Litigation related to drug errors in anaesthesia: an analysis of claims against the NHS in England

Transcription

1 doi: /j x Litigation related to drug errors in anaesthesia: an analysis of claims against the NHS in England J. Cranshaw, 1 K. J. Gupta 2 and T. M. Cook 2 1 Consultant Anaesthetist, Royal Bournemouth Hospital, Bournemouth, UK 2 Consultant Anaesthetist, Royal United Hospital, Bath, UK Summary Ninety-three claims (total cost ) filed under anaesthesia in the NHS Litigation Authority database between 1995 and 2007, alleging patient harm directly by drug administration error or by an allergic reaction, were analysed. Alleged errors were categorised using systems employed by the National Coordinating Council for Medication Error Reporting and Prevention, the American Society of Anesthesiologists Closed Claims Project and the UK Health and Safety Executive. The severity of outcome in each claim was categorised using adapted National Patient Safety Agency definitions. Sixty-two claims involved alleged drug administration errors (total cost ) and 15 resulted in severe harm or death. Half alleged the administration of the wrong drug, in most (16) a neuromuscular blocker. Of the claims alleging the wrong dose had been given (25), nine alleged opioid overdose including by neuraxial routes. The most frequently recorded adverse outcomes were awake paralysis (19 claims; total cost ) and respiratory depression requiring intensive care treatment (13 claims; total cost ). Thirty-one claims involved allergic reactions (total cost ). In 20 claims, the patient allegedly received a drug to which they were known to be allergic (total cost ). All claims in which it was possible to categorise the nature of the error involved human error. Fewer than half the claims appeared likely to have been preventable by an ideal double checking process.... Correspondence to: Dr J. Cranshaw Julius.Cranshaw@rbch.nhs.uk Accepted: 15 August 2009 Preventable medication errors have been suggested to cause 10 20% of in patient adverse events [1] and four of the nine World Health Organization (WHO) solutions for improving patient safety involve reducing such errors [2]. However, few studies of medication error simultaneously report their cause, frequency and outcome. Furthermore, inconsistent reporting and recording methodology make verifying the relative importance of different adverse events difficult. Anaesthesia is unusual in requiring the administration of several potent, rapidly acting drugs in a relatively brief timeframe. These drugs would be harmful if given without considerable care and attention to dose, timing and order of administration. In the UK, these drugs are almost exclusively administered by physicians and the drugs are rarely checked by anyone other than the anaesthetist before administration. Drug error in relation to anaesthesia may therefore be of particular interest both to the specialty and the wider population. The National Health Service Litigation Authority (NHSLA) has records of all clinical claims made against the NHS in England by patients and their families since 2002 and of substantial claims made between 1995 and For each claim, their database provides details of the year of the incident and its notification to the NHSLA, a brief description of the circumstances, the alleged cause, the resulting injury and the medical specialty involved. The database classifies claims as open (claim still active) or closed (claim settled or withdrawn). By inference the NHSLA database contains examples of adverse outcomes that were of importance to patients. This paper analyses claims in the database alleging medication (including blood) errors and allergic reactions to drugs in association with anaesthesia. The claims are grouped according to their cause and severity of outcome Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 1317

2 J. Cranshaw et al. Æ Drug errors in anaesthesia Anaesthesia, 2009, 64, pages and the paper considers which errors might have been amenable to prevention. It also presents the relative financial importance of the groups of claims in the context of the NHSLA database. It is important to note that the cost associated with each claim recorded in the NHSLA data set represents the combined cost of legal fees and settlements in respect of damages ; these are not separated in the data set. However, in a recent NHSLA report, the total cost of legal fees was slightly higher than settlements [3]. In addition, the available data does not state if a case was defended, or settled in or out of court. Methods All 1067 claims in the NHSLA database ( ) allocated to anaesthesia by specialty code were independently scrutinised by three consultant anaesthetists who excluded misclassified and uninterpretable claims. The remaining 841 claims were then searched for claims alleging patient harm directly from drug administration or from an allergic reaction. At each stage, if all the consultants agreed sufficient relevant data existed in a claim, they was included in the data set. If