DVT and PE cause significant mortality and morbidity in adult patients (1). The important sequelae of DVT and PE include recurrent thrombosis, PPS, and complications of anticoagulant therapy (2, 3). Because of the clinical importance of and frequency with which DVT and PE occur in adults, their epidemiology and response to a multitude of therapies have been extensively studied (1).

Until recently, therapy for children with DVT or PE was directly extrapolated from adult guidelines. The major reason for this is that whereas DVT and PE in children represent a significant clinical problem, the frequency with which they occur was such that clinical trials on the same scale as those performed in the adult population were not possible. However, justification for the continued use of adult protocols in children requires that the epidemiology and the clinical consequences are similar for both patient populations (4).

In 1994, the first analysis of the Canadian Registry of Venous Thromboembolic Complications in Children (5) was published. The analysis described 137 consecutive patients, providing the first prospective data outlining the incidence, age distribution, associated conditions, diagnostic modalities, location, and treatment patterns for children with DVT and PE. The results defined important differences in the epidemiology of DVT/PE in children compared with adults. Specifically, high-risk pediatric populations, such as children with CVLs, were identified. However, data on outcome were lacking, because of the short duration of follow-up available at that time (median, 18 mo; range, 6 to 36 mo). Through the continued development of the thrombophilia network, the Canadian Registry has now collected prospective data for 6 y on children with DVT/PE. The focus of the current report is follow-up of these patients to determine their clinical outcome. The incidence and features of the major complications of DVT/PE, which are mortality, recurrent thrombotic disease, and PPS, were assessed.


The Canadian Paediatric Thrombophilia Registry was initiated in 1990, with the co-operation of 15 tertiary-care pediatric centers throughout Canada (see Appendix). Patients in this report were entered from May 1990 until November 1996 for a total of 6 y. Children from 1 mo to 18 y were eligible for entry if a PE, or DVT in the upper or lower venous system, was objectively documented. Newborns and infants <1 mo old were entered into a separate international registry and are not included in this analysis. Children with thrombotic occlusion of the veins in the CNS, portal system, renal veins, or other nonextremity venous systems were not included in this registry. There were no other restrictions for entry into the registry. Enrollment in the registry did not mandate the use of specific diagnostic or therapeutic protocols. The methods of initial data collection were as previously described (5). Regular follow-up data were obtained at yearly intervals. In cases of death, cause of death was documented along with status of the thrombotic event at time of death when possible. For survivors, details on maintenance therapy, complications of treatment, and complications related to the primary thrombotic event were recorded. A detailed history and physical examination for evidence of PPS was performed. Clinical evidence of recurrence was noted, and if indicated, radiologic confirmatory studies were performed.


Cohort characteristics.

Four hundred five patients were entered from May 1990 to November 1996. The male to female ratio was 1.1:1. Follow-up data were available on 356 of the 405 patients (88%). Follow-up data on 49 patients were not available because of change of address, transfer to adult units, and failure to keep appointments. Some patients represent those with newly registered diagnosis who had not reached their first review period. Median follow-up was 2.86 y, with a range of 2 wk to 6 y. Patient demographics were not dissimilar to those previously reported (5). The age distribution is shown in Figure 1.

Figure 1
figure 1

Age distribution of total cohort of 405 children.

Almost two thirds (60%) of thrombosis was upper system thrombosis, frequently related to CVLs. Sixty-nine children had PE, 39 of which were associated with documented thrombosis at another site. In 39 cases, PE was CVL-related (upper and lower systems), with the next biggest group being cardiac surgery patients with central or intracardiac thrombosis. In all cases, PE was symptomatic. Ultrasound or echocardiography was used as a diagnostic test in 75% of patients, venography in 37%, ventilation-perfusion scans in 20%, angiography in 5%, autopsy in 2.7%, and other diagnostic tests in 6.4%. Only 3% of children had spontaneous DVT/PE, with >75% of children having two or more risk factors for thrombosis.

The underlying conditions for the entire cohort, compared with the children with adverse outcomes, are shown in Table 1. The treatment received for DVT/PE varied greatly. Sixty percent of children received standard heparin/warfarin therapy of variable duration. Only 10% received thrombolytic therapy.

