Pediatr Nephrol (2005) 20:636–643
DOI 10.1007/s00467-004-1785-5
O R I G I N A L A R T I C L E
Nicholas L. Everdell · Malcolm G. Coulthard ·
Jean Crosier · Michael J. Keir
Received: 27 April 2004 / Revised: 13 October 2004 / Accepted: 19 November 2004 / Published online: 17 March 2005
IPNA 2005
Abstract Babies weighing under 6 kg are difficult to dialyse, especially those as small as 1 kg. Peritoneal dialysis is easier than haemodialysis, but is not always possible, and clears molecules less efficiently. Two factors complicate haemodialysis. First, extracorporeal circuits are large relative to a baby’s blood volume, necessitating priming with fresh or modified blood. Second, blood flow from infants’ access vessels is disproportionately low (Poiseuille’s law), causing inadequate dialysis, or clotting within the circuit.
These problems are minimised by using single lumen access, a very small circuit, and a reservoir syringe to separate the sampling and dialyser blood flow rates. Its manual operation is tedious, so we developed a computercontrolled, pressure-monitored machine to run it, including adjusting the blood withdrawal rate from poorly sampling lines. We have dialysed four babies weighing
0.8–3.4 kg, with renal failure or metabolic disorders. The circuits did not require priming. Clearances of creatinine, urea, potassium, phosphate and ammonia were mean (SD)
0.54 (0.22) ml/min using one dialyser, and 0.98 (0.22) ml/ min using two in parallel. Ammonia clearance in a 2.4 kg
N. L. Everdell ( ) )
Department of Medical Physics and Bioengineering,
Malet Place, Engineering Building, University College London,
London, WC1E 6BT, UK e-mail: everdell@medphys.ucl.ac.uk
M. G. Coulthard
Department of Paediatric Nephrology,
Royal Victoria Infirmary,
Newcastle, NE1 4LP, UK
J. Crosier
Department of Paediatric Nephrology,
Royal Victoria Infirmary,
Newcastle, NE1 4LP, UK
M. J. Keir
Department of Medical Physics,
Royal Victoria Infirmary,
Newcastle, NE1 4LP, UK baby had a 9 h half-life. Ultrafiltration up to 45 ml/h was achieved easily. This device provided infants with immediate, effective and convenient haemodialysis, typically delivered for prolonged periods.
Keywords Dialysis · Ultrafiltration · Preterm baby ·
Renal failure · Renal clearance · Renal replacement
Though improvements in neonatal management have increased the survival of preterm babies, acute renal failure affects up to 8% of babies requiring intensive care, and its treatment remains problematic. Most cases are secondary to other medical conditions rather than inherent kidney disease [1]. If renal replacement becomes necessary, peritoneal dialysis is preferred [1], but sometimes this is not feasible. Some infants require haemodialysis because they have a congenital defect of the abdominal wall, or necrotising enterocolitis or abdominal surgery. Other neonates with inherited metabolic disorders require haemodialysis to augment the removal of toxic chemicals
(such as ammonia) that are generated more quickly than the kidney or peritoneal dialysis can clear them.
The minimum haemodialysis circuit volume [2] (typically up to 49 ml) is large in proportion to a baby’s blood volume of approximately 85 ml/kg. This makes it preferable to prime the circuit with blood for larger infants, and essential for smaller ones, to prevent a dramatic dilution of the baby’s blood with saline. However, stored blood is unsuitable for priming, having a high potassium concentration, an excess of citrate which chelates calcium and magnesium ions, and a low pH. This necessitates another approach. The circuit blood may be dialysed before use [3], or the red cells washed, resuspended in albumin, and chemically modified, or heparinised blood may be drawn from walk-in donors, tested and used fresh
[1].
Also, Poiseuille’s law dictates that the baby’s access blood flow rates are disproportionately low compared to

637
Fig. 1 Diagram of the dialysis circuit their dialysis requirement. This factor alone makes it about four times harder to dialyse a term baby than an adult, and around seven times harder to dialyse a 1 kg neonate [4]. Resultant poor flows may prevent effective dialysis, and lead to clotting within the dialyser. Though haemodialysis has been reported in infants of around 2 kg, it is certainly technically difficult [5], and has been considered impossible in smaller preterm infants [6].
