously described¹³ (Table 1 ). For the purpose of the RAM, MIS‐C patients were attributed a starting score of + 2 points and COVID‐19 patients were attributed a starting score of + 1 point in case of at least “moderate” disease severity. Our approach was compared with the ISTH‐endorsed recommendations. Asymptomatic venous thromboembolic events (VTE) were not screened for by means of ultrasonography or computed tomography imaging.

Table 1.

Institutional risk assessment model

Item	Scores
Age/pubertal stage	IF over 10 years old or Tanner stage above 2, + 1 pt; otherwise −1 pt
Mobilitya	IF BQM 1, +1 pt; IF BQM 2‐3, 0 pt; IF BQM 4, −1 pt
CVC	IF Port or tCICC, +1 pt; IF PICC, +1,5 pts; IF ntCICC or inserted within the previous 2 weeks, +2 pts; IF TPN or hyperosmolar drugs infusion, add + 0,5 pt; IF catheter infectionb, add + 1 pt; IF > 2 catheter blockages requiring fibrinolytic infusion, add + 1 pt
Infections	IF bacterial sepsis or severe systemic viral infection or severe localized bacterial infection (eg osteomyelitis) +1 pt
Thrombophilia	IF personal history of VTE or known major thrombophiliad, +3 pts; IF known minor thrombophilia, + 1pt; IF positive family historye + 1 pt
Trauma/surgery/PICU	IF PICU admission lasting over 48 hours within the previous 2 weeks or major surgical operation within the previous 2 weeks: +1; IF major orthopaedic operation (upper extremity excluded) within the previous 2 weeks, + 2 pts; if major trauma within the previous 2 weeks,+2 pts
Cardiovascular comorbidities	IF previous cardiac disease requiring anti‐platelet agents, +3 pts; IF chronic pulmonary hypertension, +2 pts; if acute decrease of EF to 35%‐45%, + 2 pts; if acute decrease of EF to 25%‐‐35% or inotropes requirement, + 4ptsc; if acute decrease of EF to less than 25% or persistent arrhythmia, +6c
Other comorbidities	Obesity (BMI over 95e centile), +1 pt Sickle cell disease, +1 pt; IF vaso‐occlusive crisis or acute chest syndrome, +2 pts Status‐post splenectomy for underlying hematological disease, +1 pt Underlying inflammatory disease (eg, SLE, IBD), +1 pt; IF acute flare of underlying disease, +2 pts Nephrotic syndromef, +2 pts Active malignancy: leukemia, +2 pts; localized solid tumor, +1 pt; metastatic solid tumor, +2 pts; vascular compression/invasion by solid tumor, +3 pts; CNS tumor, 0 pt; vascular tumor or malformation with LIC, +3 pts Respiratory exacerbation due to cystic fibrosis, +1 pt Pro‐thrombotic drugsg, +1 pt Mechanical ventilation, + 1 pt Cigarette smoking, +1 pt
Bleeding risks	Known bleeding disorder or personal/family history suggestive of a bleeding disorder Severe uncontrolled hypertension Moderate‐to‐severe head trauma or neurosurgical operation in the previous 72h Major surgical operation in the previous 24h Active gastrointestinal lesions
If sum of scores 3 pts or higher in the absence of bleeding risks, consider prophylactic anticoagulation and haematology consult

Abbreviations: BMI, body mass index; BQM, Braden Q mobility score; CNS, central nervous system; CVC, central venous catheter; EF, ejection fraction; IBD, inflammatory bowel disease; LIC, localized intravascular coagulation; ntCICC, non‐tunnelled centrally inserted central catheter; PICC, peripherally inserted central catheter; PICU, pediatric intensive care unit; tCICC, tunnelled centrally inserted central catheter; TPN, total parenteral nutrition; SLE, systemic lupus erythematosus.

^aCompared to baseline mobility status.

^bBased on positive cultures or characteristics of the skin surrounding the catheter exit site.

^cConsider therapeutic anticoagulation.

^dProtein S, protein C, or antithrombin deficiency; homozygosity or compound heterozygosity for factor V Leiden and/or factor II G20210A; persistent triple‐positive antiphospholipid antibodies.

