PEDIATRICS Vol. 107 No. 6 June 2001, pp. 1309-1312

Outcome in Children Receiving Continuous Venovenous Hemofiltration

Stuart L. Goldstein, MD^*, Helen Currier, RN, CNN, Jeanine M. Graf, MD^§, Carmen C. Cosio, MD^§, Eileen D. Brewer, MD^*, and Ramesh Sachdeva, MD^§

From the ^* Department of Pediatrics, Renal Section, Baylor College of Medicine;  Texas Children's Hospital Renal Dialysis Unit; and the ^§ Department of Pediatrics, Critical Care Section, Baylor College of Medicine, Houston, Texas.

	ABSTRACT

Top Abstract Methods Results Discussion References

Objective.  Continuous venovenous hemofiltration (CVVH) alone or with dialysis (D) has become an important supportive therapy for critically ill children with acute renal failure. Previous reports of pediatric patient outcome either mix CVVH/D with other renal replacement modalities or do not examine severity of illness. The current study examines only outcomes of children receiving CVVH/D using Pediatric Risk of Mortality (PRISM) scores to control for severity of illness.

Patients.  Twenty-one patients (mean age: 8.8 ± 6.3 years; mean weight: 28.3 ± 20.8 kg) received 22 courses of CVVH/D.

Outcomes.  Nine (42.8%) of 21 patients survived. Nine (75%) of 12 deaths occurred within 25 days of pediatric intensive care unit (PICU) admission. Mean PRISM score at PICU admission and CVVH initiation were 13.1 ± 5.8 and 15.4 ± 8.9, respectively. Mean patient weight, age, PRISM score at PICU admission and at CVVH/D initiation, maximum pressor number, estimated glomerular filtration rate at CVVH/D initiation and change in mean airway pressure did not differ between survivors and nonsurvivors. The degree of fluid overload at CVVH/D initiation was significantly lower in survivors (16.4% ± 13.8%) compared with nonsurvivors (34.0% ± 21.0%), even when controlled for severity of illness by PRISM score. Mean cost of providing CVVH/D accounted for only 1% of total PICU cost per patient.

Conclusions.  The pattern of early multiorgan system failure and death, minimal relative cost of CVVH/D provision, and potential for improved outcome with initiation of CVVH/D at lesser degrees of fluid overload are factors that may support early initiation of CVVH/D in critically ill children with acute renal failure.  Key words:  continuous venovenous hemofiltration, continuous renal replacement therapy, dialysis, critically ill children.

Continuous renal replacement therapy (CRRT) has become an important supportive therapy for critically ill children with acute renal failure (ARF), oliguria, and fluid overload. Although a conventional hemodialysis (HD) treatment occurs over 3 to 4 hours, CRRT treatments occur continuously, allowing for slower fluid removal rates than with HD. Thus, CRRT is a therapy better suited for fluid removal in children with hemodynamic instability. A poll of pediatric nephrologists revealed that nearly 50% of pediatric programs prefer CRRT for acute renal replacement therapy in critically ill children and that 95% of pediatric nephrologists had experience with CRRT.¹ Although the use of CRRT is becoming more prevalent in critically ill children, few published data exist that examine the outcome of children who receive CRRT.

To date, all published pediatric CRRT data have been retrospective reviews of a single-center experience that compare and/or mix the acute renal replacement therapy modalities of continuous venovenous hemofiltration (CVVH/dialysis [D]), continuous arteriovenous hemofiltration (CAVH/D) and acute intermittent HD. A few studies have considered the effect of some clinical variable on outcome. Lane and colleagues² noted that mortality was greater in children after bone marrow transplant who had >10% fluid overload at the time of HD initiation. Smoyer et al³ evaluated a large number of pediatric patients (n = 98 including neonates) and found that mortality was higher when patients receiving CRRT required intravenous pressors or were treated with ultrafiltration alone not combined with D, but did not find a difference in outcome with CAVH versus CVVH. Maxvold and colleagues⁴ noted that children with ARF who received intermittent HD required fewer pressors and had better survival rates than children who received CRRT. In fact, the low survival seen in pediatric patients receiving CRRT has been attributed to a selection toward the use of CRRT in more critically ill children,⁴ but only 1 pediatric study actually used a severity of illness scoring system as part of the analysis. Zobel et al⁵ demonstrated that among children, who received CRRT, those with worse illness severity by the Pediatric Risk of Mortality (PRISM) score had an increased mortality rate. Children in this study received 1 or more modes of CRRT, yet the authors did not assess for an effect of modality on outcome.

