A response to the problem of spinal misconnection errors

Following the publication of this report, new guidelines and an award-winning DVD were produced for the safe administration of vincristine.8 9 As a result, there were widespread changes to the management of drugs and the training and registering of staff who administer these drugs. However, the possibility of error has not been entirely eliminated and the new procedures do not affect spinal injections in anaesthesia where errors also occur. To meet the target set out in an “An Organisation with a Memory”10 to reduce the number of patients dying or being paralysed by maladministered spinal injections to zero by the end of 2001, an engineered solution was acknowledged to provide the most effective way of completely eliminating errors from the “system”.11 12 A further report on Luer connectors, for the European committee for Standardization (CEN),13 cited a number of incidents that involved the misconnection of Luer connectors and suggested that the only way to achieve a safe system would be to design different male and female connectors that will not allow inter-changeability between intended routes of delivery. In summary, there seemed little doubt about the need for investigations to assess the feasibility and safety of a system of non-Luer spinal connectors.

In 2002, the Department of Health (DH) called for manufacturers to develop a prototype non-Luer spinal connector. This was followed in 2004 with the publication by the National Patient Safety Agency (NPSA) of a risk assessment of spinal procedures for the DH.6 This document detailed the risks present in spinal procedures even with the recently introduced DH guidelines in place, and how these risks would be affected by a potential engineered solution. The risk assessment concluded that “replacing all current spinal equipment with devices with unique spinal connectors would be the most effective method to minimize the risk of maladministration of vincristine, and would help to prevent wrong route errors with other medicines and spinal procedures” (p. 4, par. 18). Thus, the NPSA report acknowledged that the solution to the maladministration of vincristine could also be used to prevent other errors in the administration of medicines (bolus or infusion) via either the intrathecal or epidural route.

The NPSA risk assessment also highlighted the need to assess the potential hazards associated with the implementation of this option that may arise from supply, storage and other logistical problems. In response, the DH commissioned a research project at the University of Leeds to analyse prospectively the risks associated with the implementation of a non-Luer spinal connector. It should be noted that the manufacture or testing of devices for epidural infusions did not fall within the remit of this work. The remainder of this article reports on the findings of this work.

In response to a DH advertisement in 2002 and following two rounds of laboratory tests conducted at the Surgical Materials Testing Laboratory in Bridgend, two design solutions were put forward by the DH non-Luer steering committee for further non-clinical testing. The testing was of two kinds: usability testing at the Bath Institute of Medical Engineering and a prospective hazard analysis (conducted by Asher Consulting in collaboration with Institute of Psychological Sciences at the University of Leeds). The process of design and testing was iterative, with findings from the testing providing feedback to the manufacturers, which were then used to make modifications to the prototypes.

Initial laboratory and design testing

In response to a DH/Purchasing and Supplies Agency tender exercise in 2002, a range of engineering and user tests were performed on three prototype systems at the Surgical Materials Testing Laboratory, Princess of Wales Hospital, Bridgend. The engineering tests were designed to ensure that the devices met certain performance requirements which standard Luer connectors are expected to meet based on BS EN 1707:1997. These tests assessed a range of parameters, including air/liquid leakage and separation force.

For the user tests, a panel of clinicians were asked to assess the ability of the devices to cross-connect with standard Luer connectors. A surprising number of the early prototypes were found to leak and to cross-connect with standard Luer connectors, therefore a second round of testing was performed on modified devices to ensure that appropriate design changes had been implemented to address these issues.

This phase of testing, which concluded in 2006, produced two design solutions which were assessed as being suitable to progress to further testing. The prototype connector solutions were both similar in design to the traditional Luer connector, but were fitted with a different locking mechanism to prevent connection of equipment not intended for spinal use (eg, an intravenous syringe).