^eOne first‐ and/or multiple second‐degree relative with unprovoked/minimally provoked deep venous thrombosis, pulmonary embolism, myocardial infarct, or stroke < 50 years or recurrent unexplained late‐occurring abortions (> 2).

^fCurrent spot urinary protein to creatinin ratio above 2.

^gEstrogen‐containing oral contraceptive pill, systemic steroids (> 1‐‐2 mg/kg/die of prednisone equivalent) for over 2 weeks, L‐asparaginase in the previous 3 weeks.

2.4. Statistical analysis

Statistical analyses were performed with Microsoft Office Excel 2019 and R version 4.0. Data are shown as median for continuous variables and as percentage for categorical variables. The Shapiro‐Wilk test was performed to assess if continuous variables were normally distributed. The Kruskal‐Wallis test was performed for comparison of the median values of continuous variables. Proportions of categorical variables between groups were compared with Fisher's exact test. A two‐sided P‐value less than 0.05 was judged statistically significant. Missing values were imputed by mean substitution.

3. RESULTS

Between 01 March and 15 October 2020, 36 children with SARS‐CoV‐2‐‐related illnesses were admitted to the Infectious Disease Unit of Regina Margherita Children Hospital, including 30 COVID‐19 cases (median age 3 years, range 10 days to 19 years) and 6 MIS‐C cases (median age 6.8 years, range 4.5 to 12.5 years; Table 2 ). Among the COVID‐19 cases, 11 (40%) patients were female, 14 (47%) patients followed a mild course, 10 (33%) presented with a moderate disease, 3 (10%) showed a severe illness, and 3 (10%) required admission to the PICU. In all, we identified 6 MIS‐C cases (17%). Among the MIS‐C cases, 3 (50%) were girls and only 1 (16.5%) required admission to the PICU. Overall, mild COVID‐19 cases were younger (median age 0.8 versus 3.9, P = .116) and showed a lower rate of comorbidities (14% versus 56%, P = .064) compared to the rest of the COVID‐19 cohort. Comorbidities included hemato‐oncologic diseases (six patients), congenital heart diseases and obesity (two patients each), chronic respiratory diseases, diabetes mellitus, and rheumatological diseases (one patient each). Two patients had multiple comorbidities. A single patient was excluded from further analysis, as her clinical and laboratory picture was likely confounded by the co‐occurence of pneumococcal sepsis and multi‐organ failure in the context of previously undiagnosed sickle cell disease.

Table 2.

Clinical and demographic characteristics of pediatric hospitalized patients with SARS‐CoV‐2–related manifestations

Clinical category	Age	Gender (F/M)	Comorbidities (Y/N)	D‐dimera (ng/mL)	CRPb (mg/L)
Mild COVID‐19 (14)	9 m (10 d‐‐17 y)	4/10	2/16	800 (200‐‐1800b)	4 (0‐‐20)
Moderate COVID‐19 (10)	3.5 y (2 m‐‐5.5 y)	2/8	3/7	900 (200‐‐1700)	5 (0‐‐76)
Severe‐critical COVID‐19 (6)	7.5 y (9 m‐‐19 y)	3/3	6/0	800 (100‐‐2750c)	25 (3‐‐42)
MIS‐C (6)	6.8 y (4.5‐‐12.5 y)	3/3	0/6	1900 (1300‐‐4400)	215 (120‐‐300)

Abbreviations: CRP, C‐reactive protein; MIS‐C, multi‐inflammatory syndrome in children.

^aValues recorded upon hospital admission, data displayed as median and overall range.

^bThe D‐dimer value of 1800 ng/mL was recorded in a mildly affected newborn.

^cThe D‐dimer value of 2750 ng/mL was recorded in a critically affected sickle cell disease patient.