Because of these problems we developed a novel manual dialysis system for tiny infants which employed a much smaller circuit, and separated the blood flow requirements of the sampling and dialysis phases of the cycle [7]. Single lumen access was used to improve access efficiency [4], and the circuit consisted merely of two syringes, a small dialyser, and a series of three-way taps.
The first syringe had a reservoir function which allowed blood to be sampled slowly from the child’s line, but to be passed quickly across the dialysis membrane. Ultrafiltration was achieved by generating a positive transmembrane pressure by fixing an elastic band on the distal syringe to resist opening. Though effective, this manual system was extremely tedious to perform, continuously requiring an extra nurse. We therefore developed an automated device which would drive the system, and allow us to provide practical haemodialysis and ultrafiltration for babies continuously over prolonged periods of time.
m
2 surface area hollow-fibre dialyser (Miniflow 10, Hospal Ltd,
CV2 1PB, UK), and connecting lines. Standard clinical polypropylene syringes with neoprene bungs stiffened after prolonged use with blood, so we used 25 ml gas-tight syringes with glass barrels and Teflon-tipped bungs (Hamilton Ltd, LA5 9EA, UK).
Unlike conventional circuits, the haemodialysis blood flow is independent of a continuous supply from the child. Instead, blood is first aspirated from the patient at a rate determined by the vascular access, and then passed repeatedly through the dialyser to allow dialysis and ultrafiltration, before being returned to the baby. This cycle is then repeated.
The circuit extracorporeal blood volume is small, at the syringe stroke volume plus 5 ml (3.5 ml within the dialyser). If the stroke volume was set at 3 ml for a 1-kg baby, the maximum extracorporeal circuit would be 8 ml, or 9.4% of its blood volume (assuming 85 ml/kg). For larger babies, any particular sample volume would provide a proportionately greater stroke volume. For example, sampling 20 ml from a 3 kg baby would give an extracorporeal circuit volume of 7.8% of the blood volume, and a 6.7 ml/kg stroke volume. Because the syringe volume is limited to 20 ml, the stroke volume/kg and percentage extracorporeal volume would fall for larger babies, reaching 3.3 ml/kg and 4.9% in a 6 kg baby. Two dialysers can be used in parallel for larger infants, increasing the circuit volume from 5 to 13 ml. Babies of 4–6 kg with 20 ml sampling volumes would then have extracorporeal circuits between
6.5% and 9.7% of their blood volumes.
Overview of the circuit
Figures 1 and 2 show a diagram and a photograph of the circuit.
Apart from the syringes, it is constructed from standard sterile disposable clinical equipment with screw locks, including threeway taps (Connecta Plus 3, Beckton Dickinson, SE-251 06, Sweden), a pressure transducer (Transpac IV, Abbott, Ireland), a 0.042
The operating cycle
The flow within the circuit is governed by the three-way taps. The circuit is initially primed with heparinised saline, and cycled to simulate clinical use, drawing fluid from a reservoir bag and returning it to waste, through tap 1, and a Y-connector with one-way valves. To commence dialysis, tap 1 is turned manually to disconnect this fluid system, and connect the baby’s blood access line instead. Taps 3 and 5 remain permanently open in all directions, and taps 2 and 4 are turned by the machine. During blood withdrawal, tap 2 connects the baby to syringe A and obstructs the flow of intravenous fluid infusion, and tap 4 disconnects syringe A from the dialyser.
Next, the blood is passed back and forth between the syringes, through the dialyser. Tap 2 turns to disconnect the baby’s access line from the circuit, and allows the intravenous infusion to resume.

638
Fig. 2 Photograph of a circuit for a larger baby with two dialysers in parallel. Note that extra syringes were used with this particular circuit to enable easy removal of clots, and that extra taps were used compared to Fig. 1 for ease of connection of infusion and heparin lines, etc.
