Optimizing care of ventilated infants by improving weighing accuracy on incubator scales

1 Introduction

Delivering optimal neonatal intensive care requires accurate neonatal weights for correct medication, fluid and electrolyte administration. Medication errors are common in the neonatal intensive care unit (NICU) [1–6] and errors with the potential to cause harm are eight times more likely to occur in the NICU in comparison with adult hospital settings [6]. Studies have looked at the stage at which medication errors occur in NICU, varying from incorrect recording of weight, medical ordering, transcribing, and pharmacy dispensing through to nurse administration; however, none have looked at the use of estimated weights often used in NICU [7].

Fluid and electrolyte imbalance is commonly encountered in preterm infants requiring intensive care in the first week of life. Appropriate monitoring is essential, and perhaps the most useful parameter for monitoring fluid balance is the weight of the baby [5]. Frequent accurate measurements of body weight and of plasma sodium are the two most reliable indicators of changes in water balance [7]. Clinical trials have looked at fluid and sodium balance in preterm babies, with the ultimate aims of optimizing care and outcomes [8–16]. Review articles on managing neonatal fluid balance stress the need for accurate daily weighing [17–20]. The impact of the inaccurate calculation of medication, fluid and electrolytes will be greater in the extreme low birth weight and most sick infants, where estimated or inaccurate weights have the potential to lead to under or over administration of medication, fluid or electrolytes.

The accuracy of integral incubator scale weights in ventilated infants in clinical practice has not been established. Our objectives were to determine, the accuracy of weighing ventilated infants on integral scales and whether the accuracy could be improved by nullifying the force exerted by the ventilator tubing by the use of a ventilator tube compensator (VTC) device.

2 Methods

3 Patients

All clinically stable ventilated infants requiring weighing were eligible for the study. The study was conducted between June 2009 and February 2011 in the tertiary level NICU at Salford Royal NHS Foundation Trust, Salford, UK.

Infants on our NICU were supported by either the VIP Gold Ventilator or SensorMedics 3100 A Oscillator. The VIP Gold ventilator uses a standard infant ventilator circuit also used with other infant ventilators including the SLE 5000, whilst the ventilator tubing for the SensorMedics oscillator is specific for it. Patients recruited to the study ranged in gestation from 23 weeks to full term. Their birth weights ranged from 500 g to 3400 g (with the exception of one 23 week gestation infant weighing 383 g, who was weighed in combination with a 500 g standardized weight in order to bring its weight into the measurable Atom incubator weight range of ≥500 g). The majority of weights recorded, 82%, were on infants weighing <1500 g and 66% were babies weighing <1000 g.

Paired-data series of neonatal weights for 30 infants receiving HFOV and 30 infants receiving positive-pressure ventilation were obtained, on integral incubator scales, with and without the VTC in situ, for comparison against the infants’ electronic standalone scales weight. A trend series of weights, of up to 4 weights, were measured on 24 infants, undergoing HFOV and positive-pressure ventilation.

Prior to weighing the electronic and incubator scales were calibrated with standardized 1 kg and 500 g weights. Each infant had an individual VTC that was cleaned prior to each use with disinfectant wipes. Infants when disconnected from the ventilator for weighing on standalone scales were supported by T-piece ventilation that was discontinued momentarily for weighing. The infant was then immediately reconnected to the ventilator within the incubator and weighed with and without the VTC attached. The order of weighing on integral or standalone scales first, and with or without the VTC was according to nurse preference which varied related to clinical need (including need to change the incubator so the infant was weighed initially on the electronic scales whilst the incubator was being changed). Both weights were done in rapid succession. Weighing on integral incubator scales was done in accordance with the manufacturers’ instructions. Namely: central positioning of infant within the incubator, horizontally positioned mattress, free hanging weight cable and no overriding of equipment or objects over the edges of the mattress. When the infant was weighed on the integral scales, without the VTC attached, the ventilator tubing was held above the infant by one of the nursing staff, (whilst the other nurse lifted the infant), to ensure that it did not touch the mattress and an attempt to return the ventilator tubing to the same position, (after lifting the infant for weighing), was made. The T-connection port of the standard infant ventilator tubing or the dump valve of the SensorMedics oscillator tubing, was tightened within the iris of the side incubator port hole to ensure that the same length of ventilator tubing was within the incubator on each occasion that the infant was weighed. The staff that carried out weighing of the infant on one set of scales was blinded to the weights obtained on the alternative set of scales. Throughout the study two nurses (from a group of six senior experienced nurses), weighed the infants on the standalone scales, whilst the same medical staff member weighed the infants on the integral scales (UEK) to ensure that the incubator manufacturer’s instructions were strictly adhered to.