Among COVID‐19 patients, D‐dimer values were not statistically different between disease categories (mildly affected, median value 814 ng/mL versus moderately affected, median value 916 ng/mL versus severely/critically affected 823 ng/mL, P = .46). In fact, in at least five cases, all moderately to critically ill, D‐dimer values above 500 ng/mL might have been due to a baseline condition, given the facts that D‐dimer was persistently above the upper level of normal even after complete disease resolution and that those five patients had comorbidities known to be associated with increased baseline D‐dimer values (eg, acute lymphoblastic leukemia, sickle cell disease, cystic fibrosis). We suspect also that two mildly affected newborns might have had baseline increased values, although we did not collect a sample upon symptoms resolution. Nonetheless, repeating calculations upon removal of these patients did not change the overall results (data not shown).

Furthermore, coagulation and inflammatory markers did change significantly over time in moderately‐to‐critically ill COVID‐19 patients (Figure 1 ): D‐dimer levels increased slightly at the peak of clinical manifestations, returning to normal upon disease resolution (median values of 916 versus 1200 versus 416 ng/mL, respectively, P = .159 and < 0.001). Concurrently, fibrinogen values stayed at the upper level of normal during the course of the disease, returning to normal upon disease resolution (median values 373 versus 348 versus 233, respectively; P = .101 and P= .003), consistently with a mild inflammatory acute phase response (CRP levels of 9 versus 26 vs 1 mg/L, respectively; P = .330 and P = .0048). Platelets remained globally unchanged over the whole course of the disease, with a trend to higher counts upon recovery (223 versus 225 versus 292 x 10⁹/L, P = .938 and P = .153); PT values were within normal values at all time points. Eight patients, seven in the moderate disease category and one in the severe disease category, presented the peak of clinical manifestations upon diagnosis; in these cases, only two values were recorded for each blood parameter (included in the intermediate and last time points, respectively), and the third one was imputed as missing value.

Figure 1.

Trends of laboratory assessments over the hospital stay. Coagulation and inflammatory response parameters (A, D‐dimer levels; B, C‐reactive protein; C, fibrinogen; D, platelet counts) as they progressed in COVID‐19 patients (moderately to critically ill) and in multi‐inflammatory syndrome in children patients at diagnosis (= ADMISSION), at peak of clinical symptoms (= PEAK), and upon disease resolution (= DICHARGE)

Comparing D‐dimer values upon admission, MIS‐C cases showed significantly higher values than COVID‐19 cases as a whole (1906 versus 817 ng/mL, P < 0,001). In fact, a D‐dimer value above 1000 ng/mL (two times the upper limit of normal) showed a good diagnostic accuracy to distinguish between COVID‐19 and MIS‐C cases (sensitivity 100%, 95% confidence interval [CI] 54.07% to 100.00%; specificity 83.33%, 95%CI 62.62% to 95.26%). On the other hand, CRP values were actually more accurate (median values 215.5 versus 5 mg/L, with values > 100 mg/L being 100% sensitive and specific). With respect to the changing of coagulation and inflammatory parameters over time in MIS‐C patients (Figure 1), D‐dimer (1906 versus 3980 ng/mL, P = .05466), fibrinogen (580 versus 650 mg/dL, P = .10931), and CRP (215 versus 289 mg/L, P = .0.20018) increased paralleling a worsening of the patient's clinical condition, while platelets slightly decreased (144 versus 110x10⁹/L, P = .10931)‐‐‐although this did not reach statistical significance due to the low number of cases. All values normalized upon disease resolution, with a trend toward higher than normal platelet values (D‐dimer 296 ng/mL, fibrinogen 221 mg/dL, CRP 1 mg/L, and PLT 370x10⁹/L, P < .001). Again, PT values were basically within the range of normal at all time points (PT ratio 1.1 versus 1.2 versus 1,1, P = .458).

Prophylactic anticoagulation (PA) was started in six patients (14%; Table 3 ). These patients received enoxaparin 100 U/kg every 24 hours, except for two patients, who received unfractionated heparin (UFH) at 10 U/kg/h given the concurrent high bleeding risk. PA was continued until discharge or thrombotic risk factor resolution/attenuation, whichever came earlier. No patient received acetylsalicylic acid. No patient died, and no thrombotic or bleeding complications were observed. If we had applied the ISTH‐endorsed recommendations, we should have offered PA to 14 patients (40%), including all our MIS‐C cases; also, we should have suggested our single patient with persistently elevated D‐dimer values to continue enoxaparin at home.¹¹

Table 3.

