Background Intrathecal administration of vincristine is a rare event but catastrophic for the patient, family and clinical team involved. Analysis of this source of harm shows it to be a classic systems error which has proved intractable for nearly 40 years. Failure to learn from history, communicate safety solutions nationally and internationally, create safety agencies which effectively and reliably prevent adverse events, conduct investigations and enquiries which fully reveals how to mitigate system error, develop robust physical design solutions to prevent harm to patients, make effective solutions universal and preparing for the unexpected are all major challenges.
Conclusions The elimination of rare yet catastrophic errors like this remains one of the tests of whether we can make healthcare safer. In this paper, we discuss why effective learning has been so slow and illustrate lessons for other fields of patient safety.
- Safety culture
- adverse event
- clinical guidelines
- drug therapy
- health policy
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
The Madagascar periwinkle (Catharanthus roseus) is a traditional remedy used to treat diabetes. In the 1950s, a scientific study on the 70 alkaloids within it isolated the vinca alkaloid, vincristine sulfate, which obstructs the spindle mechanism within human cells, arresting mitosis during metaphase. An application in the treatment of rapidly dividing cells was identified.
In 1963, vincristine was tested on 35 people.1 Overall, 40% of patients with Hodgkin's disease showed remission lasting more than 4 weeks. Yet, in general, 50% suffered severe side effects (mainly neuropathy) and had to discontinue treatment.
Later in 1963, the therapeutic use of vinca alkaloids was reviewed: ‘to date vincristine has not been administered intrathecally to any human being.’2 Intracisternal administration to a dog in a dose of 5 μg/kg had resulted in death.3
In 1968, vincristine was first administered intrathecally in error to a young patient with acute lymphocytic leukaemia. She died 3 days later. The histopathology was recorded. A study of intrathecal administration of vincristine to 20 New Zealand white rabbits yielded similar clinical and pathological features of severe toxicity.4
Survival, particularly for patients with blood cancers, has improved substantially over the last four decades, since vincristine became a major therapeutic agent. Over the same period the procedure for administering this drug has remained error-prone and a source of catastrophic harm. The specific risks of maladministration of vincristine sulfate were clearly recognised from the early experience in the 1960s. However, over almost 40 years, 58 cases of intrathecal vincristine errors are known to have occurred. At the time of preparing this paper, the most recent officially recorded was in Hong Kong in June 2007.5 Several others will have gone unreported.
Errors have long been recognised as a problem with therapeutic medicines. These have been shown to occur in 5.3% of all prescriptions. Route delivery errors, of which intrathecal vincristine delivery is one example (there are many other examples such as intravenous delivery of benzathine penicillin), account for 5% of medication errors.6
The delivery of intrathecal vincristine is a rare yet catastrophic error in the same category as a major unexpected airline engine failure, such as the incident with the BA038 Beijing-to-London flight in January 2008.7 These errors attract special investigation because of their catastrophic effects, even though they are likely to occur only once in each institution. Reason's Swiss Cheese model of error causation has been previously applied to this error to fully understand the accident trajectory.8 In this paper, we discuss why effective learning has been so slow and illustrate lessons for patient safety in five principal domains.
Failure to learn from adverse events
During the mid 1960s–1970s, despite vincristine being used clinically as an antitumour drug, relatively little was understood about its pharmacological effects. Partly this was because of a lack of appropriate labelling techniques: metabolites of unknown effect were still being discovered as late as 1977.9
There is a 10-year gap between the first recorded death due to intrathecal vincristine administration in 1968 and the second case in 1978 (there was an unpublished case in the UK in 1976). During these 10 years, use of vincristine in clinical practice escalated, and toxicity due to overdoses was recognised.
The published death in 1978 of a 5-year-old boy with acute lymphocytic leukaemia in the USA, who received intravenous and intrathecal chemotherapy at the same time, defined the hallmark mix-up error characteristic of intrathecal vincristine delivery. It raised the standard expected for safe administration.10 The case report is one of the first documents which makes clinical recommendations about safe practice. A third death was reported in 1980.11
During the 1980s and early 1990s, several publications advocated similar precautions to prevent intrathecal administration of vincristine.12–15 However, some continued to focus on successfully treating the consequences,14 16 others on less well studied neurology—for example, a coma which lasted 12 months.17 By the early 1990s, the guidelines recommended in early case reports were being quoted in current case studies.15 However, despite this addition to the literature, errors continued to occur internationally, and it appears as if lessons from the past were not being heeded.
Failure of international translation
International reporting and learning comparable with other high-risk industries has long been the goal of patient safety.18 Realising a system that allows fast and translatable knowledge to prevent error is common in aviation and other services where risk is a daily quantifiable experience. However, healthcare has yet to live up to this standard.
