Management of arterial lines and blood sampling in intensive care: a threat to patient safety

Anaesthesia 2013, 68, 1114 1119 Original Article doi:10.1111/anae.12389 Management of arterial lines and blood sampling in intensive care: a threat to patient safety R. A. Leslie, 1 S. Gouldson, 2 N. Habib, 2 N. Harris, 2 H. Murray, 2 V. Wells 3 and T. M. Cook 4 1 Specialist Registrar, 2 Core Trainee, 3 Clinical Fellow, 4 Consultant, Department of Anaesthesia and Intensive Care Medicine, Royal United Hospital, Bath, UK Summary In 2008, the UK National Patient Safety Agency (NPSA) made recommendations for safe arterial line management. Following a patient safety incident in our intensive care unit (ICU), we surveyed current practice in arterial line management and determined whether these recommendations had been adopted. We contacted all 241 adult ICUs in the UK; 228 (94.6%) completed the survey. Some NPSA recommendations have been widely implemented use of sodium chloride 0.9% as flush fluid, two-person checking of fluids before use and their practice was consistent. Others have been incompletely implemented and many areas of practice (prescription of fluids, two-person checking at shift changes, use of opaque pressure bags, arterial sampling technique) were highly variable. More importantly, the use of the wrong fluid as an arterial flush was reported by 30% of respondents for ICU practice, and a further 30% for practice elsewhere in the hospital. Our survey provides evidence of continuing risk to patients.... Correspondence to: T. M. Cook Email: timcook007@googlemail.com Accepted: 11 July 2013 This article is accompanied by an Editorial by Smith, pp 1099 1103 of this issue. In critically ill patients, indwelling arterial cannulae are routinely inserted for continuous blood pressure monitoring and to obtain arterial blood samples for blood gas, glucose and electrolyte analysis. However, these lines are not without risks. In 2008, the UK National Patient Safety Agency (NPSA) published a Rapid Response Report after it identified 84 incidents, including two deaths, in the National Reporting and Learning Service (NRLS) where the wrong infusion flush fluid was attached to the arterial line and a further 76 incidents of faulty sampling technique [1]. The wrong infusion fluids most frequently used in these incidents were glucose 5% (22.6%) and mixtures of glucose and saline (19.0%). One of the deaths in this report resulted from the accidental use of glucose 5% as the arterial line flush fluid. This resulted in artificially elevated blood glucose values in arterial line samples and subsequent inappropriate insulin administration, causing severe hypoglycaemia, neuroglycopenia and irreversible brain injury [2]. A further analysis by the NPSA after their 2008 alert (July 2008 to May 2011) identified another 169 reports where the wrong infusion fluid was attached to arterial lines (Prof D Cousins, Senior Head of Patient Safety for Safe Medication Practice and Medical Devices, NHS England, personal communication) (Table 1). Glucose 5% accounted for 29.0% of these errors. In total, glucose-containing solutions accounted for the wrong infusion fluid in 102 (60.4%) errors: 30 (29.4%) reports where a glucose-containing fluid was used included a glucose monitoring error. The 2008 NPSA Rapid 1114 2013 The Association of Anaesthetists of Great Britain and Ireland

Leslie et al. Survey of arterial line practices Anaesthesia 2013, 68, 1114 1119 Response Report made several recommendations for avoiding these errors (Table 2). In 2011, an incident occurred on our intensive care unit (ICU), described elsewhere [3], where sodium chloride 0.9% with glucose 5% was used as an arterial flush. As in the case described above [2], this led to falsely high blood glucose readings, the prescription of insulin and iatrogenic hypoglycaemia. At the time of this incident, we judged our ICU s policies and practice to be compliant with the NPSA s 2008 recommendations. Our local investigation of the event identified a number of latent risks not all captured by the NPSA alert [3]. The Table 1 Number and severity of 169 arterial line errors identified as reported to NRLS between July 2008 and May 2011. Values are number (proportion). Severity of injury Death 0 Severe harm 1 (0.6%) Moderate harm 