Clinical description of patients who received prophylactic anticoagulation

	Totale score	Description
Patient 1	3	14‐year‐old post‐pubertal girl (TS 4) with COVID‐19‐‐driven respiratory exacerbation of underlying cystic fibrosis, who scored 2 on the BQM in the early course of her hospital admission
Patient 2	8	7‐year‐old boy with MIS‐C, severe myocardial involvement requiring continuous inotropic support and placement of a non‐tunnelled right‐jugular CVC, who scored 1 on the BQM over the first 10 days of his hospital admission
Patient 3	3	15‐year‐old obese post‐pubertal boy (TS 5) with moderate COVID‐19, who scored 3 on the BQM over the course of his hospitalization
Patient 4	6	19‐year‐old boy (TS 5) with critical COVID‐19 requiring mechanical ventilation, who had recently undergone matched unrelated donor hematopoietic stem cell transplantation due to relapsed acute lymphoblastic leukemia and had suffered from multiple transplant‐related complications (including EBV reactivation and fungal pneumonia). A tunnelled CVC had been inserted approximately 6 weeks before the current presentation
Patient 5	3	6‐year‐old obese girl with MIS‐C and reduced ejection fraction (40%) but preserved mobility
Patient 6	6	9‐month‐old with a previously undiagnosed sickle cell disease with critical COVID‐19‐‐induced acute chest syndrome who required mechanical ventilation and was later escalated to veno‐venous extra‐cardiac circulation (switching at that point from PA to full therapeutic anticoagulation)

Abbreviations: BQM, Braden Q mobility score; CVC, central venous catheter; EBV, Epstein‐Barr virus; PA, prophylactic anticoagulant; TS, Tanner stage.

4. DISCUSSION

Multiple investigators have reported clinical and anatomopathological evidence of an increased incidence of thromboembolic events in adult COVID‐19 patients,2., 15. with up to 57% of critically ill individuals developing symptomatic venous or arterial thrombotic complications despite pharmacological prophylaxis.2., 16., 17., 18., 19. From what could be gained so far, COVID‐19–related coagulopathy is a multi‐pronged process consisting of endothelialitis and inflammatory‐driven endothelial dysfunction,²⁰ platelet activation²¹ and increased interaction with neutrophils and monocytes,²² increased tissue factor expression by monocytes,²³ increased neurotrophic extracellular trap (NET) production and delayed removal,²⁴ deregulated complement activation,²⁵ and fibrinolysis shutdown,²⁶ to name a few. Even though the exact sequence of events has not been clarified, a deregulated immunological process starting at the interface between alveoli and lung endothelium seemingly determines a local activation of hemostasic processes leading to platelets, neutrophils, and fibrin deposition (ie, in situ micro‐ and macrovascular pulmonary thrombosis), which likely contributes to the overall respiratory insufficiency and facilitates the systemic spread of inflammation and coagulopathy.

Many study groups have tried to define laboratory markers that could better describe and predict disease evolution, in order to guide treatments, among which one of the most studied is D‐dimer.²⁷ Unfortunately, D‐dimer testing has numerous well‐known drawbacks, including lack of specificity,²⁸ and its role in terms of VTE prediction is still controversial²⁹: it tends to increase in all inflammatory conditions, which are per se associated with an increase in the thrombotic risk, without a clear association between the intensity of D‐dimer alteration and the magnitude of the thrombotic risk; in fact, D‐dimer is not currently included in the usually applied VTE RAMs in the adult population.