No pediatric study has evaluated factors affecting outcome and controlled for severity of illness in children treated only with CVVH or CVVH/D, the most common CRRT modality used in critically ill children today. In this study, we examined the potential effect of clinical variables on the outcome of critically ill children with ARF treated only with CVVH during their acute illness. The aim of this study was to determine whether specific variables have a significant effect on patient outcome. PRISM scores were then reviewed to assess whether the observed effect was independent of severity of illness.

	METHODS

Top Abstract Methods Results Discussion References

Patient Population

The medical records of 21 patients who received CVVH with or without concurrent hemofiltration (D) added to the CVVH circuit (hereafter referred to as CVVH/D) in the Texas Children's Hospital pediatric intensive care unit (PICU), Houston, Texas from February 1996 through September 1998, were reviewed retrospectively. Each decision to initiate CVVH/D was made solely by the attending pediatric nephrologist on call for the Renal Service at Texas Children's Hospital. During the course of study, no set protocol was used to determine the modality of renal replacement therapy to be provided for a particular patient. The pediatric nephrologist did not know the patient PRISM score. In general, the practice during that time was to choose CVVH/D over HD for patients receiving >1 intravenous pressor for blood pressure support. Two patients who did not tolerate initiation of CVVH/D therapy as evidenced by profound hypotension and cardiopulmonary arrest in the first 2 hours after CVVH/D initiation were not included in the study. Both of these patients were receiving 3 pressors at CVVH initiation. Neither CVVH circuit used an AN-69 membrane.

Clinical Variables

The following pre-CVVH/D initiation data were analyzed: patient age, patient gender, primary disease leading to need for CVVH/D, comorbid diseases, reason for CVVH/D use, PICU admission weight (kg) and height (cm), fluid intake in liters (fluid in) from PICU admission to CVVH/D initiation, fluid output in liters (fluid out) from PICU admission to CVVH/D initiation, estimated glomerular filtration rate (GFR; by Schwartz formula) at CVVH initiation and urine output 24 hours before CVVH initiation (mL/kg/hour).

The percent fluid overload (%FO) at the time of CVVH/D initiation was calculated by the following formula: We used the %FO as a relevant marker for the time to initiation of CVVH/D (ie, early vs late initiation). Reporting the actual time between PICU admission and CVVH/D could lead to spurious conclusions because patients can potentially be stable for a few days in the PICU before developing renal failure and fluid overload. Because all patients in this study were intubated and mechanically ventilated, insensible losses were minimal and not factored into the %FO calculation.

The following post-CVVH/D initiation PICU data were obtained for each patient: fluid intake in liters (fluid in_CVVH), fluid output including ultrafiltration by CVVH/D in liters (fluid out_CVVH), maximum number of pressors used, pressors completely weaned (yes/no), mean airway pressure (P_aw) at CVVH/D initiation and termination, PICU length of stay (days), CVVH complications, and patient outcome (death or survival to PICU discharge).

The percent of pre-CVVH/D accumulated fluid removed by CVVH/D was calculated using the following formula:

Equipment

Choice of equipment was based on patient size and available sites for venous access. Either the BM-11 (Baxter Corporation, McGaw Park, IL) or the PRISMA (Cobe-Gambro Healthcare, Lakewood, CO) hemofiltration machines were used. Dual-lumen uncuffed catheters (Quinton Corporation, Bothell, WA or Medcomp Corporation, Harleysville, PA), in sizes ranging from 7 to 12.5 French, were used for intravenous access. Reconstituted whole blood (hematocrit: 35%) was used to prime circuits with an extracorporeal blood volume >10% of a patient's total blood volume. Blood flows ranged from 50 mL/minute to 200 mL/minute and were prescribed to not exceed a rate of 400 mL/minute/1.73 m² of patient body surface area.

For CVVH/D, a premixed lactate-buffered diafiltration fluid (Baxter Corporation) was used except in patients with liver failure who were unable to metabolize lactate to bicarbonate. A bicarbonate buffered D fluid (100 mg/dL of dextrose, 100 mEq/L of NaCl, 40 mEq/L of NaHCO₃, 3 mEq/L of KCl, and 1 mEq/L of MgSO₄) was prepared daily by the hospital pharmacy for patients with hepatic dysfunction.