6 (3.6%) Mild harm 34 (20.1%) No harm 128 (75.7%) Year of event 2008 27 (16.0%) 2009 55 (32.5%) 2010 64 (37.9%) 2011 23 (13.6%) Table 2 The 2008 NPSA Rapid Response Report recommendations 2008 [1]. Arterial blood sampling should be performed only by competent and trained staff. Arterial lines must be clearly identified by use of a labelling system such as continuous coloured lines. Any infusion attached to the arterial line must be prescribed and checked before administration. Further checks should be made at regular intervals such as shift handover. Only sodium chloride 0.9% should be used as the infusion fluid for arterial lines Labels should clearly identify contents of infusion bags, even when pressure bags are used. Over time, manufacturers should develop a universal system to address this problem. inquest following this patient s death focused in great detail on arterial line use and practices. We conducted this survey to assess current practice surrounding arterial line use, to determine whether the NPSA recommendations had been adopted nationally and to identify whether other latent risks remained widespread. Methods We undertook a telephone survey of all adult ICUs in the UK between December 4, 2012 and January 24, 2013. We posed questions to the nurse in charge of the unit, or if they were unavailable, to a senior nurse or senior doctor. We asked structured questions assessing: the unit s preparation of the arterial line flush system; the technique for arterial blood sampling; the safety checks of the arterial line whilst in place; and the practice regarding glycaemic control and glucose sampling (Table 3). We entered data into an Excel spreadsheet for analysis (Microsoft Excel 2010; Microsoft Corporation, Redmond, WA, USA). Results We contacted all 241 adult ICUs in the UK. The questions were answered by respondents in 228 (94.6%) units; 13 (5.4%) units declined to take part. The nurse in charge responded in 118 (51.5%) units, ICU sister in 99 (43.4%) units, other nurse in eight (3.5%) units, ICU consultant in two (0.9%) units, and the respondent was not recorded in one (0.4%) unit. Out of the 228 responses, 223 (97.8%) were from mixed ICUs, three (1.3%) surgical ICUs and one (0.4%) a medical ICU. All percentages described below refer to percentages of respondents answering that question. Use of written arterial line policy In 188 (82.5%) units, there was a written policy for arterial line use. Arterial line flush fluid In 218 (95.6%) units, sodium chloride 0.9% was used as the flush fluid for arterial lines; nine (3.9%) units used heparinised saline and for one (0.4%) unit the response was not recorded. In 22 (9.6%) units, fluids used for flushing arterial lines were stored separately 2013 The Association of Anaesthetists of Great Britain and Ireland 1115

Anaesthesia 2013, 68, 1114 1119 Leslie et al. Survey of arterial line practices Table 3 Questionnaire used in the survey. What is your job title? What type of ICU do you work in? Regarding arterial line management, do you have a written policy or standard operating procedure? Which fluid is used as a flush for arterial lines? Is the fluid prescribed? Do you take blood samples for blood glucose analysis from the arterial line? What volume do you discard? Why this volume (especially do they indicate 3x deadspace)? Do you know what the deadspace for your system is? Do all nurses always withdraw and discard the same volume? (If arterial line used for blood glucose estimations) Do you check a peripheral blood glucose regularly? If yes, how often? How are fluids for arterial vs intravenous use stored/segregated? Do nurses perform a two-person check and sign arterial line fluids before attaching to arterial lines? How is it checked at change of shift? Take out of bag? Two-person check? Both sign that check complete? Describe your unit s pressure bag for use with arterial lines? (clear / opaque / mesh / combination) Do you have an insulin policy to control blood glucose in sick patients? Is this a locally created policy or one from another centre? What is your blood glucose target range? Is there a maximum blood glucose reading trigger for informing a doctor? Is there a maximum insulin rate trigger for informing a doctor? Are you aware of the NPSA s 2008 Rapid Response Report on problems