Non-clinical evaluation

Usability testing

Using two prototype systems, three different spinal procedures were simulated—an intrathecal chemotherapy procedure, placement of an epidural catheter and delivery of a spinal anaesthetic bolus. In the first round of testing 30 participants who perform these spinal procedures (comprising 11 consultants, 12 specialist registrars, 6 staff grade doctors and 1 specialist nurse) were recruited from Bristol and Bath hospitals, both large Trusts covering the necessary specialties (ie, haematology and anaesthesiology). The procedures were conducted using a spinal dummy (see fig 2) and recorded using a video camera to capture difficulties and other comments. Feedback was sought from clinicians in two forms: (1) concurrent verbalisation while completing the task to highlight any usability concerns and (2) brief structured interviews immediately following each scenario to discuss ease of use and clinical suitability of each of the non-Luer spinal connectors. At the end of the interviews, participants were asked to evaluate the performance of both prototypes in each testing scenario by giving an overall satisfaction rating out of 10.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Use of a spinal trainer dummy in the usability testing.

Prototype 1 took the form of a separable adaptor, which the manufacturer proposed could be attached to standard Luer syringes to render them non-Luer compatible. Despite positive satisfaction scores (mean = 7.2/10), it was the view of the participating clinicians and of the DH non-Luer steering group, that this prototype did not meet the requirements of the project, because of the separable non-Luer connector. It was recommended that the manufacturer provide devices with the non-Luer connector as an integral part of their design, or as permanently bonded in place. Following this guidance, the manufacture failed to provide modified prototypes for a second round of testing. Table 1 shows the mean satisfaction ratings given by clinicians following their use of the two prototypes in three simulated scenarios.

The second manufacturer provided devices which featured non-Luer compatible connectors that were permanently bonded on. In the first round of testing, prototype 2 achieved a mean satisfaction rating of 6.4 out of 10 in the simulated scenarios. Clinician comments revealed several shortcomings in this design. For instance, the colour of the needle hub made it difficult to see cerebrospinal fluid (CSF) flashback. User feedback was given to the manufacturer with a set of recommendations for improvement of the connector design. Following modifications to this non-Luer system, it was submitted for a second round of testing with 15 clinicians (10 anaesthetists, 5 haematologists). This produced improved satisfaction scores, with the modified prototypes obtaining a mean satisfaction rating of 8/10. All clinicians found this system to be clinically acceptable in its current form. Sufficiently positive feedback was received for this prototype to be considered acceptable for inclusion in a clinical implementation study.

Prospective hazard analysis

Based on information from a literature review, interviews with six clinicians and observations of clinical practice, hierarchical task analyses (detailed descriptions of the action sequences performed within a task) were produced for each of the relevant spinal procedures. These included diagnostic lumbar puncture; measurement of CSF pressure using a spinal manometer; intrathecal chemotherapy; and spinal anaesthesia. The task analyses of intrathecal chemotherapy and spinal anaesthesia formed the basis for the prospective hazard analyses (PHA) that took place during six workshops in two large NHS Trusts in the UK.

The hazard analysis method used was a combination of hazard identification (HAZID), originally developed by the Institute of Chemical Engineers, and potential human error assessment (PHEA), originally developed by British Coal. The combination of techniques is important since HAZID ensures a systematic and comprehensive approach to hazard identification and PHEA focuses on the errors that represent precursors to hazards. This hybrid PHA approach has been used extensively during the investigation of human error in the coal mining and nuclear power industries, by our collaborator at Asher Risk Consulting.

Twenty-four healthcare professionals attended the six PHA workshops, including 11 consultants, five pharmacists, four nurses, two specialist registrars and two pharmacy technicians. Each workshop lasted approximately three hours and involved two stages. First, the task analyses, which were entered into the left hand column of a spreadsheet and displayed using a multimedia projector, were reviewed by the attendees and modified where they did not accurately capture the nature of the spinal procedure in question. Second, the verified task analyses were used to review systematically the spinal procedures in an effort to identify potential hazards in their preparation or performance. This involved asking attendees to think carefully about the potential for error in each of the subtasks of the task analysis. For each action in the task analysis, attendees were asked to consider the following questions: (a) Who initiates the action? (b) Who takes the action? (c) Is there any potential for error in the action? (d) If there is potential for error, what would be the likely consequences of the error? Also, would the error be recoverable at a later stage in the process (eg, an error in drug preparation may be recovered during routine checking) and (e) would the introduction of non-Luer equipment affect the error potential favourably (eg, by preventing the error) or unfavourably (eg, by increasing the likelihood of the error occurring)? For an example of a completed PHA spreadsheet, please see online Appendix A.