4 Statistical analysis

The data were analyzed using the Bland and Altman difference-plot method to explore the level of agreement between the weights obtained using standalone and integral incubator scales. The mean difference and the standard deviation (sd) of the differences between these two weights were calculated and the results displayed on two sets of graphs (Figs. 2 and 3) indicating the limits of±2sd. Biases were calculated to determine the precision of the estimated limits of agreement. In addition, the proportion of weights exceeding a considered clinically relevant percentage difference of 2% was determined for each approach (Table 1). A 2% difference was used since an infant would have an expected weight loss of up to 2% of its body weight daily in the first 4-5 days of life. A weight loss greater than 2% daily would be clinically relevant and would mandate consideration of extra fluid administration.

5 Results

We compared 48 pairs of infant incubator scale weight data sets, with and without VTC_os in situ, versus the true weight for 30 infants undergoing HFOV. The mean difference and (95% confidence limits for the bias) between the infants’ true weight and the incubator scale weight was 60.8 g (49.1 g to 72.5 g) without VTC_os. The corresponding mean difference and (95% confidence limits) with the VTC_os decreased to –2.8 g (–8.9 g to 3.3 g) (Fig. 2). The median recorded difference was 6.4% (range, –8.7% to 17.2%) for uncompensated weights, and 0.2% (–7.4% to 5.2%) for VTC_os assisted weights.

Analysis of 54 pairs of incubator scale weight data sets for 30 infants undergoing positive-pressure ventilation, with and without VTC_pr, yielded a mean difference of 41.0 g (32.1 g to 50.0 g) for uncompensated weights; decreasing to –5.1 g (–9.3 g to –0.8 g) with VTC_pr (Fig. 3). The median recorded inaccuracy was 4.2% (range, –10.3% to 13.6%) for uncompensated weights, and –0.5% (–5.8% to 3.3%) for VTC_pr assisted weights.

Ventilated infant weights, (that had detachable weighable appendages) agreed to within less than 2% of their true weight on integral incubator scales when used in conjunction with the appropriate VTC device. If infants had umbilical or long lines in situ, the median inaccuracy increased slightly from the true weight to 2.4% in infants undergoing positive-pressure ventilation (range, 0.1 to 5.7%) and 1.5% in infants on HFOV, (range, 0.1 to 7.4%), data not shown.

6 Discussion

This is the first study using comparative weighing to report that weighing ventilated infants on integral incubator scales is substantially inaccurate for both individual and trend weights. The large discrepancy found between the weights measured using integral incubator scales and the true weight could be substantively corrected by the use of a VTC device to counterbalance the force exerted by the ventilator tubes. The most striking inaccuracy was seen in infants supported by a SensorMedics oscillator with a mean difference of 60.8 g from the true weight when weighed on integral scales, which decreased to a mean difference of –2.8 g by the application of a VTC_os.

The majority (66%) of ventilated infants studied were extreme low birth weight infants reflective of a typical tertiary level NICU. These are the very group of infants where an over- or underestimated weight can have important clinical implications for treatment. In the clinical management of a SensorMedics oscillated 500 g infant, a mean difference of 60.8 g would be 12% different from its true weight potentially resulting in substantial over or under dosing of fluid, medication or electrolytes.