An initial bolus of fluid flushes the line because the infusion pump continues to operate against the closed tap. Tap 4 also turns to connect the circuit between the two syringes. Syringe A’s plunger is then depressed and syringe B’s is simultaneously withdrawn, until the blood has been transferred between syringes through the dialyser. Their directions are then reversed, and so on for a predetermined time. If ultrafiltration is required, the receiving syringe opens less than the volume being emptied from the driving syringe.
This generates a rise in transmembrane pressure which drives plasma water out of the circuit. For chemical clearance, the flow of a commercially available bicarbonate-based dialysis fluid is controlled through the dialyser cylinder using an inflow intravenous infusion pump, and passive outflow. Finally, taps 2 and 4 return to their sampling positions, and the processed blood is returned to the baby, and another volume aspirated.
During treatment, clotting within the circuit is prevented by a slow heparin infusion into tap 5. This provides relatively higher concentrations within the dialyser than in the baby’s circulation because some effect is lost with passage through the dialyser. The infusion is adjusted to maintain the glass-activated clotting times, measured on 0.1 ml blood from the patient line, between 180 and
220 s.
Recirculation is inevitable in single lumen circuits because dialysed blood is left in the sampling line at the end of each cycle.
For this circuit, the impact of this is best expressed by calculating the effective cycle volume for any particular aspirated volume, and will depend critically on the volume of the connecting tubing and intravenous line lumens. These totalled about 1.0 ml (0.4 ml in taps,
0.4 ml in connecting line, 0.2 ml in venous line) for our two smaller infants, and 1.4 ml in the two larger babies who had longer connecting lines. In our clinical practice, the 5 ml stroke volume used for the smallest baby was therefore reduced by 20% to an effective cycle of 4 ml, and the 20 ml aspirate for the largest baby was effectively cut by 7% to 18.6 ml. Recirculation due to aspirating freshly returned blood from within the vein is likely to be minimal because the returning and sampling phases are sequential, unlike conventional dual lumen dialysis where they occur simultaneously.
To prevent babies becoming clinically unstable during periods of fluid removal or dialysis, treatment is performed either continuously throughout the 24 h, or at least slowly over prolonged periods, and not in short treatment runs. At the end of a treatment period, the machine is stopped with both syringes empty, and the blood remaining in the circuit returned to the baby. The taps are positioned to allow heparinised saline to flush the blood, under visual control, from the distal end of the circuit to the patient line.
The circuit can then be disposed of, or preserved by cycling with heparinised saline until flushed completely clear, and used later (for example, the next day). This saved the time of assembling new circuits (which are not yet available as pre-packed units) and was not associated with any bacterial contamination, judged by negative blood cultures.
This prototype device has neither air nor blood leak detectors built in. We have relied upon the fact that air leaks would be likely to result in the accumulation of bubbles within the clear glass operating syringes to monitor these, and have observed the clarity of the effluent dialysate to diagnose blood leaks.
Description of system
The syringe plungers are driven by 1 W stepper motors, similar to those used in a clinical syringe pump, but allowing the plunger to be driven considerably faster. We initially used powerful motors to overcome the stiffness that polypropylene/neoprene syringes developed with use, but this was an unsatisfactory solution because ultimately the bungs lock within the barrel, and the force of the motor detaches them from the plungers. Glass/Teflon gas-tight syringes were slightly stiffer than standard syringes initially, but were unaffected by use.
The torque required to turn taps 2 and 4 between two positions,
90 apart, during each operating cycle is relatively high because they also stiffen as traces of blood seep into their bearings. This problem was also overcome by the use of more powerful electric motors.
The system is controlled by a desktop PC. The software controls the desired volumes and flow rates for blood sampling, dialysis and ultrafiltration. It monitors the blood circuit pressures (user interface, Fig. 3). It also displays cumulative information, such as the total treatment time and ultrafiltrate volume.