The above information was elicited from attendees and entered into the PHA spreadsheets during the workshops. Following the workshops, the potential errors identified were classified by the research fellow and the risk consultant in two ways: (1) in terms of whether they were errors in the input, decision or output stage of a task and (2) according to whether they could be characterised as slips, lapses, mistakes or violations.14 Slips and lapses occur when an action sequence appropriate to the situation is executed incorrectly. Slips are observable, whereas lapses are not. A mistake is the successful execution of a faulty plan of action and a violation is a deliberate breach of rules or procedures. These classifications are important as different types of errors may be remedied using different interventions.

The PHA process revealed certain potential errors (eg, mistakes in drug dose preparation) that could occur regardless of the type of equipment used, be it Luer or non-Luer. The focus of this exercise, however, was on the way in which the introduction of non-Luer equipment would impact upon the spinal procedures carried out and the changes that might be required to work practices, storage of drugs, etc. to facilitate the implementation.

The prospective hazard analysis identified two potential risks that will need to be managed as part of a successful implementation programme. These were the possible failure in the supply chain and the potential implications of additional devices being stored in clinical areas alongside Luer equipment. These issues will be further investigated in the final phase of the project, where the selected non-Luer system will be evaluated in clinical use. In addition, the analysis indicated minor procedural changes that would be necessary to facilitate implementation. For instance, where pharmacy drug preparation sheets detail the equipment needed for each intrathecal chemotherapy preparation, these will need to be modified to include non-Luer devices.

The PHA also revealed support by anaesthetists for the extension of this design solution, which although not introducing new risks, does not address the risks posed by misconnection errors in epidural anaesthesia. As epidural infusion lines may be left in situ for several days, during which time patients are removed to the more hazardous ward environment, the likelihood of cross-connection is perhaps greater here than in spinal anaesthesia. Indeed, between 2000 and 2004, three patients died following the misconnection of epidural bupivicaine (a local anaesthetic) to an intravenous port.15 Thus, following the successful implementation of non-Luer equipment for spinal injections, the design solution of dedicated spinal connectors may be extended for use in epidural infusions.

Based on the findings of these two pre-implementation testing procedures, a device has been identified that meets the requirements of the users, being very similar to the existing devices in performance yet not allowing for connection with the Luer. Moreover, the device requires no additional effort to use, that is, it is not an add-on solution. Feedback from clinicians has also resulted in minor modifications to the devices, for example, changes were made to the translucency of the connector to improve CSF visibility.

What next?

In the current phase of this project, clinicians at Leeds NHS Trust will have the opportunity to use the new devices. The implementation of the non-Luer connector into practice will be evaluated by the project team from the Institute of Psychological Sciences at University of Leeds and Product Design at Sheffield Hallam University. Following this phase of the project, and together with the lead clinicians in adult and paediatric oncology and obstetric anaesthetics, we will report on the implementation of the non-Luer devices in a follow-up article. A set of implementation guidelines for the wider roll-out of the non-Luer devices will also be produced. The final stage of the project will involve the launch of the product onto the market and an evaluation of its uptake.

Conclusion

In conclusion, an engineered solution to the problem of maladministration of spinal injections has been extensively tested and piloted. For a full report of the pre-implementation evaluation of the non-luer connector in clinical practice please see http://www.pcpoh.bham.ac.uk/publichealth/psrp/documents/PS049_Non-Luer_Finor1_Report.pdf. This process has taken time and effort from manufacturers, clinicians and policy makers, but the potential benefits to patient safety may represent one life saved per year and many fewer patients injured as a result of spinal administration errors. The extension of the non-Luer connection system to epidural infusion equipment would further reduce the likelihood of misconnection errors. The novel spinal connectors under evaluation demonstrate the potential significant benefits of incorporating judicious equipment design within a robust systems approach to patient safety. Considerable reductions in litigation costs to institutions and individuals may also be expected as a positive outcome of a widespread introduction of a safe non-Luer system for use in spinal procedures.