In 1994, 26 years after the first intrathecal vincristine death, the first international advisory was issued by the World Health Organization.19 This followed the Food and Drug Administration of the USA notifying the WHO of multiple case reports that had been accumulating internationally. The WHO reported that these events had been eliminated in the USA since 1991 due to improved labelling requirements recommended by the FDA. The UK Medicines Control Agency implemented a safety scheme involving a colour-coding mechanism after a death in 1990.
The next case in the USA in 1995 became the first reported long-term survivor (a 6-year-old boy who suffered permanent neurological dysfunction of the bladder),20 although one previous case had survived to 3 months and died from their illness rather than the error.14
The next published case in the UK in 1997 was the second long-term survivor (likely because of aggressive neurosurgical treatment). She was a 10-year-old girl left with lower-limb paraplegia.21 Other international reports around this time came from Australia and Singapore as recorded in a chronological summary in 2005 by the Society of Hospital Pharmacists of Australia.22
Failure to achieve compliance with safety guidance
In the mid 1990s, the patient safety movement expanded in some countries' health systems. The publication of the report: To Err is Human by the Institute of Medicine in the USA was a landmark.23 This report scoped the scale of medical error, explained the underlying concepts and encouraged systems thinking as well as setting goals for harm reduction. It attracted public as well as professional interest.
In 1998, a Medication Safety Alert on intrathecal medication was released by the Institute for Safe Medication Practices,24 a body devoted to patient safety and which had previously issued advice related to vincristine in the early 1990s. This was in response to another death in the USA following the FDA guidelines in 1991.25
Despite the advent of patient safety alerts, four more incidents occurred in Canada, Saudi Arabia, Korea and the USA.26–28 Another alert was issued by the ISMP in 2000.29 They announced that the FDA and ISMP would be intensifying efforts to prevent this error in the USA. Two more deaths were reported in the USA in 2003 and 2005.30
Failure of investigations and enquiries
The UK report, An Organisation with a Memory, featured as one of its case studies the intrathecal vincristine error.31 However, before the implications of this initiative were to be fully realised, an 18-year-old boy, Wayne Jowett, died in Nottingham, England in 2001. This led to the largest and most innovative investigation of an intrathecal vincristine incident.
In the Nottingham case, a skilled non-medical accident investigator was called in. He catalogued over 40 weaknesses in the hospital system that led to the fatal error.32 This report remains the most authoritative investigation of such an incident anywhere in the world and reveals this source of catastrophic error as a classic multifactorial system failure. This validated the prevailing scientific theory of patient safety—that medical error is largely due to system failure. A separate UK report to specifically prevent the error was produced by a specialist in the field later the same year.33
In the UK, the intrathecal medication protocols require multiple safeguards to be put in place, reflecting the system nature of the error. The UK also analysed the real stories of harm caused to patients and their families; this included the stories of the first and last patients to die due to intrathecal vincristine in the UK, Lee Duggins in 1976 and Wayne Jowett in 2001.34 A similar approach was adopted by the WHO.35
However, despite this renewed focus on the ‘system error,’ the promotion of a ‘no blame’ culture and the galvanising of patient champions, deaths continued, and individual doctors faced consequences. Following the death of Wayne Jowett, the supervising doctor was suspended and charged with manslaughter.36 37 This was not the first time legal action had been taken against doctors. Two doctors had been charged with manslaughter in 1991,38 their actions described as: ‘momentary, but untypical recklessness.’ An appeal in 1993 failed. The medicolegal aspects appeared in a brief discussion following the death of a women in Singapore in 1996.39
Failure of solutions
Wrong route delivery medication errors can be prevented in three principal ways (box 1).
Box 1 Prevention of wrong route delivery medication errors
Drug alerts that warn healthcare institutions, medical practitioners and other relevant personnel about new risks.
Development of consensus and evidence-based protocols and guidelines to create a standard operating procedure that prevents against error.
Physical redesign of delivery systems such that it is impossible to deliver drugs by the wrong route.
First, learning from previous adverse events has been slow and has not abated the frequency of error. Issuing drug alerts both nationally and internationally has failed to reliably drive improvements. It raises awareness but lacks the power to enforce change.
Second, although a protocol-driven approach to preventing intrathecal vincristine has defined the 21st-century approach to delivery, this has not halted the error. Multiple government agencies and bodies, nationally and internationally, have published extensive protocols and guidance on safe vincristine delivery as well as debate in medical literature covering multiple health professions.5 22 29–33 40–50 However, incidents have continued to occur on a regular basis, including Spain and Germany.51 52
In 2004, in the UK, implementation of the guidance for the safe administration of vincristine was reviewed.53 Three years after revised protocols and guidance were issued and widely disseminated, all hospitals reported compliance with national guidelines. Yet, an independent review revealed that only 50% were actually compliant. Guidelines and protocols offer healthcare institutions and medical practitioners the opportunity to reduce risk by using professionally agreed and industry-standard operating procedures, but responsibility and management are required by the user to change practice.