Ventilated infants with central lines had an increased median weight inaccuracy of up to 2.4% recorded, when weighed on integral scales in conjunction with a VTC device. Invasive lines may be elevated when weighing infants on standalone scales or placed on the incubator mattress when weighed on integral scales, in an attempt to exclude their weight from the calculation. They may still have an effect upon the true weight, since it is difficult to legislate for their overall consequence in either weighing method.

Engstrom conducted the only other reported study addressing the accuracy of in-bed weighing procedures for intensive care infants, 61% of whom were ventilated [22]. A major limitation of the study was that it only studied the intra-and inter-examiner reliability of the in-bed weighing procedure and did not compare integral scale weights with true standalone scale weights. It was noted that weights on sick infants attached to equipment, may be less reliable, due to the unknown effect that equipment may have and the difficulty found holding stable the equipment attached to the infant during weighing was acknowledged. Intra- and inter-examiner reliability was significantly affected by the presence of several pieces of monitoring equipment and holding equipment above the infant resulted in variances of up to 100 g, depending upon the equipment and how it was held.

An online survey (unpublished data) that we undertook in 2011 of tertiary level NICU consultant neonatologists in England showed that the majority consider trend weights to be accurate; in that if the ventilator tubing and attachments remain the same on a daily basis, then any daily weight change will be due to a change in the infant’s weight. The incubator manufacturers stipulate that, ventilator tubing, and attachments, should be placed so that they return to the same relative position when the infant has been lifted off and repositioned on the mattress. However, we demonstrated that this was not feasible either in simulated patients or in clinical practice and that trend weights were inaccurate. 24 ventilated infants had trend weights taken, including a 740 g infant undergoing HFOV who had 4 weight data sets recorded. The integral scale weights showed an inaccuracy of 4.9% to 8.1%, however use of VTC_OS reduced this error to less than 1% of the true weight, nonetheless when the infant had an umbilical venous catheter in situ the weight was overestimated by 3.3%.

The study is limited in that only ventilated intensive care infants that were stable enough to disconnect from ventilation could be recruited, however there is no reason to suspect that the findings would be different in less stable infants. Another potential limitation is that each weight recording on the standalone scales was performed by two nurses from a group of six senior experienced nurses, rather than the same nurse, however this does reflect clinical practice. There were no adverse events, the procedure of attaching the VTC takes a matter of seconds and was well tolerated, despite two infants being transiently disconnected from the ventilator whilst the VTC was being connected.

Two constant force spring devices, VTC_os and VTC_pr, suspended perpendicular to the point of attachment to the endotracheal tube were designed, to counterbalance the force exerted by the ventilator tubing. The method of attachment was secure, however, the parents of two infants required additional reassurance that the device was safe prior to consenting.

In conclusion, our study demonstrated that integral incubator scales are substantially inaccurate for weighing ventilated infants and in some cases could result in suboptimal treatment being given to infants. A median inaccuracy from the true weight of 6.4% and 4.2% was shown in infants undergoing SensorMedics HFOV and positive-pressure ventilation respectively. Reliance on trend weights obtained from integral incubator scales for the purposes of calculating medication, fluids and electrolytes is of concern, particularly in the smallest and sickest of infants, due to the errors in dose this may introduce. Our findings show that a simple, safe, detachable, VTC mechanical device, used in conjunction with integral scales, can substantively improve the accuracy of weighing ventilated infants to better than 2% of the true weight of the infant for a large majority of infants, obviating the need to disconnect from the ventilator with the attendant risks of destabilization.

Conflict of interest

I own a patent on the medical device that facilitates the more accurate weighing of ventilated neonates described in this report. Whilst I own the intellectual property rights, Salford Royal Hospital would be the financial beneficiary if the medical device were to be developed. The 1% that I would receive would be assigned to further research. Dr Taylor declares no potential conflict of interest. This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.