all agreed that the claim data were irrelevant or inadequate, they was excluded. Disagreements were resolved by discussion [4]. Drug errors were categorised using the National Coordinating Council for Medication Error Reporting and Prevention (NCCMERP) taxonomy [5]. However, claims categorised under this taxonomy as drug administration at the wrong time were relabelled wrong order as all such claims alleged administration of a neuromuscular blocker before an induction agent. For comparison with studies in the USA relating to drug administration errors, claims were also categorised according to one system used by the American Society of Anesthesiologists (ASA) Closed Claims Project [6, 7]. The severity of outcome in each claim was classified using an adaptation of the National Patient Safety Agency (NPSA) definitions for grading patient safety incidents [8]. The type of human error described in each claim was categorised using the UK Health and Safety Executive categorisation of human failure [9]. Results Ninety-three relevant claims were available for analysis (total cost ). This equated to 11% of all classifiable claims in the NHSLA data set. Sixty-one (7% of all claims) alleged patient harm directly by drug administration. One alleged harm due to errors that included drug omission. Thirty-one (4%) alleged harm by an allergic reaction. Claims alleging harm directly due to drug administration error (n = 62) The total cost relating to these claims at the time of analysis was (49 closed; median ; mean ; range ; 13 open). All 16 claims made between financial years and were closed. Thirteen of the subsequent claims remained open. Classification or error Using the NCCMERP taxonomy, 30 claims alleged administration of the wrong drug; 24 that a drug was administered in the wrong dose; five that a drug was administered in the wrong order; and two that administration occurred by the wrong route (Fig. 1). In one claim there was insufficient detail to classify the error. In 16 of the 30 claims alleging the wrong drug had been administered, this was a neuromuscular blocker. Suxamethonium was named in nine. Of the 24 claims alleging the wrong dose had been given, 22 alleged overdose: 13 intravenous and nine neuraxial. In 14 claims, an opioid was cited and in half it was administered neuraxially. The wrong routes of administration were one epidural and one arterial instead of intravenous. Three claims involving neuraxial injection required discussion by the investigators for categorisation. In these, a wrong route error was a possibility that could not be excluded. By the classification system used to sort ASA closed claims, 28 alleged substitution error; 23 incorrect dose; six insertion error; two the incorrect route of administration; one was a complex case not easily classified; and one involved omission of a drug. There were no repetition errors (Fig. 2). Clinical outcomes The severity of the adverse outcome after the 62 alleged drug administration errors is presented in Table 1. Wrong route 2 (3%) Wrong order 5 (8%) Wrong dose 24 (39%) Wrong drug 30 (50%) Figure 1 Classification of the 61 classifiable drug administration errors using the NCCMERP taxonomy Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland

3 J. Cranshaw et al. Æ Drug errors in anaesthesia Insertion 6 (10%) Incorrect dose 23 (38%) Incorrect route 2 (3%) Omission 1 (2%) Other 1 (2%) Substitution 28 (45%) Repetition 0 (0%) Figure 2 Classification of the 61 classifiable drug administration errors using a classification employed by the ASA Closed Claims Project. The most frequently recorded adverse outcome was awake paralysis, the unintentional administration of a neuromuscular blocker to an awake patient (19 (31%)) claims (17 closed; total cost ; median 6000; mean ; two open). Two further claims were categorised as awareness under anaesthesia as the timing of the event was imprecisely described (both closed, cost 0 and ). The next most frequently recorded adverse outcome was respiratory depression requiring intensive care treatment (13 claims (21%); nine closed; total cost ; median ; mean ; range ; four open). In nine claims an opioid overdose was allegedly responsible. In the other four the drug was not named. Intravenous administration was noted in six, neuraxial in three. In one case the outcome was death and in 10 respiratory arrest, eight resulting in central nervous system injury. The frequency of other adverse outcomes was less than 10. Timing and location of drug errors Alleged errors occurred on induction (24; 39%), during maintenance (18; 29%), on wards (15; 24%), in the Table 1 The grade of severity of adverse outcome after alleged drug administration errors and after allergic reactions. Values are number (proportion). Severity of outcome Claims alleging harm directly due to drug administration (n = 62) Claims involving allergic reactions (n = 31) Death 5 (8%) 5 (16%) Severe 10 (16%) 10 (32%) Moderate severe 1 (2%) 1 (3%) Moderate 16 (26%) 4 (13%) Mild moderate 7 (11%) 4 (13%) Mild 20 (32%) 7 (23%) Unknown 