Heather Lambert, Shaun Hiu, Malcolm Coulthard, John N S Matthews, Ruth Wood, Jean Crosier, Rachel Agbeko, Thomas Brick, Heather Duncan, David Grant, Quen Mok, Andrew Gustaf Nyman, John Pappachan, Paul Wellman, Chris Boucher, Joe Bulmer, Denise Chisholm, Kirsten Cromie, Victoria Emmet, Richard Feltbower, Michael Grayling, Rebecca Harrison, Eva-Maria Holstein, Ciara A Kennedy, Elaine McColl, Kevin Morris, Lee Norman, Julie Office, Roger Parslow, Christine Pattinson, Shriya Sharma, Jonathan Smith, Alison Steel, Rachel Steel, Jayne Straker, Lamprini Vrana, Jenn Walker, Mike Whitaker, Jim Wightman, Nina Wilson, and Lucy Wirz
Abstract Background Critically unwell babies in intensive care units may develop acute renal failure. Options for renal replacement therapy are limited by their small size and available technology. Objectives To determine the clinical efficacy, outcomes and safety profile of the NIDUS® (a novel infant haemodialysis device) for babies under 8 kg, compared with current renal replacement therapy. Design A clinical investigation using a non-blinded cluster stepped wedge design with paediatric intensive care units randomised to sequences. Setting Paediatric intensive care units in six UK hospitals. Participants Children under 8 kg who required renal replacement therapy for fluid overload or biochemical disturbance. Interventions Continuous renal replacement therapy was provided by the usual methods: peritoneal dialysis and continuous haemofiltration (during control periods) and by the NIDUS (during intervention periods), a novel device designed for babies with a smaller circuit and filter and volumetric control of ultrafiltration. Main outcome measures Primary outcome was precision of ultrafiltration compared with prescription; secondary outcomes included biochemical clearances, accuracy of reported ultrafiltration and mortality. Data sources Bedside study data collected by weighing bags of fluid entering and leaving the device were entered into the study database along with case descriptors. Some secondary outcome data was collected via the Paediatric Intensive Care Audit Network. Results Ninety-seven participants were recruited by study closure, 62 to control and 35 to intervention. The primary outcome was obtained from 62 control but only 21 intervention patients, largely because of technical difficulties using NIDUS. The analysis comparing the available primary outcomes showed that ultrafiltration with NIDUS was closer to that prescribed than with control: standard deviations controls 18.75, intervention 2.95 (ml/hour), adjusted ratio 0.13, 95% confidence interval (0.03 to 0.71); p = 0.018. The mean clearances for creatinine, urea and phosphate were lower on peritoneal dialysis than NIDUS, which were in turn lower than continuous veno-venous haemofiltration. The variability in the clearances was in the same order. Of the 62 control patients, 10 died (2/62 on peritoneal dialysis; 7/13 on continuous haemofiltration) before discharge from paediatric intensive care unit (16%), compared with 12 out of 35 (34%) in the NIDUS group: p = 0.04, 95% confidence interval for difference (0 to 36%). Harms No important adverse events occurred and the NIDUS has an acceptable safety profile compared with other renal replacement therapies in this critically ill population with multi-organ failure. Mortality was lowest for Peritoneal Dialysis, highest for continuous haemofiltration, with the NIDUS in-between. Only one serious adverse device event which was reported to the Medicines and Healthcare products Regulatory Agency. Conclusions NIDUS works effectively, delivering appropriate blood clearances and accurate, controllable fluid removal (ultrafiltration), indicating that it has an important place alongside other dialysis modalities for infant renal replacement therapy. Future work Findings from this study indicate some modifications are required to NIDUS to improve usability. Further studies on use of the NIDUS device in other populations of babies for example those with chronic renal failure, and long-term outcomes are required. Trial registration This trial is registered as ISRCTN 13787486. