Continuous renal replacement therapy (CRRT) is currently the most widely used technique for extrarenal filtration in critically ill children, because it allows continuous and programmed fluid removal 12345.

Although a number of studies have demonstrated that these techniques are useful and safe in critically ill children of any age 4567, complications do occur 8. Children are at a higher risk than adults for developing complications associated with CRRT due to the difficulty of venous catheterization with the large-caliber catheters required for the technique, the large extracorporeal volume of the system (filters and lines), which predisposes to hypotension at the time of connection, and the need for a more accurate control of volumes in order to avoid fluid and electrolyte disturbances.

There are no studies that have prospectively analyzed the complications or risk factors in children on CRRT.

The objective of the present study was to determine the incidence of complications in children requiring CRRT and to analyze the predisposing risk factors.

An analysis was performed of the data from a prospective, single-center register of critically ill children treated using CRRT. The study was approved by the local Institutional Review Board and due to the characteristics of the study inform consent of patients was not considered to be necessary. Between January 1996 and June 2009, CRRT techniques were used in 174 children (105 boys (60.3%) and 69 girls (39.7%)) with a mean (standard deviation) age of 52.3 (63.8) months and weight of 17.6 (18.2) kg; 43.7% of the patients were under one year of age. The most common conditions in patients requiring CRRT were heart disease (55.7%), particularly during the postoperative period of cardiac surgery, and sepsis (19.5%).

Two different renal replacement pumps were used to perform CRRT: the BSM321C (Hospal^®, Barcelona, Spain) in the first 35 patients and the Prisma (Hospal^®, Barcelona, Spain) in the remaining 139. The caliber of the catheters used was between 4F and 11F and the filters were between 0.04 m² and 0.9 m², according to the age and weight of the patient. All patients received continuous anticoagulation with heparin to maintain an activated coagulation time between 130 and 200 seconds. Other anticoagulant or antiaggregant drugs (citrate, warfarin, aspirin, prostacyclin) were not administered.

The following data were gathered prospectively in all patients on starting CRRT: age; weight; sex; diagnosis; severity scores, pediatric risk of mortality (PRISM II) score 9, pediatric index of mortality (PIM I and II) score 10, pediatric logistic organ dysfunction (PELOD) score only from 2001 11; number of organ failures; blood pressure; need for vasoactive drugs; dose of dopamine and adrenaline; lactic acid levels; pH and base excess; levels of creatinine, urea, alanine transaminase (ALT), bilirubin, sodium, potassium, chloride, calcium, phosphorus, magnesium, albumin and platelets; and type of filtration pump used. The type of connection to CRRT was determined by the physician responsible for the patient. In some cases the connection was made directly to the circuit that had previously been primed using normal saline, in others, after purging with heparin, the system was flushed using 5% albumin before connection to the patient. During filtration, a daily record was kept of the technique used (hemodiafiltration or hemofiltration), maximum dose of heparin, ultrafiltration rate, life of each filter, electrolyte levels, complications related to the CRRT, and mortality during admission to the pediatric intensive care unit.

The following complications were analyzed: 1) complications of catheterization, defined as hemorrhage with a fall of more than 2 g/dL in the hemoglobin concentration and/or hypotension or the need for transfusion and/or withdrawal of the catheter from that site, thrombosis, pneumothorax, and altered limb perfusion; 2) hypotension on connection to the filter, defined as a fall in the mean blood pressure (MBP) of more than 20 mmHg over baseline and/or an MBP more than two standard deviations below the normal values for age and that required volume expansion and/or an increase in the dose of vasoactive drugs that the patient was receiving in the first 60 minutes after the connection to CRRT; 3) significant hemorrhage, defined as a fall of more than 2 g/dL in the hemoglobin concentration in the first 24 hours after bleeding and/or hypotension and that required packed red cell transfusion; and 4) electrolyte disturbances including hyponatremia (sodium <130 mEq/L), hypernatremia (sodium >150 mEq/L), hypokalemia (potassium <3 mEq/L), hyperkalemia (potassium >5.5 mEq/L), hypochloremia (chlorine <95 mEq/L), hyperchloremia (chlorine >115 mEq/L), hypocalcemia (total calcium <8 mg/dL), hypercalcemia (total calcium >12 mg/dL), hypophosphatemia (phosphate <4 mg/dL in children <6 years and phosphate <3 mg/dL in children >6 years), hyperphosphatemia (phosphate >7 mg/dL), hypomagnesemia (magnesium <1.5 mg/dL), and hypermagnesemia (magnesium >3 mg/dL). An analysis was performed of the changes in the electrolyte levels during the first three days of CRRT. The incidence of complications of CRRT between the first seven years and the second seven years of the study was compared.