Anticoagulation was achieved by adjusting a constant heparin infusion to maintain activated clotting time between 180 and 220 seconds. Heparin was withheld from patients with a bleeding disorder, an activated clotting time >180 resulting from underlying illness (eg, disseminated intravascular coagulation from sepsis), or evidence of intracranial bleeding.

PRISM Score Calculation

PRISM 2 scores, which had been recorded for patients during their stay in the PICU, were reviewed. A single study nurse calculated the PRISM 2 score for each patient at the time of PICU admission and CVVH initiation in accordance with the method established by Pollack and colleagues.⁶ The PRISM 2 score evaluates 14 clinical variables (systolic and diastolic blood pressure, heart and respiratory rate, partial pressure of arterial oxygen/forced inspiratory oxygen, partial pressure of carbon dioxide, Glasgow coma score, prothrombin time/partial thromboplastin time, pupillary reaction, total bilirubin, serum potassium and bicarbonate levels, and total serum calcium and glucose). Of note, no direct measure of renal function is factored into the PRISM score, although serum potassium and bicarbonate levels factor into the method.

Cost Analysis

Cost data were available from January 1997 onward for each patient's total PICU stay. The cost of providing CVVH/D was separated from the total PICU cost and reported as a percentage of the total PICU cost. Cost data accounted for nursing services, allied health professional services, and disposable medical equipment. Physician services and durable medical equipment costs were not factored into the cost analysis.

Statistical Methods

The effect of clinical variables on outcome (survivors vs nonsurvivors) was analyzed by Student's t test for independent samples. Patient survival curves were derived using Kaplan-Meier survival analysis. Multiple regression analysis was used to control for severity of illness with the PRISM score and evaluate the effect of fluid overload on outcome. All statistical analyses were performed using the Statistica Software Package, Version 5.0 (Statistica, Tulsa, OK). P value of <.05 was considered significant.

	RESULTS

Top Abstract Methods Results Discussion References

Pre-CVVH Data

Twenty-one patients received 22 courses of CVVH and/or CVVH/D totaling 3028 hours of circuit time over the course of study. Mean (± standard deviation) patient age was 8.8 ± 6.3 years (range: 0.5-18 years) and mean patient weight at PICU admission was 28.3 ± 20.8 kg (range: 6.7-81.4 kg). Eleven patients were <10 years old and 14 patients weighed <30 kg at the time of PICU admission. Mean length of PICU stay was 25.8 ± 18.6 days (range: 7-82 days). Mean duration of CVVH/D therapy was 5.8 ± 5.7 days (range: 2 hours to 19 days). Twelve patients received CVVH/D and 9 patients received only CVVH. The underlying illnesses leading to the need for CVVH/D included: sepsis (n = 11), cardiogenic shock (n = 4), hypovolemic acute tubular necrosis (n = 2), end-stage heart disease (n = 2), and end-stage lung disease, hepatic necrosis, viral pneumonia or small bowel obstruction (1 each).

Mean PRISM score at PICU admission was 13.1 ± 5.8 (range: 4-28) and at CVVH initiation was 15.4 ± 8.9 (range: 3-36). Mean %FO at CVVH initiation was 26.0% ± 19.8% (range: 0%-80%). Mean GFR at CVVH initiation was 20.5 ± 18.1 mL/minute/1.73 m² (range: 3-59.1 mL/minute/1.73 m²).

Survival Data

Nine of 21 patients (42.8%) survived. Five of 11 septic patients (45.5%) survived. If the 3 patients with end-stage organ failure (2 with end-stage heart disease and 1 with end-stage lung disease) are excluded from analysis, nine of the remaining 18 patients (50%) survived. Nine of 12 deaths (75%) occurred <25 days into the PICU course (Fig 1). Mean time to death in nonsurvivors was 20.7 ± 17.4 days (range: 7-52 days).

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Kaplan-Meier plot that demonstrates 75% of deaths occurred within 25 days of PICU admission. Plot represents data from all 21 patients in the study.

Clinical variable data for survivors and nonsurvivors are summarized in Table 1. The degree of fluid overload at CVVH/D initiation was significantly lower (P = .03) in survivors (16.4% ± 13.8%; range: 2%-44.8%) compared with nonsurvivors (34.0% ± 21.0%; range: 7.5%-80%). Mean patient weight, age, PRISM score at PICU admission, PRISM score at CVVH/D initiation, maximum number of pressors used, GFR, and change in P_aw did not differ significantly between survivors and nonsurvivors. Mean percentage of pre-CVVH/D accumulated fluid actually removed by CVVH/D did not differ significantly between survivors and nonsurvivors, although the median percentage of pre-CVVH/D accumulated fluid removed was higher for survivors (25.8%; range: 185%-167%) versus nonsurvivors (4.4%; range: 17.4%-86.5%).

