Of the 359 patients recruited, 23 did not survive the cardiac surgery. The remaining 336 patients (163 males) were included in the final analysis, 102 (30%) of whom had cyanotic congenital heart disease. Cardiac surgery was performed at a median age of 1.7 years (range 1 day to 22.3 years) and a median weight of 10 kg (range 2.2–66 kg). The operations performed were categorised into firstly, repair of left to right shunts (n = 195); secondly, repair of lesions with right ventricular outflow tract obstruction (tetralogy of Fallot with or without pulmonary atresia, ventricular septal defect with infundibular pulmonary stenosis) (n = 84); thirdly, arterial switch operation for transposition of the great arteries (n = 15); fourthly, Fontan-type procedures (n = 10); and fifthly, miscellaneous procedures including correction of anomalous pulmonary venous drainage, prosthetic valve replacement, repair of left ventricular outflow obstruction, Ross procedure, and Norwood stage I operation (n = 32).
Seventy seven of the 336 (23%) patients developed pericardial effusion postoperatively. The amount of effusion was small in 34, moderate in 24, and large in 19 patients. The effusion was already circumferential in 55 of 77 (71%) at first presentation. Its onset was at 11 (7) days after surgery. Overall, the diagnosis was made on or before day 14 and day 28 after surgery in 79% (61 of 77) and 97% (75 of 77) of patients who developed effusion, respectively. Importantly, 12% (4 of 34), 25% (6 of 24), and 32% (6 of 19) of patients with, respectively, small, moderate, and large amounts of effusion presented beyond the second week after operation (fig 1).
Cumulative percentage of patients who developed pericardial effusion versus the time of onset.
Clinical symptoms or signs were present in only 18 (23%) patients. These were fever in 10 patients, gastrointestinal disturbance such as vomiting or abdominal discomfort in two, signs of fluid retention in three, irritability in two, and evidence of tamponade in one. Patients who had a large amount of effusion were more likely to be symptomatic than those with only a small to moderate amount (47.4% v 15.5%, p = 0.01). Likewise, patients with circumferential effusion tended to be more symptomatic than those with loculated effusion (29.7% v 9.0%, p = 0.077).
Conservative treatment with fluid restriction and diuretic treatment in 44 (57%) patients with a small to moderate amount of effusion resulted in resolution in all of these patients. Anti-inflammatory medications were started in 21 patients (27%) who, albeit asymptomatic, had a moderate amount of effusion with evidence of progression. Aspirin was prescribed to seven patients and steroid was given to 14. Eighteen patients responded, while three eventually required pericardiocentesis. Pericardiocentesis, either by a surgical or a percutaneous approach, as the initial treatment was performed in 12 (16%) symptomatic patients with a large amount of effusion. The median duration for effusion to resolve completely was eight days (range 1–394 days), with complete resolution achieved in 83% (64 of 77) of patients within four weeks of diagnosis.
Table 1 summarises the demographic, clinical, and perioperative variables of patients with and without pericardial effusion. Significantly more female (28.9%) than male (16.6%) patients developed pericardial effusion (p = 0.009), with a relative risk of 2.05 (95% confidence interval (CI) 1.2 to 3.5). The prevalence of pericardial effusion after Fontan-type procedures (60%) was significantly higher than that after other types of surgery (p = 0.037) (fig 2). Furthermore, pericardial effusions that developed after a Fontan-type procedures were at least moderate to large. Postoperatively, a greater pericardial drain output within the first four hours tended to be associated with a higher risk of developing later effusion (p = 0.056). Warfarin, given to 18 patients who had undergone valve replacement (n = 8), Fontan-type procedures (n = 6), and patch repair of an atrial septal defect (n = 4), also conferred a significant risk of developing a pericardial effusion (relative risk 4.48, 95% CI 1.78 to 12.33, p = 0.002). Logistic regression similarly identified sex (β for male = −0.63, p = 0.037) and warfarin (β = 1.73, p = 0.009) as significant determinants. Nonetheless, it is important to note that in this cohort, warfarin was started in significantly more female than male patients (8.7% v 1.8%, p = 0.007). Furthermore, patients who had undergone arterial switch operation, which had the lowest incidence of pericardial effusion, were mostly male (73.3% v 26.7%, p = 0.068).
Comparisons of demographic, clinical, and perioperative variables between patients with and without development of pericardial effusion
Prevalence of pericardial effusion after open heart surgery for different types of congenital heart disease. Solid areas represent patients who developed pericardial effusion. ASO, arterial switch operation; RVOT, right ventricular outflow tract.