Table 1 Underlying conditions for thrombosis Abbreviations: TE, thromboembolic event; BMT, bone marrow transplant; MVA, motor vehicle accident; BCP, birth control pill.

All-cause mortality.

All-cause mortality was 16% (65 of 405).The major cause of death in 56 of these 65 (86%) patients who died was the underlying disease. The most common underlying disorders causing death were cancer (n = 22, 34% of total deaths), cardiac disease (n = 13, 20%), infection (n = 8, 12%), and other causes (n = 13, 20%).

Mortality directly attributable to DVT/PE.

Nine patients (14% of total deaths) died as a direct result of thromboembolic disease for an overall mortality rate caused by DVT/PE of 2.2% (9 of 405). Seven of the deaths attributable to DVT/PE occurred between 1990 and 1993, with two between 1994 and 1996. The ages of the children who died were distributed evenly from 2 mo to 12 y. The mechanism of death in these children was either hypoxia or obstruction of cardiac output. The sites of the fatal thrombosis were pulmonary in seven, central venous in one, and intracardiac in one.


Thirty-three (8.1%) patients had a total of 35 local recurrences or second thrombotic episodes that were objectively documented. The mean time from first thrombotic event to recurrence was 6 mo (range, 3 mo to 5 y). In seven patients, one of their thrombotic events was in fact a CNS thromboembolism (stroke).The age distribution at first thrombosis for patients with recurrence is shown in Figure 2. Five patients were on warfarin at the time of recurrence, but only three had therapeutic international normalized ratios (INRs) at that time.

Figure 2
figure 2

Patient age at time of diagnosis of initial thrombosis for patients with recurrence (cross-hatched bars) and with PPS (black bars). Multiple ages have been grouped to improve clarity.


PPS was diagnosed clinically on the basis of signs and symptoms of pain, swelling, and brawny discoloration of the limb involved. Fifty (12.4%) children had evidence of PPS during their follow-up. The average duration of follow-up for those with PPS was 3.11 y (compared with 2.86 y for entire cohort). This difference is not statistically significant. PPS occurred secondary to lower venous system thrombosis in 70% of cases (35 children). Nine children (18%) with PPS had had recurrent thrombosis. The age distribution at time of thrombosis for the children who developed PPS is shown in Figure 2.


Recent advances in tertiary-care pediatrics have led to a dramatic improvement in survival for children with serious illnesses such as congenital heart disease and cancer (6). Successful therapy for these previously deadly illnesses requires aggressive surgical and pharmacologic intervention, which has introduced a number of secondary complications previously not common in children (7). One of the most frequent complications is DVT and PE (5). The current report provides intermediate follow-up data on the outcome of DVT/PE in children (8). This information is crucial to the development of rational guidelines for the management of DVT/PE in children.

The Canadian Childhood Thrombophilia Registry was established in 1990. The first analysis of the registry provided prospective data detailing the epidemiology of venous thrombosis in children (5). The epidemiology of neonatal thrombosis has also been recently described in a report by an international registry (9). These studies confirmed major differences in the incidence, underlying diseases, location, and treatment profiles of DVT/PE in infants and children compared with adults.

Further recent evidence shows that children, and in particular infants, respond differently to anticoagulant and thrombolytic drugs compared with adults (1013). Likely, these differences reflect normal developmental changes in the hemostatic system (14). Dose requirements for heparin and oral anticoagulants are age and weight dependent (11, 12). The frequency of adverse effects of anticoagulant and thrombolytic therapy also differ in children.

These differences mean that optimal management of venous thrombosis in children is likely to be different from optimal management of venous thrombosis in adults. However, to develop pediatric trials of either prophylactic or therapeutic strategies, we must know the clinical consequences of DVT/PE in children.