From the earlier reports of adult COVID‐19 cases, D‐dimer levels upon hospital admission and/or its increase during hospital stay have been linked to worse clinical outcomes and thrombotic complications. These observations led many clinicians to use increasing D‐dimer values as a trigger to intensify their anticoagulation prophylaxis protocol, increasing the administered dose of anticoagulant up to full therapeutic levels,³⁰ in the hope of counteracting the procoagulant status so frequently observed. This practice lacks a solid evidence‐based foundation and may be associated with increased bleeding risk,31., 32. which is well known to accompany hyperinflammatory conditions in general and microvascular complications in particular. Also, the fact that the thrombotic process is mainly inflammatory‐driven (ie, immunothrombosis) and that inflammatory biomarkers (eg, CRP, interleukin‐6, and soluble triggering receptor expressed on myeloid cells [s‐TREM]) have proven to be even better predictors than D‐dimer in terms of clinical outcome³³ point to a multimodal approach, possibly combining agents active at different levels of the immuno‐thrombotic process,34., 35. as better suited to overcome this difficult challenge. Luckily, multiple randomized trials are ongoing to provide the practicing community with solid data.

Pediatric COVID‐19 patients usually follow a less severe course, with a very low mortality rate (less than 0.1% according to data released by the Italian National Institute of Health [ISS]); also, critically ill children have better outcomes than their adult counterparts.9., 10., 36. Thrombotic complications do not appear to occur more often than what is expected in hospitalized children: a Italian observational study (unpublished data, courtesy of Dr Garazzino) recorded a prevalence of 1 VTE in more than 350 hospitalized pediatric COVID‐19 cases, compared to an estimated incidence of hospital‐acquired VTE in the general pediatric population of approximately 1/200.37., 38. MIS‐C apparently conveys a risk of thrombosis of approximately 3.5% (which may be overestimated, given the paucity of reports that detail the observed numbers of VTE),³⁹ likely driven by a pro‐coagulant highly inflammatory milieu and increased venous stasis secondary to decreased myocardial function. In fact, this observed risk is also lower than what would be expected in pediatric myocarditis (approximately 6%⁴⁰).

Our single‐center cohort appear to be similar in composition to the comprehensive report by Duarte‐Salles et al⁴¹ (available in pre‐print), with a prevalence of severe respiratory manifestations of 17% among hospitalized patients with COVID‐19. In terms of laboratory alterations, our COVID‐19 cohort displayed only mildly increased D‐dimer values across all severity categories, consistent with previous descriptions.9., 10., 42. Moreover, our dynamic evaluation of laboratory data confirmed that clinical worsening‐‐‐albeit mirrored by spiking D‐dimer values, especially in the MIS‐C cohort‐‐‐was not associated with fibrinogen consumption, with all patients recovering without thrombotic sequelae. On the other hand, our MIS‐C cohort showed markedly increased values of both CRP and D‐dimer, as previously reported, which rapidly decreased within 48 to 72 hours of corticosteroid therapy administration; only a single patient, who never received steroids as she did not show significant cardiovascular involvement, maintained increased D‐dimer values upon disease resolution (2637 ng/mL). Again, these observations corroborate the view of D‐dimer in children being a mere low‐specificity marker of the intensity of the inflammatory response, without a direct link with increased thrombotic risk. Indeed, contrary to the adult experience, D‐dimer has generally proved to be an unreliable biomarker for diagnosing or predicting thrombotic complications in children: it shows low accuracy in identifying pulmonary embolism⁴³ and in predicting venous thrombosis recurrence.⁴⁴ Finally, a recent report by Al‐Ghafry et al⁴² described viscoelastic testing consistent with a pro‐thrombotic state (increased maximum clot firmness [MCF] in both EXTEM and FIBTEM) in eight COVID‐19 pediatric patients, with no correlation between D‐dimer values and MCF. Thus, D‐dimer may be a poor choice as a parameter to guide therapeutic decisions in terms of anticoagulant prophylaxis.