Because the degree of fluid overload at CVVH initiation was the only clinical variable that had an effect on outcome, we used multiple regression analysis to adjust for severity of illness (PRISM at PICU admission) to determine whether worsening fluid overload was itself solely a marker of disease severity. A lesser degree of fluid overload was associated with improved outcome even when the data were controlled for severity of illness by PRISM score (P = .03).

Cost Analysis

Complete treatment cost data were available for 16 patients (8 survivors and 8 nonsurvivors). The mean total PICU cost was $124 496 per patient (range: $24 149-$405 199) and the mean cost of providing CVVH/D was $884 per patient (range: $148-$2215). The mean percentage cost of providing CVVH/D accounted for only 0.9% ± 0.5% (range: 0.1%-1.5%) of total PICU cost for patient PICU stay and did not differ between survivors and nonsurvivors.

	DISCUSSION

Top Abstract Methods Results Discussion References

The current study reports the first set of pediatric CRRT data that examined only CVVH/D and used a pediatric scoring system to control for severity of illness. To date, all other pediatric reports have combined data for at least 2 different methods of renal replacement therapy or have not taken patient severity of illness into account. CAVH was the first mode of CRRT used widely in children, because extracorporeal blood volume is minimal in a CAVH circuit. In the 1990s, CVVH circuits with sufficiently small extracorporeal volumes have been manufactured to make CVVH available to children of all sizes.^7,8 Although a previous pediatric study found no effect of modality (CAVH vs CVVH) on outcome,³ CVVH is now often the preferred method of CRRT for children, because circuit blood flow is regulated mechanically and does not depend on a patient's cardiac output.

Compared with previous pediatric studies,^3,4 we evaluated the potential effect of many more clinically relevant variables on survival in children who received CVVH/D. Our data support the finding of Lane and colleagues² that a worse degree of fluid overload at the time of HD initiation was associated with poorer outcome. Survivors in our study were significantly less fluid overloaded at CVVH/D initiation compared with nonsurvivors, whereas none of the other clinical indices examined were associated with differences in outcome in our study. Clearly, many critically ill patients require significant fluid resuscitation to maintain blood pressure and some may not be able to tolerate minimal fluid removal by CVVH. Although this degree of hemodynamic instability may itself portend a worse prognosis, a greater degree of fluid overload was still associated with worse outcome when the data were adjusted for severity of illness by PRISM score. It is possible that in some cases CVVH/D may be a prevention, rather than a treatment, for worsening degrees of fluid overload. For example, a patient with a bone marrow transplant might have ARF from a number of causes, including venoocclusive disease, cyclosporine toxicity, and/or bone marrow transplant nephropathy. Prevention of fluid overload and pulmonary edema is crucial for prevention of mortality in this patient population. Early initiation of CVVH to allow for sufficient blood product and nutrition administration, while preventing fluid overload may improve patient survival in this type of situation.

It was not the purpose of this study to validate the prognostic ability of the PRISM 2 score in children with ARF. We used the PRISM 2 score solely to adjust for severity of illness when evaluating variables observed to have an effect on patient outcome. The developers of the PRISM score correctly caution against applying severity of illness scoring systems to individual patients⁹a point underscored by the observation of Fargason and Langman¹⁰ who found that the PRISM score had limited ability to predict mortality in children who received intermittent D. A potential danger of applying severity of illness scoring systems to individual patients would be to withhold therapy from patients who are at greater than acceptable risk for death as predicted by PRISM, especially if in an effort to reduce cost. The benefit-to-cost ratio of any new therapy for critically ill patients needs to be closely scrutinized.¹¹ Unlike adults, however, the majority of children develop multisystem organ failure early in their PICU course.^12,13 Our data revealed that 75% of nonsurviving children with ARF who received CVVH/D died within the first 25 days of PICU stay. Our cost analysis agreed with a previous analysis that showed CVVH/D did not represent a significant cost addition to the care of critically ill children.¹⁴

Based on the observations of our study, we propose that earlier initiation of CVVH/D (eg, at 10% fluid overload vs 25% fluid overload) may prevent morbidity and potentially improve survival in some critically ill children with ARF. Net fluid removal may not be necessary or may even be harmful at early stages of disease in which patients have capillary leak. However, earlier initiation of CVVH/D could provide for a more fluid-balanced state instead of a fluid overloaded state and allow for administration of nutritional fluids and necessary blood products. The predilection for early multiorgan system failure and death in critically ill children with ARF and the potential for improved outcome with initiation of CVVH/D at lesser degrees of fluid overload are factors that argue for early and aggressive initiation of CVVH/D.