Third, there have been attempts to create delivery systems whereby intravenous and intrathecal drugs can only physically be given by their designated routes. Yet, international standards and a mosaic of different tubal connection possibilities within and between countries have kept this seemingly simple solution at bay. The WHO has set the challenge to make it: ‘physically impossible to attach an intravenous syringe containing vincristine to a spinal needle (‘lock and key design’).’49 This represents a forcing function making it virtually physically impossible to take the wrong action. This has had success in other areas—for example, the antihypoxic connections and devices for anaesthetic machines. The Spinal Injection Safety System developed in Canada has shown some promise but has not yet been used widely.47
The luer multicompatible connection system is used commonly in the UK National Health Service. The Department of Health in the UK has been pursuing a plan to implement a system to eliminate all Luer connections associated with spinal delivery systems. This initiative has progressed slowly due to the complexities of designing an entirely new connection system.54 Manufacturers have also been reluctant to invest in development when they believe there is little demand and no major market for a new product. To date, to our knowledge, no data exist comparing the cost of implementing a design solution throughout an entire health system versus the costs of continued medication errors. Yet, it has been reported that in Japan, following the death of a patient, a significant change was made to connectors on syringes in under a year.48
In recent years, the use of a minibag to deliver vincristine has become increasingly discussed and advocated.22 30 40 41 43 44 46 49 50 This is not a full physical design solution or forcing function but offers an attractive interim measure. It involves not using syringes to deliver vincristine at all and instead diluting the drug in a minibag, working on the premise that it would be virtually impossible to deliver this to a patient through a spinal needle. An alert issued by the National Patient Safety Agency in 2008 discovered that no incidents had been reported where a minibag had been used.50 The ISMP conducted a survey attracting 418 responses in December 2005. It determined that a quarter of respondents used minibags.55 However, as the error remains rare, it cannot be regarded as a definitive solution.
The WHO, in its latest drug alert on safe delivery of vincristine, has recommended using the minibag to prevent errors in the absence of a formal design solution which separates intravenous and intrathecal routes physically.49 The WHO published the alert immediately following the death of a patient in Hong Kong in July 2007. The Hong Kong Hospital Authority held an enquiry and later changed its policy to use the minibag for vincristine delivery.5 However, the minibag is not yet used everywhere.
Emergence of a new source of harm
Error-prone situations always continue to elude and surprise. In China, in 2007, the drug company Shanghai Pharmaceutical Group contaminated, at the factory level, methotrexate and cytarabin hydrochloride (drugs normally given intrathecally) with vincristine.56–60 The Chinese government recalled the drugs and closed the plant. At first, a few children in Shanghai and Guangxi Zhuang Autonomous Region suffered neurological symptoms. Eventually the outbreak was believed to have led to 193 patients nationwide having intrathecal vincristine unintentionally as part of their normal medication plans. Paralysis was common, and these incidents increased the world total from 58 to 251 overnight. The company has previously exported drugs to 70 countries worldwide.
Unusual events like this are outwith the scope of any existing error reporting system, and even the best design solution would not prevent such an unpredicatble and ‘upstream’ form of error. Yet, these events keep us alert to the inherent dangers in every healthcare establishment and add another vulnerability to this long-running story of unsafe care.
One of the greatest expectations in patient safety is to demonstrate that risks can be quickly and reliably reduced once identified. So far, this has proved very difficult. Many patient safety solutions are embryonic in their conceptualisation, translation and implementation.
Effective approaches to reducing risk include: risk identification through reliable and universally applicable reporting systems, prioritising initiatives and interventions, effective use of standard operating procedures and forcing functions, and demonstrating a measurable reduction in harm. However, unlike aviation, nuclear and other high-risk industries, healthcare organisations remain unfocused and unable to rapidly reduce potentially fatal risks, perhaps in part due to the lack of penalties able to be enforced in healthcare compared with other industries.
We have explored five domains of failure to learn from for this rare yet catastrophic error (table 1).
Consensus is needed internationally to halt this error. The WHO has a pivotal role to play in developing an international alert system for fast, accurate and translatable information to be quickly disseminated throughout the world's healthcare organisations. A healthcare version of the aviation industry's Commercial Aviation Safety Team (CAST)—a public–private partnership, with significant buy-in from all relevant bodies, which is responsible for reducing fatalities in the aviation industry—may be an effective way of gaining consensus and policy-making power to overcome the market's drive away from seemingly expensive design solutions, the principal intervention called for by WHO.61 WHO has a significant role to play in convening such initiatives.
Intrathecal administration of vincristine is a rare event but catastrophic for the patient, family and clinical team involved. Analysis of this source of harm shows it to be a classic systems error which has proved intractable for nearly 40 years, since the therapeutic use of vinca alkaloids came into being. The ultimate solution may be through design, but until then, this error remains one of the test cases of whether we can make healthcare safer.
The authors alone are responsible for the views expressed in this publication, and they do not necessarily represent the decisions, policy or views of the World Health Organization.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.