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Efficacy and Mechanism Evaluation Programme (NIHR award ref: 14/23/26) and is published in full in Efficacy and Mechanism Evaluation; Vol. 11, No. 1. See the NIHR Funding and Awards website for further award information. Plain language summary Why do this study? Some children in intensive care are so poorly that their kidneys do not work well, and they need help, called dialysis, to get rid of fluid and chemicals from their blood. For babies, we currently use peritoneal dialysis, where fluid is cycled in and out of the tummy, or adapted machines designed for bigger children (continuous veno-venous haemofiltration). A new machine, the NIDUS® (Allmed, www.allmedgroup.com), was developed specifically for babies weighing under 8 kg with much smaller tubing. NIDUS worked well when studied in Newcastle but needed testing elsewhere. What was the question? How well does NIDUS work compared to other dialysis methods? What are the problems? What did we do? The study was done in six paediatric intensive care units who used their usual dialysis methods (=control) in the first part of the study and then later swapped to using the NIDUS (=intervention). What did we find? We recruited 97 participants, 62 to control (49 peritoneal dialysis, 13 continuous veno-venous haemofiltration) and 35 intervention (NIDUS). We found NIDUS provided much better control of fluid removal. The CVVH machines were more efficient at blood cleaning than NIDUS, which was better than peritoneal dialysis. What does this mean? We learnt a lot about babies needing kidney support in paediatric intensive care units and that all methods have advantages and disadvantages. We showed that NIDUS could be very useful for some participants because it cleans blood effectively and gives accurate, controllable fluid removal. We have gathered important information to help us improve NIDUS to make it easier to use and run. Many parents responded to our questionnaire and most told us they felt it was acceptable to be approached about taking part in research despite the circumstances. This is very important for future research studies. We are very grateful to families for their generosity in becoming involved in this study. Scientific summary Background Critically unwell babies in paediatric intensive care units (PICUs) may develop acute renal failure and require management with renal replacement therapy. Although mortality and morbidity vary and are related to the underlying diagnosis, survival of babies in paediatric intensive care is worse for those with fluid overload. Babies requiring renal replacement treatment present specific therapeutic challenges because of their small size and the current technology available. Difficulties with vascular access and blood flows, fluid balance, loss of circuits, filter clotting and hypotensive episodes at initiation are all described in the literature. The need for new solutions and improved technology is well recognised. Continuous veno-venous haemofiltration (CVVH) machines in use in the UK at the time of this study are not approved for use in babies weighing 80%) were similar. The median (IQR) age in controls 10.5 (7, 38) days was similar to that in the intervention group 11 (7, 61) days; the range of age of participants was between 1 and 477 days (approximately 15 months). The median (IQR) weights 3.2 (2.9, 3.9) and 3.7 (3.1, 5.6) kg were similar between control and intervention. Availability of primary outcome The primary outcome was available on all 62 control patients but only 21 of the 35 intervention patients. This was due to a range of reasons including difficulties in obtaining the information needed to compute the UF rate (accurate timing and weighing data) and technical difficulties using the NIDUS: full details are in the report. Precision of UF Analysis comparing the 62 control patients with the 21 intervention patients with a primary outcome showed that UF with the NIDUS was closer to that prescribed than with control: standard deviations (SDs) controls 18.75, intervention 2.95 (ml/hour), adjusted ratio 0.13, 95% confidence interval (0.03 to 0.71); p = 0.018. For the NIDUS and CVVH devices, an important measure was to compare the difference between the actual fluid removal measured and that reported by the device. This had a mean closer to zero for the NIDUS than CVVH (means −0.44 vs. 11.6 ml/hour, respectively), with less variation in NIDUS than CVVH (SDs 3.2 vs. 28.4 ml/hour). Biochemical clearances The initial intention was to compare clearance rate on NIDUS with the control group. However, for these variables combining PD and CVVH in this way proved to be misleading because NIDUS clearances rates were intermediate between those of PD and CVVH. The clearance for creatinine on PD was smaller and less variable (mean 0.08, SD 0.03 ml/min/kg) than on the NIDUS (mean 0.46, SD 0.30 ml/min/kg), which was in turn smaller and less variable than for CVVH (mean 1.20, SD 0.72 ml/min/kg). The pattern was repeated for urea: PD (0.12, 0.06), NIDUS (0.48, 0.30) and CVVH (1.15, 0.67), all in ml/min/kg, and also for phosphate: PD (0.07, 0.04), NIDUS (0.44, 0.27) and CVVH (1.16, 0.71), all in ml/min/kg. All pairwise treatment comparisons of means and of SDs gave p