The system is classified as class I equipment by IEC 60601 (BSI
1990) [8], and is powered through an isolation transformer to significantly reduce the risks of electric shock. Much of the software and hardware is concerned with providing safe single fault conditions, and a single fault analysis was performed in accordance with
IEC 60601 part 1, and part 2 which deals specifically with the safety requirements of dialysis machines (BSI 1990; BSI 1998) [8,
9]. Each time the machine is switched on, an initialisation procedure tests the entire system’s functions.

639
Fig. 3 Graphical user interface
The circuit pressure is continuously monitored using a disposable sterile transducer. The circuit pressure and the phase of the dialysis cycle are constantly displayed on the monitor. If the pressure limits are exceeded, an audible and visible alarm is triggered. If necessary the operating cycle is halted.
The transducer position enables all the phases of the operating cycle to be monitored. During blood withdrawal it senses the negative pressure generated by the resistance to flow. With adequate access, this drop is minimal, but obstructions produce a marked fall. As blood is returned, obstructions generate a pressure rise. During dialysis, the transducer monitors the patency and resistance to flow through the lines and hollow fibres, registering a pressure rise as blood is passed from syringe A to B, and a fall as it is returned. Gradual changes in pressures with time could indicate problems such as clot accumulation within the dialyser.
The computer adjusts the blood sampling rate if babies have poorly flowing lines. Typically, blood withdrawal at 20 ml/min produces minimal negative pressure, but may generate a marked fall and lead to a complete obstruction. This is probably because rapid sampling from a vein or the right atrium virtually empties it in a small infant, occluding the line by sucking its sampling hole against the vessel wall. This is a familiar experience when taking blood with a syringe, when clinicians release the suction, and then sample more slowly. In this computer controlled algorithm, the syringe stops if the negative pressure reaches a trigger value, briefly reverses, and restarts 2 ml/min slower. This sequence repeats until the rate allows sampling without reaching the trigger pressure.
After 15 min, the sampling rate increases by 2 ml/min each cycle in case the sampling conditions have improved.
Ultrafiltration, dialysis and haemolysis testing
The system was tested in vitro on fresh heparinised blood. Using syringe stroke volumes of 3–20 ml, eighteen ultrafiltration rates up to 50 ml/h were each tested over 1 h, and dialysis clearances were measured to determine optimal combination of working blood volume and dialysate flow rates. To increase measurement precision, urea and creatinine were added to the blood to produce concentrations seen in renal failure. Clearances were measured by the removal of chemicals from blood, and their accumulation in the effluent dialysate.
Red blood cells may be haemolysed by mechanical forces during dialysis. Fresh heparinised blood was gently agitated in a reservoir by a mechanical stirrer, dialysed, and free plasma haemoglobin concentrations measured hourly for 12 h. Simultaneous controls were measured from blood that was agitated, but not dialysed.
Clinical use of the system
We have used this device only for babies where conventional options were clearly inferior or non-existent (Table 1). Parents were always fully aware that this device was a prototype. The Research
Ethics Committee did not consider the individual clinical decision to use this device to be within their area of jurisdiction. The Hospital Trust deemed treatment appropriate if the clinicians considered it to be the child’s best option.
Machine reliability
The device was run for frequent prolonged periods during development, and clinically almost continuously for
2 months, and had no failures of the computer control, the machine, or any circuit components. We had no air or blood leaks. Some awkward computer programme control sequences need modification.
In vitro testing
Ultrafiltration
This was consistently achieved to within €4.5% of the prescribed rate, which compares well with other pub-

640
Table 1 Details of four babies dialysed using the automated syringe driven device
Case number, sex, gestational age, weight and age at dialysis
Sex wks kg days
Clinical details
1 F
2
3
4
F
M
F
32
38
24
37
1.63
0.2
Lactic acidosis from a mitochondrial cytopathy, anticipated from family history and antenatal testing. Died of cardiomyopathy.
3.84
34 End stage renal failure from dysplasia, treated initially with peritoneal dialysis.
Also had a protein-losing enteropathy, and developed an ileal perforation, so
0.80
30 required her Tenckhoff to be removed. Planned to haemodialyse until abdomen recovered to allow PD again, but enteropathy shown to be due to a severe intestinal abnormality, so treatment withdrawn.