3 (5%) 0 % recovery area (three; 5%), and during administration of epidural therapy (two; 3%). Fifteen (24%) claims were made peri-partum (13 closed; total cost ; median ; mean 9460; range ; two open). Eight reported awake paralysis with seven alleging suxamethonium had been substituted for another named drug after delivery: in five for Syntocinon Ò (Alliance Pharmaceuticals, Chippenham, Wiltshire, UK), in one for gentamicin; and in one for ondansetron. Two alleged that Syntocinon was given in error before delivery, precipitating unnecessary emergency caesarean section. Four alleged overdose of neuraxial opioids and or local anaesthetic, and while none claimed respiratory depression, one claimed pneumonia had ensued. Type of human error involved in erroneous drug administration In eight claims, limited clinical detail meant it was impossible to determine the nature of the alleged error. However, in all the other claims, events could be ascribed to human error rather than equipment or monitoring failure. Of those that could be categorised, 26 claims apparently involved satisfactory anaesthetic plans that deviated in some unintended way, errors categorised as slips (unintended actions due to failure of attention) or lapses (unintended actions due to memory failure). In six claims, errors could be categorised as mistakes (errors of judgment or decision-making in the application of knowledge or rules). Twenty-eight claims apparently involved the misidentification of an intravenous drug, including one misidentification of blood. Claims involving allergic reactions (n = 31) Thirty-one claims involved allergic reactions (26 closed; total cost ; median 2425; mean ; range ; five open). The severity of outcome in these claims is presented in Table 1. Claims related to previously reported allergens In 20 (65%) claims, the patient allegedly received a drug to which they were known to be allergic (15 closed; total cost ; median 3346; mean 6540; range ; five open). In six claims, the reported allergen was a penicillin. Four claims involved severe adverse outcomes but no long-lasting sequelae were reported (two closed; and ) and five involved rashes (all closed ) but no deaths, cardiorespiratory arrests or central nervous system damage were recorded. Claims related to unreported or unknown allergens All 11 claims related to previously unknown allergens involved severe anaphylactic or anaphylactoid reactions Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 1319

4 J. Cranshaw et al. Æ Drug errors in anaesthesia Anaesthesia, 2009, 64, pages (10 closed; total cost ; median 802; mean ; range ; one open). The outcomes were death in five, cardiorespiratory arrest with subsequent central nervous system injury in four but in two, no long-lasting sequelae were documented. In two, involving central nervous system injury, the claims included allegations of inadequate management of the reaction. Type of human error involved in claims involving allergic reactions Human error in this group of claims could not be categorised. In claims where the allergen was not previously identified, no error was attributed. However, in claims where an allergen had been previously reported, it was not possible with the available clinical details to determine the nature of any error that might have led to that drug s administration. Discussion The limitations of the NHSLA database and our methodology in its analysis have been discussed elsewhere [4]. The database was designed for the financial management of claims and not to store clinical details. Thus the financial data contained are more complete than the clinical information. Of the original anaesthesia data set, approximately 10% of the claims were excluded from analysis as there were insufficient clinical data to categorise them. A further 10% were deemed misclassified, being unrelated to anaesthesia. It is therefore feasible that claims pertinent to alleged anaesthetic drug error were also misclassified and missed in our analysis. Even severe adverse outcomes do not always generate claims. For example, in 12 years of NHSLA records, only one claim alleged fatal transfusion error, yet over the same time period there were 24 UK reports of transfusion incompatibility causal or contributing to death [10]. How many occurred in relation to anaesthesia in England is unknown. Nevertheless, the factors that determine whether an incident leads to litigation remain ill defined. Until 2009, initiating litigation generally precluded participation in the NHS complaints procedure and this may have acted as a deterrent. However, notwithstanding these acknowledged limitations, it may be possible to make some cautious inferences from the claims related to alleged anaesthetic drug error contained in the NHSLA database. We hypothesised that the adverse outcomes contained in the database would be of importance specifically to patients on the basis that they or their families had instigated litigation. In the subgroup analysed in this report, the finding that nearly one third of