In our large cohort of 405 patients, the all-cause mortality in children with DVT/PE was 16%. This is similar to the 2-y all-cause mortality in adult patients with DVT (15). Underlying cancer is frequently the cause of death in children with DVT, similar to the cause of death in adults with DVT. Nine children (2.2% of the total cohort) died as a result of their thrombosis, all of whom had CVL-related thrombosis in the upper venous system. Most CVLs in children are placed in the upper system, with the CVL tip in the superior vena cava or right atrium. A large thrombus can easily extend into the heart and pulmonary vasculature. Death occurs because of obstruction of cardiac output, with resultant hypotension or hypoxia. Four of the children who died as a result of their thrombosis received no antithrombotic treatment. This was because of sudden catastrophic thrombosis. These findings emphasize the importance of CVL-related thrombosis and the need for adequate and safe prophylactic protocols to be developed. The majority of the deaths occurred in the first 3 y of the study.

Recurrent thromboembolic disease occurred in 8.1% of the patient cohort, compared with a 2-y recurrence rate of 17.5% recently reported in adult patients with their first DVT (15). The decreased rate of recurrence in this study may be because the underlying contributing factors in children with DVT are more transient than those in adults. Children with congenital heart disease frequently return to normal health and activity after surgery. Childhood cancers are frequently cured, compared with adults in whom cancer is frequently controlled but less commonly cured.

The distribution of age at original diagnosis for children with recurrent thrombosis differed from that of the entire cohort in that there were fewer young infants who had recurrent disease. This may result from different risk factors in younger compared with older children, or perhaps different abilities to report symptoms of recurrence. Further studies will be required to determine whether the recurrence rate in children continues to increase with increased duration of follow-up, as is the case in adults with venous thrombosis. The lack of standardized investigation and treatment among patients enrolled in the registry makes identification of predictive risk factors for recurrence (for example, presence of congenital thrombophilia) impossible. Congenital prethrombotic disorders were found in four of 33 children with recurrent disease but in only seven of the entire cohort of 405. However, less than one third of the cohort was assessed for congenital prethrombotic disorders, and children with recurrent thrombosis were far more likely to be investigated. Similarly, the comparative effects of the different treatments used cannot be assessed adequately in an observational cohort in which there is no control group for confounding factors. These questions are best answered in prospective cohort or randomized treatment trials (16).

The frequency of PPS in the cohort (12.4%) was most likely an underestimate. Multiple studies in adults show that PPS frequently develops late after DVT (15). We anticipate that the incidence of PPS will increase during subsequent years. PPS was most frequently diagnosed secondary to lower limb DVT. The venous drainage of the lower limbs is more adversely affected by gravitational dependence than the upper limbs, increasing the likelihood of dependent swelling and pain. PPS is a clinical diagnosis, which depends to some degree on patients reporting the classic symptoms. PPS was likely underdiagnosed in young infants because of the infant's inability to report symptoms and the difficulties of accurately assessing swelling in small infants. Underdiagnosis in infants may explain the skewed age distribution of patients with PPS, with a predominance of teenagers. An alternative explanation is that infants have protective mechanisms that reduce the risk of PPS, although such mechanisms remain to be elucidated. As stated previously, this study did not have the design or sufficient power to determine predictive factors or the effects of treatment in reducing PPS (16). The overall impact of PPS on growing children who are still developing motor skills is potentially increased compared with adults.

In conclusion, the Canadian Childhood Thrombophilia Registry has followed a large cohort of children with DVT/PE for a mean follow-up of 2.86 y. This study has provided prospective data specific to children with DVT/PE. The thrombosis-specific mortality was 2.2%, with significant morbidity in terms of recurrent thrombosis (8.1%) and PPS (12.4%). The thrombosis-specific mortality and frequency of recurrent thrombosis are similar to those reported in the original cohort of 137 patients. The frequency of PPS is reduced compared with that reported in the original cohort of 137 patients. Given the substantially increased patient numbers and increased duration of follow-up, the current results more likely represent the true frequency of mortality and morbidity from childhood DVT. These results confirm that venous thrombosis, including CVL-related thrombosis, have serious consequences for children affected. Further studies are required to identify specific risk factors for these complications and to devise effective prophylactic and therapeutic strategies. Multicenter international randomized clinical trials are required to achieve these goals.