	ACKNOWLEDGMENTS

We thank M. Michelle Mariscaclo, MD, for providing insight with respect to the pattern of multiorgan dysfunction in children.

We thank Carolyn M. Smith, Senior Decision Support Analyst with the Texas Children's Hospital Support Service, for providing ICU and CVVH cost data.

	FOOTNOTES

Received for publication Jul 14, 2000; accepted Sep 28, 2000.

Reprint requests to (S.L.G.) Texas Children's Hospital, 6621 Fannin St, MC 3-2482, Houston, TX 77030. E-mail: stuartg{at}bcm.tmc.edu

	ABBREVIATIONS

CRRT, continuous renal replacement therapy; ARF, acute renal failure; HD, hemodialysis; CVVH, continuous venovenous hemofiltration; D, dialysis; CAVH, continuous arteriovenous hemofiltration; PRISM, Pediatric Risk of Mortality; PICU, pediatric intensive care unit; GFR, glomerular filtration rate; %FO, percent fluid overload; P_aw, mean airway pressure.

	REFERENCES

Top Abstract Methods Results Discussion References

1. Belsha CW, Kohaut EC, Warady BA Dialytic management of childhood acute renal failure: a survey of North American pediatric nephrologists. Pediatr Nephrol 1995; 9:361-363 [CrossRef][Medline]
2. Lane PH, Mauer SM, Blazar BR, Ramsay NK, Kashtan CE Outcome of dialysis for acute renal failure in pediatric bone marrow transplant patients. Bone Marrow Transplant 1994; 13:613-617 [Medline]
3. Smoyer WE, McAdams C, Kaplan BS, Sherbotie JR Determinants of survival in pediatric continuous hemofiltration. J Am Soc Nephrol 1995; 6:1401-1409 [Abstract]
4. Maxvold NJ, Smoyer WE, Gardner JJ, Bunchman TE Management of acute renal failure in the pediatric patient: hemofiltration versus hemodialysis. Am J Kidney Dis 1997; 30:S84-S88 [Medline]
5. Zobel G, Kuttnig M, Ring E, Grubbauer HM Clinical scoring systems in children with continuous extracorporeal renal support. Child Nephrol Urol 1990; 10:14-17 [Medline]
6. Pollack MM, Ruttimann UE, Getson PR Pediatric risk of mortality (PRISM) score. Crit Care Med 1988; 16:1110-1116 [Medline]
7. Bunchman TE, Maxvold NJ, Kershaw DB, Sedman AB, Custer JR Continuous venovenous hemodiafiltration in infants and children. Am J Kidney Dis 1995; 25:17-21 [Medline]
8. Bunchman TE, Donckerwolcke RA Continuous arterial-venous diahemofiltration and continuous veno-venous diahemofiltration in infants and children. Pediatr Nephrol 1994; 8:96-102 [CrossRef][Medline]
9. Ruttimann UE, Albert A, Pollack MM, Glass NL Dynamic assessment of severity of illness in pediatric intensive care. Crit Care Med 1986; 14:215-221 [Medline]
10. Fargason CA, Langman CB Limitations of the pediatric risk of mortality score in assessing children with acute renal failure. Pediatr Nephrol 1994; 7:703-707
11. Jakob SM, Frey FJ, Uehlinger DE Does continuous renal replacement therapy favourably influence the outcome of the patients? Nephrol Dial Transplant 1996; 11:1250-1255 [Abstract/Free Full Text]
12. Proulx F, Gauthier M, Nadeau D, Lacroix J, Farrell CA Timing and predictors of death in pediatric patients with multiple organ system failure. Crit Care Med 1994; 22:1025-1031 [Medline]
13. Proulx F, Fayon M, Farrell CA, Lacroix J, Gauthier M Epidemiology of sepsis and multiple organ dysfunction syndrome in children. Chest 1996; 109:1033-1037 [Abstract/Free Full Text]
14. Gilman CM, Coffel BE, Gunn SK Continuous venovenous hemofiltration: a cost-effective therapy for the pediatric patient. Am Neurol Nurs Assoc J. 1997; 24:337-341