outcomes were associated with severe harm or death supports this view, and also indicates that events of potentially great clinical significance are contained in the database. The proportion of harmful events reported in this subgroup is much greater than in a similar subgroup reported to the National Reporting and Learning Service (NRLS). Of the 93 NHSLA claims, 32% involved severe harm or death and 62% mild to moderate harm. By contrast, of 1120 NRLS reports involving anaesthesiarelated drug errors made between 2004 and 2006, 1% reported severe harm or death, 20% reported mild or moderate harm and 79% reported no harm [11]. This contrast may reflect different priorities between claimants and clinical reporters. The NRLS reports contain nearmiss incidents in which the patient was unharmed. However, it is a matter for further audit, research and debate whether learning from near-miss drug errors will lead to a meaningful reduction in actual harm to patients. In financial terms, at the time of this investigation, claims of anaesthesia-related alleged drug error in the NHSLA database were the fourth most costly ( 5 million) after regional ( 11 million), obstetric ( 7.5 million) and airway-related claims ( 5.2 million), and represented 16% of the total anaesthesia-related payments. This group of claims is therefore of relative financial importance. Giving due care and consideration to the adverse effects on the claimants (almost a third of the outcomes were fatal or categorised as severe, a higher proportion than in regional and obstetric claims), if the frequency of this type of claim could be reduced in a comparatively cost-efficient manner, it should be given serious consideration on the basis of this cost alone. Human error was apparent in all the claims where categorisation was possible and approximately half of these claims apparently involved drug misidentification. A drug double checking process, such as the one proposed by the NPSA, might theoretically have prevented 44% of the claims in the data set, but it is notable that the total cost of these claims ( ) represented less than 6% of the cost of the whole subgroup. This may suggest that these claims were more easily defended or settled at lower cost. In the allergic reaction subgroup, previously recorded allergens were administered in 58%. These claims might also have been avoided by a double checking process for documented allergies. However, double checking is clearly not infallible. In two claims totalling involving mismatched blood transfusion and the adjustment of an infusion device, a double checking process should have already been in place. Double checking processes do not provide a robust defence against human fallibility, in part because they require care [12]. Eighty-four per cent of the drug errorrelated claims in an ASA Closed Claims analysis were 1320 Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland

5 J. Cranshaw et al. Æ Drug errors in anaesthesia judged as involving less than appropriate care. Indeed, inadequate care may be a specific feature of drug errors as care was considered less than appropriate in only 35% of the rest of the claims in that database [7]. A possible improvement in patient safety through drug checking and administration would be a process that failed safe i.e. a drug could not be administered unless the checking process had been correctly completed. However, it is easy to anticipate that such a system might have unintended consequences. Alternatively, one can accept that it is inevitable that human checking processes will fail and some authorities have suggested that errors should be seen primarily as system failures and not as failures of individuals [13]. Systems additional to double checking may be needed to detect failures as they occur and prevent or mitigate their consequences. The number of anaesthetics administered in England per year is not known. However, one conservative approximation is six million [14]. A similarly modest estimate is that a patient receives five peri-operative drug administrations under the remit of anaesthesia. Therefore, a very conservative estimate is that in the region of 360 million relevant drug administrations may have occurred over the 12 years that this analysis covers. Although the confidence intervals surrounding these estimates may be wide, a cost-benefit analysis before the introduction of any risk reduction strategy is still reasonable. We have identified, on average, only eight NHSLA claims per year that were triggered by anaesthesia-related alleged drug errors. We have also suggested that an effective double checking process might have only prevented fewer than half of these. If such a process were mandatory for 30 million events per year, perhaps four claims might be prevented and payments totalling around avoided. Two hundred and five drug error-related claims were filed between 1980 and 2003 in the ASA Closed Claims database [7]. These claims represented 4% of the total (compared to 11% of 841 NHSLA claims collected over 12 