Very preterm baby who had had a perforated bowel from necrotising enterocolitis, and developed acute renal failure secondary to liver abscess. Died of overwhelming sepsis.
2.40
4 Hyperammonaemia due to methyl-malonic acidaemia, presenting after 4 days of feeding. By start of dialysis her ammonia levels were already 2,300 m mol/l and she was unconscious and fitting. Good metabolic clearances, but definite evidence of having sustained brain damage, so treatment stopped.
Time on dialysis days
0.8
58
1.0
1.2
Fig. 4 Mean plasma clearances of creatinine, urea and potassium, over a range of syringe stroke volumes and dialysate flow rates, measured in vitro on whole blood, and in baby 2 lished data [10, 11]. To avoid producing hyperviscosity, the ultrafiltration rate was always adjusted to avoid producing packed red cell volumes greater than 70%. Rates of 50 ml/h were achieved without a discernible increase in operating pressures.
Dialysis
Clearances of creatinine, urea and potassium were similar, and absolute values achieved using from 3 to 20 ml syringe stroke volumes ranged from 0.15 to 1.05 ml/min.
Assuming a clinical stroke volume of 5 ml/kg, the clearances equated to 0.21 to 0.36 ml/kg/min (Fig. 4).
Clearances increased by 17% as the dialysate flow rates were increased from 10 to 20 times the blood volume
(paired t -test, P =0.001), but increased little beyond that
(Fig. 4). In practice, there is little to gain by reducing the dialysate flow rate below 400 ml/h. At this rate, a 5-l bag of dialysate will last for 12 h, which is the stability limit advised for commercially available bicarbonate-based solutions.
Clearances followed single exponential kinetics, dekt scribed by the equation C t
= C
0
(1– e ), where C
0 and
C t are the concentrations in the plasma at the start and time t , and k is the clearance. This was used to predict how poor blood sampling would affect clearances. Prolonged cycles become relatively inefficient as the concentration gradient falls, but in short cycles a greater proportion of the time is taken up with returning and sampling blood. Figure 5 shows examples of this. In practice, we therefore used 4–5 min dialysis times.

641
Fig. 5a–c Calculated clearances (ml/min) using a 10 ml syringe stroke volume for various values of k assuming a sampling speed from the patient of a 20, b 10, and c 5 ml/min
Haemolysis
The free haemoglobin concentrations remained around
50 mg/l in the undialysed control blood, but increased to
920 mg/l during 12 h dialysis. With a haemoglobin of
14 g/dl, this would represent destruction of < 1% of the cells, and is therefore an acceptable complication of an essential clinical procedure.
Clinical use
Four babies were dialysed continuously (Table 1), two newborns with inborn errors of metabolism, an extremely small infant who was unable to be peritoneally dialysed and too small to be conventionally haemodialysed, and a larger baby with end stage renal failure and peritoneal dialysis failure who was predicted to need prolonged haemodialysis. She was continuously dialysed whilst clinically very unstable for 3 weeks, and for 12 h daily thereafter.
Fig. 6 Clearances achieved in four babies dialysed with the syringe driven device pressure alarms even after the machine reduced the sampling rate to 2 ml/min. This led us to successfully change access to a 1 mm internal diameter central line.
Dialysis and ultrafiltration
Access and sampling
Baby 1 was initially dialysed through a 6 French (1.9 mm external diameter) umbilical venous line which sampled only intermittently, and then through a 5 French (1.6 mm external diameter) umbilical arterial catheter which gave excellent flows. Two babies had sialastic, surgically placed, 1 mm internal diameter superior vena cava lines.