We and others have observed that the pro‐coagulant state induced by COVID‐19 is unlikely to translate in clinically relevant thrombotic complications in children. Pediatrics’ central dogma is that children are not just little adults. This tenet holds particularly true in the field of thrombotic disorders, as it is a widely recognized notion that children develop far fewer thrombotic complications compared to adults, even in high‐risk scenarios (eg, severe trauma, complex orthopedic surgeries, critical diseases).⁴⁵ The reasons behind this reduced susceptibility are still not completely understood and likely lie on both a more robust array of natural anticoagulants (ie, increased levels of alpha‐2‐macrogloulin) and healthier vascular linings (ie, reduced cumulative exposure to substances that are toxic to the endothelium‐‐‐pollution, smoking, metabolism by‐products).⁴⁶ So, contrary to adult indications, we believe that universal pharmacologic prophylaxis in the pediatric population is unwarranted and that raised D‐dimer values should not be taken into consideration, given its mere role as a marker of the acute‐phase response in children. On the other hand, as hinted by our data and corroborated by recent case series36., 47. this disease tends to follow a more severe course in the presence of comorbidities (ie, obesity, active malignancy, sickle cell disease) that enhance the baseline thrombotic risk, especially by priming the vascular surface toward endothelial dysfunction. This baseline risk would be boosted by COVID‐19 and could be counteracted by a targeted PA course with enoxaparin or UFH, which might also contribute anti‐inflammatory activity.⁴⁸ Therefore, we suggest to evaluate every child hospitalized with COVID‐19 for possible concurrent pro‐thrombotic risk factors, and to consider a personalized PA strategy accordingly. Also, given the concerning report by Duarte‐Salles et al⁴¹ of a bleeding rate of 2% to 3% in hospitalized children with COVID‐19 (in a cohort in which more than 30% of patients received some sort of anticoagulation), we suggest to carefully balance the thrombotic and the hemorrhagic risks for every child.

Our study has several limitations. First, it comprises a small number of patients, so its results may not be generalizable and should be interpreted with caution. Second, our patient cohort was quite heterogeneous, including three patients with acute lymphoblastic leukemia at different stages of their treatment protocol, which impacted the platelet valued recorded (both in terms of being surreptitiously low and of being confounded by transfusions). It must be noted, though, that consistent with the absence of a consumptive process, none of those children showed platelet transfusion refractoriness in the course of COVID‐19. Third, D‐dimer assays in pediatrics are prone to pre‐analytic confounders (eg, quality of the blood draw, age, baseline conditions), which may limit their reliability. Though acknowledging these possible drawbacks, we tried to minimize them by performing multiple assessments for each patient: the fact that our data are consistent and show a time‐dependent and pathophysiologically plausible trend corroborates their validity; also, excluding patients with persistently raised D‐dimer values even after disease resolution did not change the overall results. Also, as the clinical significance of asymptomatic VTE in children is controversial, we decided not to systematically screen for thromboembolic complications, relying instead only on clinical judgment; thus, we cannot exclude that an asymptomatic event might have been missed. Finally, we did not consistently check for abnormalities in other coagulation parameters (eg, protein C, protein S, antithrombin, viscoelastic tests), thus we may have overlooked important aspects of COVID‐19–associated coagulopathy in children. Given the speculative nature of this study without an immediate clinical benefit for our participants, however, we opted not to burden patients with excessive blood draws. Following the results of this preliminary analysis, a protocol was devised to include these tests in the prospective assessment of moderate to critically ill cases.

In conclusion, D‐dimer values apparently do not distinguish mildly affected COVID‐19 patients from more severely affected cases. On the other hand, D‐dimer and CRP values accurately distinguish MIS‐C cases from COVID‐19 cases, as reflected by a heightened immunological/inflammatory component in the former. We did not find evidence in either case that a consumptive coagulopathy was in place, given only slight alteration in platelet counts and fibrinogen values. Contrary to adult guidance, universal anticoagulant prophylaxis in hospitalized children suffering from COVID‐19 is not advised, but might be considered in highly selected cases with multiple concurrent pro‐thrombotic risk factors without taking into account D‐dimer values.

CONFLICT OF INTEREST

No author has any real or potential conflict of interest to disclose.

AUTHOR CONTRIBUTIONS

G. Del Borrello designed the study, interpreted the datam and drafted the manuscript; I. Giraudo, C. Bondone, G. Pruccoli, M. Spadea collected and analyzed the data, and contributed to drafting the manuscript; M. Denina, S. Garazzino, C. Linari, F. Mignone, B. Pollio, and P. Saracco helped interpret the data and critically revised the manuscript. All authors approved the final version of the manuscript.