years [4]) and this percentage changed little year on year. The ASA analysis determined that substitution error occurred in 24% (NHSLA data 43%); incorrect dosing in 31% (NHSLA data 37%); insertion error in 17% (NHSLA data 10%); other complex problems in 24% (NHSLA data 5%) and use of an incorrect route of administration in 2% (NHSLA data 3%). The difference in the number of cases categorised as complex problems may reflect the ASA s access to clinical notes and additional evidence unavailable in the NHSLA database. This aside, the proportions are arguably similar and might reflect broadly comparable anaesthetic practices and problems. In 2003, of 205 claims alleging anaesthesia-related drug administration errors in the ASA Closed Claims database, there were 14 claims of awake paralysis [7]. Similarly, these were caused by syringe misidentification or administration of a neuromuscular blocker and induction agent in the wrong order. In 12, the neuromuscular blocker was suxamethonium. Nevertheless, 24% of claims in the ASA analysis had a fatal outcome (NHSLA data 8%) and frequent misidentification of epinephrine ampoules was noted, an error not recorded in any of the NHSLA claims where notably the same drug is termed adrenaline. Thus, despite some similarities it remains uncertain if lessons can be learnt in England or elsewhere in the UK from this type of error analysis in the USA and if similar error prevention strategies would be of equal value in both healthcare systems. Approximately one quarter of alleged drug error claims in the NHSLA database originated in the obstetric environment. Although the proportion of anaesthetics administered in this setting is unknown, the frequency of these claims may be disproportionately high. In keeping with other data [4], the obstetric claims appear higher in volume but lower in cost to the NHSLA, perhaps indicating an intrinsically litigious area of anaesthetic practice. Three quarters involved peri-operative drug administration and more than half involved substitution of suxamethonium resulting in awake paralysis. Drug misidentification may be a particular problem during regional anaesthesia for caesarean section. If the obstetric environment does have a higher drug error rate than other areas, it may be an appropriate location to pilot prevention strategies. Although two claims remained open, only 3% of the total cost of claims (excluding allergy) were incurred peripartum. However, of claims involving awake paralysis, 67% were peripartum. More than one third of all drug error claims alleged drug overdose and two thirds of these claims reported death or severe adverse outcomes. In 64%, an opioid was named, half administered neuraxially and again two thirds reported fatal or severe adverse outcomes. Although iatrogenic opioid overdosing is a recurrent error reported to the NRLS [14], it seems rarely reported in the context of anaesthetic care. Our data cannot help determine the frequency or clinical importance of such errors in these circumstances. However, the financial importance of their adverse outcomes in the database is clear. Although three claims remained open at the time of analysis, the total cost of claims alleging opioid overdose was , 70% of the cost of all claims. This stands in stark contrast to the total cost of claims alleging non-opioid overdose (3% of the cost of all claims) and of claims associated with awake paralysis (4% of the cost of all claims). Two thirds of the claims alleging that a drug error precipitated an allergic reaction originated from patients Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 1321

6 J. Cranshaw et al. Æ Drug errors in anaesthesia Anaesthesia, 2009, 64, pages who allegedly received a drug to which they were known to be allergic. Nine of these 20 patients were given a penicillin despite a history of penicillin allergy. It was not possible with the available clinical details to determine if the antibiotic was administered in error or deliberately despite potential risk. The WHO Surgical Safety Checklist includes a check for known allergy before induction of anaesthesia. However, not all responses to drugs that patients or their carers consider allergic represent immunological reactions although they may still be of clinical importance. The appropriate action in circumstances where a clinician believes that there is a risk rather than a certainty that the patient may be harmed by administration of a penicillin seems open to debate, as is the risk:benefit ratio of selecting another antibiotic [15] or omitting an antibiotic altogether. A published survey has shown that the feelings of the public and the medical profession may be influenced by the severity of an allergic reaction after the administration of a known allergen in error [16]. However, with our limited supportive data, we only conject that the apparent relationship between the payments for claims made after the administration of a recorded allergen and severity of outcome is real. In claims made after anaphylactic reactions to an