In one of these, the standard sampling rate of 20 ml/min produced negative pressures of more than 200 mmHg, which initiated the automatic slowing process, and subsequent attempted increase. The access had spontaneously improved, and the standard rate was subsequently achieved. One baby had a percutaneous long line of
0.6 mm internal diameter initially which caused negative
The clearances obtained clinically were similar to those achieved in vitro. The three smaller babies treated with one dialyser had mean (SD) clearances for creatinine, urea, phosphate, potassium, and in one case ammonia, of
0.54 (0.22) ml/min. Baby 2 was nearly 4 kg and had two dialysers in parallel which gave clearances of 0.98
(0.22) ml/min (Fig. 6). She maintained stable electrolytes, a mean creatinine concentration of 220 m mol/l, and a mean urea of 13 mmol/l over several weeks. She required continuous phosphate supplements to compensate for high clearances. Varying the dialysate flow rate conferred little extra benefit once its flow rate in ml/h was 10 times greater than the sampling volume of the syringe (Fig. 4).
The machine required little nursing input, apart from

642
Fig. 7 Clearance of ammonia from the plasma of case 4 using syringe driven automated haemodialysis, plotted a on a linear, and b on a log scale.
Clearance half-life was 9 h.
Shaded areas show the period of dialysis
Fig. 8 Theoretical clearances produced by dialysis with the syringe driven device, compared a to the predicted normal
GFR [15] expressed in ml/min, and b expressed as a percentage of normal measuring clotting times and adjusting the heparin infusion. Baby 2 was mostly dialysed for 12 h during the day, and the circuit repeatedly cleaned and reused. Despite the theoretical possibility that the babies’ blood would tend to cool towards room temperature whilst in the extracorporeal circuit, none had a discernible fall in central body temperature, probably because the rate of blood removal from the babies was relatively slow, and perhaps because they were mainly nursed under overhead heaters.
Ammonia was cleared efficiently from baby 4 using one dialyser. The plasma concentration quadrupled in the
14 h before dialysis, and with treatment it fell with a halflife of 9 h (Fig. 7). Unfortunately, the pre-treatment concentration had already caused brain damage.
Babies 1 and 4 were not in renal failure, so no attempt was made to ultrafilter them. Babies 2 and 3 were ultrafiltered without any rise in dialysis working pressure, baby 2 for many days. It was easy to precisely remove fluid to maintain clinical balance regardless of her fluid intake, producing rates of up to 45 ml/h.
We have shown that this previously described manual dialysis system can be safely and effectively automated, and used for prolonged periods in babies between 800 g and 4 kg. The constant extra nursing required for the manual technique made it exceptionally tedious to perform for prolonged periods without error. By contrast, this device requires minimal nursing input, even when run
24 h/day. This blood circuit design is unique, enabling sampling from the baby to be at much slower rates than the flow needed through the dialyser. Other circuit designs with a superficial resemblance differ fundamentally
[12, 13, 14].
The babies we treated did not have satisfactory alternative treatment options. The two with inborn errors of metabolism would have needed blood-primed circuits for conventional haemodialysis, which generates significant delay since either walk-in donors or donor blood modification is needed. Our system meant no delay. Unfortunately, both of our babies died, one of her mitochondrial cardiomyopathy, and the other because her pre-treatment blood ammonia had already reached damaging levels. The clearances achieved were similar to those previously described for conventional haemodialysis [15].

643
The two babies with renal failure could not be peritoneally dialysed because of recent laparotomies for intestinal perforations. Baby 3 was too small to allow conventional haemodialysis to be considered, and died of overwhelming sepsis. We used our device instead of a standard circuit for baby 2 because it required considerably less nursing time, and avoided the need to prime the lines. This is a major advantage in prolonged and therefore repeated use. We now consider this device to be the treatment of choice for haemodialysing infants up to 6 kg.
Both biochemical and fluid volume control were easily maintained extremely stable. It is unfortunate that she also had an intrinsic abdominal pathology which led to withdrawal of treatment.