The statistical analysis of the results was performed using the SPSS statistical package version 14.0. Pearson's chi-squared test and Fisher's exact test were used to compare percentages and the Mann-Whitney test to compare values with a non-parametric distribution. Significance was taken as a P value less than 0.05.

Our study is the first that has prospectively investigated complications related to CRRT in critically ill children and that has analyzed the factors associated with these complications.

The percentage of complications of venous catheterization was similar to that found in other studies of central line catheterization in pediatric patients, despite the fact that the catheters necessary for CRRT are larger 131415. The complications of catheterization were more common in smaller children because catheterization is more difficult in these patients and because the caliber of the catheter used in infants is proportionally larger than in older children. In contrast to other series, we did not find differences in the rate of complications between the use of veins in the upper body (jugular and subclavian) or lower body (femoral) 1315. A recent study in adults that compared jugular and femoral venous access for acute CRRT found that the incidence of hematomas was higher in jugular than in femoral access, with no significant differences in the rates of infection secondary to catheterization 16. The incidence of catheter-related infection was not analyzed in the present study. A recent study has shown that ultrasound-guided central venous catheter placement decreases the complications of catheterization, although we have not used this method in our patients 17.

Hypotension after connection to the CRRT system was the most common complication; it is more common in children because the extracorporeal volume of the circuit and filter used for CRRT represents 10 to 5% of a patient's blood volume 5. The circuits used in our study have a priming volume (including the filter) of 50, 100, and 130 mL depending on the surface area of the filter used (0.04, 0.6, or 0.9 m², respectively). Although the circuit priming volume is proportionally larger in children of lower weight, we did not find any relation between the frequency of hypotension and age, weight, or surface area of the filters. The design of filters and circuits with a low priming volume is an essential factor in the reduction of hemodynamic complications at the time of connection to the system.

Patients with previous hemodynamic alterations theoretically could have more hypotension after connection to the CRRT. However, surprisingly in our study, we have not found any individual factors associated with hypotension after CRRT connection. It is possible that, although individually each risk factor is not associated with hypotension, the combination of several risk factors such as the extracorporeal volume of the circuit and filter and the previous hemodynamic alterations could influence the development of hypotension after connection to CRRT.

There are a number of techniques that can be used to attempt to reduce the risk of hypotension at the time of connection, such as priming the circuit with whole blood or colloids, although there are no studies that have analyzed their efficacy. Patients on CRRT usually received many blood transfusions. To reduce the risks of transfusion, we decided to more frequently prime the circuit with 5% albumin rather than with whole blood if the hemoglobin is not very low. However, in our study, we did not record in which children the circuit was primed with albumin and we cannot therefore analyze the efficacy of this measure. Further studies are necessary to determine the efficacy of circuit-priming methods in the reduction of hypotension at the time of connection.

The treatment of hypotension was different depending on the situation of each patient. Generally, we used volume expansion with colloids 10 to 20 ml/kg as the first measure. If the hypotension was severe we also increased the vasoactive drugs that the patient received, and when the haemoglobin level was low we also transfused packed red cells.

Some authors have reported the onset of a bradykinin release syndrome when using filters with the AN69 membrane primed with blood; the syndrome presents as acute hypotension and can be avoided by raising the pH 18. Although we use AN69 membranes, we have not had this complication because we do not prime the circuit with blood.

Hypotension can also occur if excessive ultrafiltration is programmed 5 or if the machine systems that measure the volumes function incorrectly. To prevent this, both the fluid balances measured by the CRRT machine and the clinical state of the patient should be monitored continuously. According to our protocol, nurses measured hourly the input and output fluid balance and checked the ultrafiltrate volume registered by the machine. Furthermore a continuous clinical vigilance was performed. According to these data the programming of the ultrafiltration was changed by the intensivist. We think that for this reason we have not found any complications of excessive ultrafiltration in our patients.