antigen unknown before administration, we found it impossible to attribute error. When the outcomes were fatal or severe, most claims appear to have been made on this basis alone. That four of these claims had a cost of 0 and that the median was less than 1000 despite death or central nervous system damage may reflect similar judicial disinclination to attribute error in this circumstance. In summary, medication error is acknowledged as a significant but reducible clinical and financial burden to healthcare providers. However, uncovering which errors are the most important in these respects is challenging. Healthcare providers and patient organisations are confronted by a lack of systematically acquired information about drug errors that has the face validity to influence clinical practice [17]. A database is required that reliably records errors, grades consequential adverse outcomes and provides a suitable denominator. Until this exists, it will be difficult to discriminate the relative importance of different types of medication errors and intervene appropriately to reduce harm in a rational and patient-centred manner. In the meantime, identifying adverse outcomes of importance to patients could help prioritise areas for intervention. Analysis of the NHSLA database as a whole suggests alleged anaesthesia-related medication errors, avoidable reactions to allergens and severe allergic reactions are important causes of claims. The database gives some insight into patient-defined adverse outcomes and simultaneously provides a measure of a related financial burden. Awake paralysis, particularly in the obstetric theatre, followed by opioid-induced severe respiratory depression with neurological sequelae, appear important adverse outcomes to patients. However, the cost to the NHSLA of the latter is much greater than the former. Our analysis suggests that robust double checking would have been unlikely to prevent more than half the alleged medication errors in this database. These observations might provide additional data for modelling the cost-benefit of drug error reducing strategies including double checking. They may also lend weight to specific suggestions for further study. Acknowledgements The authors acknowledge and thank Dr R. Mihai and Dr S. Scott for the preliminary sorting of the claims and Ruth Symons and the NHSLA for the original data. References 1 Smith J. Building a Safer NHS for Patients: Improving Medication Safety. Department of Health en/publicationsandstatistics/publications/publicationspolicy AndGuidance/DH_ (accessed ). 2 The World Health Organization. Patient Safety. Launch of Nine Patient Safety Solutions. patientsafety/events/07/02_05_2007/en/index.html (accessed ). 3 NHSLA Annual Report and Accounts. London: HMSO, Accounts2007.pdf (accessed ). 4 Cook TM, Bland L, Mihai R, Scott S. Litigation related to anaesthesia: an analysis of claims against the NHS in England Anaesthesia 2009; 64: National Coordinating Council for Medication Error Reporting and Prevention. About Medication Errors. Taxonomy of Medication Errors. taxo pdf (accessed ). 6 Webb RK, Currie M, Morgan CA, et al. The Australian Incident Monitoring Study: an analysis of 2000 incident reports. Anaesthesia and Intensive Care 1993; 21: Bowdle TA. Drug Administration Errors from the ASA Closed Claims Project. ASA Newsletter 2003; 67: NHS National Patient Safety Agency. Being Open: Communicating Patient Safety Incidents with Patients and their Carers. Appendix A: glossary of terms and list of acronyms and abbreviations GatewayLink.aspx?alId=5594 (accessed ). 9 Health and Safety Executive. Introduction to the Inspectors Human Factors Toolkit. October 2005 (Updated 27 February 2009). kitintro.pdf (accessed ) Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland

7 J. Cranshaw et al. Æ Drug errors in anaesthesia 10 Serious Hazards of Transfusion Report (accessed ). 11 Catchpole K, Bell MDD, Johnson S. Safety in anaesthesia: a study of reported incidents from the UK National Reporting and Learning System. Anaesthesia 2008; 63: Armitage G. Double checking medicines: defence against error or contributory factor? Journal of Evaluation in Clinical Practice 2008; 14: Kohn LT, Corrigan JM, Donaldson MS (eds). To Err Is Human. Building a Safer Healthcare System. Washington DC: National Academy Press, NHS National Patient Safety Agency National Reporting and Learning Service. Rapid Response Report: Reducing Dosing Errors with Opioid Medicines 4th July reducing-dosing-errors-with-opioid-medicines/ (accessed ). 15 Frumin J, Gallagher JC. Allergic cross-sensitivity between penicillin, carbapenem, and monobactam antibiotics: what are the chances? The Annals of Pharmacotherapy 2009; 43: Blendon RJ, DesRoches CM, Brodie M, et al. Patient safety: views of practicing physicians and the public on medical errors. New England Journal of Medicine 2002; 347: Vincent C, Aylin P, Franklin BD, et al. Is health care getting safer? British Medical Journal 2008; 337: Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 1323