The device achieved clearances of around 0.5 ml/min with one dialyser, and 1 ml/min with two. To avoid blood priming, and minimise the extracorporeal circuit volume, we anticipate using one dialyser in babies under 3 kg, and two for larger infants. Although recirculation is inevitable with single lumen vascular access, this is not a major problem and can be readily deduced by subtracting the sample line dead space from the cycle volume. Recirculation is also likely to be relatively significant when using conventional double lumen circuits in small sick neonates, because relatively low flows in small vessels will allow greater re-aspiration of recently returned blood, even when the return lumen exits distally. Because the glomerular filtration rate (GFR) in babies does not increase linearly [16], the proportion of the GFR achieved by this device varies with body size. Its clearances equal the GFR in babies of 1 kg, falling to about 15% of this by
3 kg (Fig. 8), and then range from 30% to 10% using two dialysers in babies weighing from 3 to 6 kg. Clearances produced by peritoneal dialysis vary with the molecule, but are typically about 10%.
This device is a prototype. We believe its simplicity, ease of use, safety and reliability of dialysis and ultrafiltration, and the fact that it does not require blood priming, make it the method of choice for treating children under
6 kg with inborn errors of metabolism, and renal failure if peritoneal dialysis is not feasible. We think it opens up the possibility of haemodialysing very tiny preterm babies, even those below 1 kg. We are therefore currently producing a second prototype with a more clinician-friendly interface, and to conform to commercial and internationally recognised standards, including having reliable automated air and blood leak detectors.
1. Coulthard MG, Vernon B (1995) Managing acute renal failure in very low birth weight infants. Arch Dis Child Fetal Neonatal
Ed 73:F187–192
2. Gitomer JJ, Khan AM, Ferris ME (2001) Treatment of severe theophylline toxicity with hemodialysis in a preterm neonate.
Pediatr Nephrol 16:784–786
3. Pasko DA, Mottes TA, Mueller BA (2003) Pre dialysis of blood prime in continuous hemodialysis normalizes pH and electrolytes. Pediatr Nephrol 18:1177–1183
4. Coulthard MG, Sharp J (2001) Haemodialysing infants: theoretical limitations, and single versus double lumen lines. Pediatr
Nephrol 16:332–334
5. Bock GH, Campos A, Thompson T, Maher SM, Kjellstrand CM
(1981) Haemodialysis in the premature infant. Am J Dis Child
135:178–180
6. Coulthard MG (1989) Renal function. In: Harvey D, Cooke R,
Levitt G (eds) The baby under 1000 g, 1st edn. John Wright,
London, pp 211–225
7. Coulthard MG, Sharp J (1995) Haemodialysis and ultrafiltration in babies weighing under 1000 g. Arch Dis Child Fetal
Neonatal Ed 73:F162–165
8. Medical Electrical Equipment (1990) Part 1: General requirements for safety. IEC 60601-1. British Standards Institute,
London
9. Medical electrical equipment (1998) Part 2: Particular requirements for the safety of haemodialysis, haemodiafiltration and haemofiltration equipment. IEC 60601-2-16. British Standards Institute, London
10. Roberts M, Winney RJ (1992) Errors in fluid balance with pump control of continuous hemodialysis. Int J Artif Organs
15:99–102
11. Jenkins R, Harrison H, Chen B, Arnold D, Funk J (1992)
Accuracy of intravenous infusion pumps in continuous renal replacement therapies. ASAIO J 38:808–810
12. De Virgiliis GM, Vanin M, Buoncristiani U (1985) A new single needle dialysis system. Trans Am Soc Artif Int Organs
31:116–118
13. De Wachter D, Verdonck P, Verhoeven R, Hombrouckx R
(1993) Comparison of a new and a standard single-needle dialysis system using a mathematical model. Artif Organs
17:328–338
14. Hombrouckx R, Bogaert AM, Leroy F, Beelen R, De Vos JY,
Van Overmeeren G, et al. (1989) Limitations of short dialysis are the indications for ultrashort daily auto dialysis. Trans Am
Soc Artif Organs 35:503–505
15. Schaefer F, Straube E, Oh J, Mehls O, Mayatepek E (1999)
Dialysis in neonates with inborn errors of metabolism. Nephrol
Dialysis Transplant 14:910–918
16. Coulthard MG (1985) Maturation of glomerular filtration in preterm and mature babies. Early Hum Dev 11:281–292