The need for anticoagulation of the CRRT system, associated with the frequent alterations of coagulation that occur in these patients, increases the risk of hemorrhage. Both CRRT and heparin can produce a fall in the platelet count, as found in our study, or an alteration of platelet function. Heparin continues to be the most widely used method of anticoagulation in CRRT 19, although it has been suggested that anticoagulation using sodium citrate could reduce the risk of heparin-related hemorrhage; however, sodium citrate increases the risk of hypocalcemia and alkalosis 520.

Although premature coagulation of the CRRT filter is more common in children 21, 10% of our patients presented clinically significant hemorrhage, and there was a higher mortality among these patients. In our study, we found no relationship between the incidence of hemorrhage and the platelet counts or doses of heparin used. However, an important limitation in our study is that no analysis was performed of a possible relationship between hemorrhage and other disturbances of coagulation. Moreover, it is also possible that patients with high risk of haemorrhage received a low dose of heparin and this fact could influence to not find relationship between heparin dose and bleeding. Hemorrhage in critically ill patients on CRRT is probably the consequence of several factors: a coagulation disorder, altered tissue perfusion caused by the underlying disease, and the alterations of coagulation caused by the extracorporeal circuit and anticoagulation 822.

Electrolyte disturbances are very common in critically ill children 23. CRRT can be used to correct severe electrolyte disturbances, but can also produce them 24. The risk is higher if inappropriate dialysis and/or replacement fluids are used 25, as occurred in one of our patients 12. In our study, despite using balanced solutions, there was a significant fall in the levels of sodium, chloride, and phosphate, leading to the need to increase the concentration of these electrolytes in the dialysis and replacement fluids or to administer intravenous supplements. When balanced solutions are used high dose of dialysis and/or replacement fluids should not produce more electrolytes disorders. However we have not analyzed if electrolytes disorders were associated to the intensity of fluid dose prescriptions. Hyponatremia may develop if the dialysis and replacement fluids do not compensate the negative sodium balance 8. In a previous study we found a very high incidence of hypophosphatemia in children on CRRT; this was due to the high efficacy of these techniques and the fact that the usual replacement and dialysis fluids do not contain phosphate 26. The addition of phosphates to replacement and dialysis fluids did not cause any instability of the solutions or other complications, and reduced the incidence of hypophosphatemia and the need for intravenous phosphate supplements 26. Therefore, as electrolyte disturbances are common in children on CRRT, periodic controls of their blood levels should be performed and the concentration in the replacement and dialysis fluids should be monitored closely in order to detect errors in the preparation of the fluids.

Other complications have been reported in patients on CRRT, such as alkalosis secondary to the bicarbonate content of the dialysis and replacement fluids 27, and errors of drug dose 28. These complications were not analyzed in our study. Another limitation of our study is that we did not determine the incidence of hypothermia, which is more common in children on CRRT due to extracorporeal radiant heat exchange, or catheter-related infection 8.

We conclude that the frequency of complications in children on CRRT is high, and some of these complications can be serious. The most common are hypotension at the time of connection and electrolyte disturbances. The hemodynamic state of children on CRRT should therefore be monitored closely and frequent controls of the electrolyte concentrations should be performed.

• The frequency of complications in children on CRRT is high, and some of these complications can be serious.

• The most common complications are hypotension at the time of connection and electrolyte disturbances. The hemodynamic state of children on CRRT should be monitored closely and frequent controls of the electrolyte concentrations should be performed.

ALT: alanine transferase; CRRT: continuous renal replacement therapy; MBP: mean blood pressure; PELOD score: paediatric logistic organ dysfunction score; PIM score: pediatric index of mortality score; PRISM score: pediatric risk of mortality score.

The authors declare that they have no competing interests.

MJS and JLH conceived the study and participated in the design, data collection and analysis, and drafting of the manuscript. JU, MJS, YB and MB participated in the data collection and analysis of data, and drafting of the manuscript. JMB participated in the design of the study and performed the statistical analysis. All authors read and approved the final manuscript.