with arterial lines and glucose sampling? Have you have had errors (wrong bag, wrong flush etc) on ICU. Please describe if you are happy to Have you have had errors (wrong bag, wrong flush etc) in theatre or elsewhere. Please describe if you are happy to Do you admit children to your unit (even for preparation for retrieval)? Do you have any fluids specifically for paediatric use on the unit (e.g. saline with glucose)? Does this include saline 0.9% with glucose 5%? Was a formal risk assessment performed before bringing in such fluids to ICU? Do you have a reliable mechanism to inform all agency or bank nurses of your arterial line policy/standard operating procedure? from intravenous fluids. In 202 (88.6%) units, these fluids were stored in the same location as intravenous fluids, whilst in four (1.8%) units, the respondent was unable to answer. In 165 (72.4%) units, the arterial line flush fluid was prescribed by a doctor. Where it was prescribed, respondents stated that it was included on their standardised ICU infusion prescription charts. In 189 (82.9%) units, the nurses used a two-person double-check and signed for the arterial line flush fluid before attaching it to the arterial line. Overall procedures for checking and documenting flush fluids at shift handovers were highly variable, with few following best practice. At nursing shift handover, two nurses double-checked the fluid attached to the arterial line in 111 (48.7%) units, both nurses signed to record this check in 46 (20.2%) units and the fluid was routinely removed from the pressure bag to check its contents in 17 (7.5%) units. All units used a pressure bag to enable infusion of the flush fluid. The respondents described this as follows: clear by 94 (41.2%); opaque by 22 (9.6%); meshed by 58 (25.4%); and a combination of these by 54 (23.7%). Other high-risk fluids One hundred and two (44.7%) units admitted children, although in many cases only for stabilisation before transfer. Of these units, 61 (59.8%) stocked specific paediatric fluids and 53 (86.9%) of these units stocked sodium chloride 0.9% with glucose 5%. Sampling technique Two hundred and eighteen (95.6%) units used an open arterial line transducer-sampling system and 10 (4.4%) units a closed system. The volume of blood discarded before taking a sample varied: 2 5 ml in 160 (73.4%) units and < 2 ml in 20 (9.2%) units. In 151 (69.3%) 1116 2013 The Association of Anaesthetists of Great Britain and Ireland

Leslie et al. Survey of arterial line practices Anaesthesia 2013, 68, 1114 1119 units, the volume of blood discarded was consistent, while practice varied in 64 (29.4%) units. One hundred and three (47.2%) respondents stated that the discarded volume was to clear the line or clear the deadspace and 109 (50.0%) respondents were unsure as to why this volume of fluid was discarded. No respondent mentioned three times the deadspace. Of the 103 respondents who described clearing the line or deadspace, 41 (39.8%) respondents knew the deadspace of their system and out of these, only nine (22.0%) withdrew and discarded a volume of at least three times this deadspace volume. Blood glucose monitoring In 220 (96.5%) units, an insulin policy was used to control blood glucose levels: 82 (36.0%) units aimed for a tight blood glucose range (4.0 6.0 mmol.l 1 ) and 135 (59.6%) units aimed for a conventional blood glucose range (4.0 10 mmol.l 1 ). Ten (4.4%) respondents did not know the target range. In 211 (92.5%) units, arterial lines were used for glucose monitoring; 70 (33.2%) of these units also regularly took a venous or capillary blood glucose reading to validate the arterial sample result. The frequency of venous/capillary sampling varied between once every hour and once every 24 hours, with only five (7.2%) units sampling more frequently than once every four hours. One hundred and two (44.7%) respondents reported that their units had a pre-defined elevated blood glucose level that triggered review by a doctor; similarly, 60 (26.3%) units had a pre-defined insulin infusion rate that triggered a medical review. Awareness of the NPSA Rapid Response Report Sixteen (8.7%) respondents had heard of the NPSA Rapid Response Report on problems with infusions and sampling from arterial lines. Problems with arterial lines In 69 (30.3%) units, respondents were aware of errors that had occurred with arterial line use in their ICU: 21 (30.4%) stated that the error was due to attachment of the wrong infusion fluid and 48 (69.6%) were unsure of the type of error. Glucose 5% was the most frequently implicated incorrect fluid, being cited in 14.5% of all described errors. In a further 72 (31.6%) units, respondents were aware of errors that had occurred with arterial lines inserted outside the ICU, such as the operating theatre or the emergency department: 21 (29.2%) stated that the errors were due to attachment of the wrong infusion fluid and 51 (70.8%) were unsure of the type of error. Again, glucose 5% was the most frequently cited incorrect fluid, featuring in 9.7% of all described errors. Agency or bank nurses In 99 (43.4%) ICUs, the respondent stated that there was a reliable mechanism in place to inform agency and bank nurses about the local arterial line policy. Of the others, 103 (45.2%) units did not have a reliable mechanism. Twenty-six (11.4%) units stated that they never used agency or bank nurses. Discussion There are many potential causes of error in the preparation of arterial line flush systems and in sampling from them. Factors that may contribute include: similar packaging and labelling of intravenous infusion fluids; inadequate checking of infusion bags before connection to arterial lines; the use of pressure bags that obscure fluid labels; poor sampling technique; contamination of samples by infusates; and confusion between arterial and venous lines [1]. Only a minority of respondents were specifically aware of the 2008 NPSA Rapid Response Report, although of course this does not mean the recommendations therein had not been acted on. For instance, amongst the recommendations made in the report was that any fluid attached to an arterial line must be prescribed. Our survey shows that this was done only in about three quarters of units. Including the prescription on the standardised infusion prescription chart was reported by respondents to improve compliance. The NPSA also recommended checking of arterial flush infusions before attaching them to the patient and at regular intervals thereafter. Checking could be improved by requiring signatures on this chart at shift changes. This and other recommendations are summarised in the report of our recent incident [3]. To avoid contamination of arterial line blood samples with flush solutions, it is necessary to discard at least the flush fluid that lies in the arterial line between 2013 The Association of Anaesthetists of Great Britain and Ireland 1117

Anaesthesia 2013, 68, 1114 1119 Leslie et al. Survey of arterial line practices the patient and the sampling port. International guidelines, now almost two decades old, recommend discarding more than three times the system deadspace [4]. However, a study conducted in 2010 reported that when glucose 5% is used as the arterial line flush, discarding five times the system deadspace did not prevent clinically significant sample contamination and elevation of blood glucose measurements [5]. Therefore, although three times the deadspace is adequate if sodium chloride 0.9% is used, it becomes inadequate if a solution containing a high concentration of glucose is used as the arterial line flush. The deadspace volume will depend on the type of cannula and the connecting arterial line system used. Our survey suggests that the technique used in many ICUs is inadequate to prevent contamination with infusate. The survey also indicated variations in sampling practice between and within individual units and poor knowledge of local equipment and of the reasons for discarding blood before sampling. Lack of knowledge and variations in practice by senior nursing staff may increase the risk sampling errors in many ICUs. Sampling venous or capillary blood glucose for comparison with arterial readings could theoretically help identify arterial flush errors; however, capillary blood glucose samples may themselves give inaccurate readings in the critically ill [6] and sampling would have to be very frequent, as less than three hours of hypoglycaemia can cause fatal neuroglycopenia [3]. Fewer than one in 15 of those units performing peripheral samples did so at this frequency. Trigger values of blood glucose or insulin infusion, where advice is sought from medical staff, may help prevent flush errors leading to patient harm by promptly identifying physiological derangements. Tight glucose control (which generally relies on protocols that will rapidly reduce elevated glucose levels and consequently may lead to high rates of insulin infusion [7]) was reported in more than one third of units. The current evidence does not support use of tight glucose control [8]. In 2007, the NPSA released a Patient Safety Alert aimed at reducing the risk of hyponatraemia in children [9]. This alert discouraged use of fluids with low sodium content in sick children and suggested, amongst others, sodium chloride 0.9% with glucose 5%. In this survey, we note that most units admitting children stock this fluid. This fluid may pose a particular risk of an arterial flush error: it is usually presented as a 500-ml bag; its label can readily be misread/mistaken for sodium chloride 0.9%; and its very high glucose content (280 mmol.l 1 ) will give falsely high blood glucose levels even if arterial samples are taken correctly. It was also the misadministered fluid in our ICU incident [3]. In conclusion, this survey has shown wide variation in practice, failure to comply with some recommendations of the 2008 NPSA alert, and the presence of other latent risks not covered by the alert. The data from the NRLS make it clear that arterial line incidents are still being reported at a rate of five per week, with the true number probably being up to 20 times higher [10]. This presents a major threat to patient safety and further national debate on standards of practice around arterial lines is warranted. The Association of Anaesthetists of Great Britain and Ireland (AAGBI) has established a working party to address this. Competing interests No external funding declared. TMC is a member of the AAGBI Working Party on preventing iatrogenic hypoglycaemia due to inappropriate insulin therapy in critical care practice. References 1. National Patient Safety Agency. Problems with infusions and sampling from arterial lines. Rapid Response report. NPSA/ 2008/RRR006. http://www.nrls.npsa.nhs.uk/resources/?entryid45=59891 (accessed 29/04/2013). 2. Sinha S, Jayaram R, Hargreaves CG. Fatal neuroglycopaenia after accidental use of glucose 5% solution in a peripheral arterial cannula flush system. Anaesthesia 2007; 62: 615 20. 3. Gupta KJ, Cook TM. Inadvertent hypoglycaemia caused by an arterial flush drug error: a case report and contributory factors critical analysis. Anaesthesia 2013 DOI: 10.1111/anae.12388. 4. Burnett RW, Covington AK, Fogh-Andersen N. Recommendations on whole blood sampling, transport, and storage for simultaneous determination of ph, blood gases, and electrolytes. International Federation of Clinical Chemistry Scientific Division. Journal of the International Federation of Clinical Chemistry 1994; 6: 115 20. 5. Brennan KA, Eapen G, Turnbull D. Reducing the risk of fatal and disabling hypoglycaemia: a comparison of arterial blood sampling systems. British Journal of Anaesthesia 2010; 104: 446 51. 1118 2013 The Association of Anaesthetists of Great Britain and Ireland

Leslie et al. Survey of arterial line practices Anaesthesia 2013, 68, 1114 1119 6. Jacobi J, Bircher N, Krinsley J, et al. Guidelines for the use of an insulin infusion for the management of hyperglycemia in critically ill patients. Critical Care Medicine 2012; 40: 3251 76. 7. van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. New England Journal of Medicine 2001; 345: 1359 67. 8. NICE-SUGAR Study Investigators, Finfer S, Chittock DR, et al. Intensive versus conventional glucose control in critically ill patients. New England Journal of Medicine 2009; 360: 1346 9. 9. NPSA safety alert: NPSA/2007/22. 28th March 2007 0409 Reducing the risk of hyponatraemia when administering intravenous infusions to children patient Safety alert -2007-03-28 V1 http://www.nrls.npsa.nhs.uk/resources/?entryid45=59809 (accessed 29/04/2013). 10. Sari AB, Sheldon TA, Cracknell A, Turnbull A. Sensitivity of routine system for reporting patient safety incidents in an NHS hospital: retrospective patient case note review. British Medical Journal 2007; 334: 79. 2013 The Association of Anaesthetists of Great Britain and Ireland 1119