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

Transcription

1 Anaesthesia, 2009, 64, pages doi: /j x Litigation related to anaesthesia: an analysis of claims against the NHS in England T. M. Cook, 1 L. Bland, 2 R. Mihai 3 and S. Scott 3 1 Consultant Anaesthetist, Royal United Bath Hospital, Bath, UK 2 Medical Student, University of Bristol, Bristol, UK 3 Consultant Anaesthetist, John Radcliffe Hospital, Oxford, UK Summary The distribution of medico-legal claims in English anaesthetic practice is unreported. We studied National Health Service Litigation Authority claims related to anaesthesia since All claims were reviewed by three clinicians and variously categorised, including by type of incident, claimed outcome and cost. Anaesthesia-related claims account for 2.5% of all claims and 2.4% of the value of all claims. Of 841 relevant claims 366 (44%) were related to regional anaesthesia, 245 (29%) obstetric anaesthesia, 164 (20%) inadequate anaesthesia, 95 (11%) dental damage, 71 (8%) airway (excluding dental damage), 63 (7%) drug related (excluding allergy), 31 (4%) drug allergy related, 31 (4%) positioning, 29 (3%) respiratory, 26 (3%) consent, 21 (2%) central venous cannulation and 18 (2%) peripheral venous cannulation. Defining which cases are, from a medico-legal viewpoint, high risk is uncertain, but the clinical categories with the largest number of claims were regional anaesthesia, obstetric anaesthesia, inadequate anaesthesia, dental damage and airway, those with the highest overall cost were regional anaesthesia, obstetric anaesthesia, and airway and those with the highest mean cost per closed claim were respiratory, central venous cannulation and drug error excluding allergy. The data currently available have limitations but offer useful information. A closed claims analysis similar to that in the USA would improve the clinical usefulness of analysis.... Correspondence to: Dr Tim Cook timcook007@googl .com Accepted: 20 January 2009 It is recognised that medical care is associated with patient harm and iatrogenic injury. In 1991 Brennan [1] in America reported avoidable patient harm in 4% of hospitalisations, while in Australia in 1995 Wilson [2] reported adverse events in 17%. In the UK, in 2001, Vincent [3] reported iatrogenic harm in 11% of hospital admissions and in 2007 Sari [4] reported a 9% rate. Up to half of all these incidents are deemed avoidable and many are associated with substandard or negligent care: around 1 in 10 contributes to patient death [1 4]. Patient harm may lead to medico-legal claims against doctors and organisations. In anaesthesia the most robust data on such claims comes from the American Society of Anesthesiologists Closed Claims Project (ASACCP), which has been running for more than 20 years. More recently Japanese [5] and Danish [6] authors have reported on the scope and costs of their anaesthesia-related complications and medico-legal claims. Although data from one country may usefully inform practice in other countries, the most relevant data are likely to be country-specific. There are little equivalent data published from the countries of the UK, and no closed claims analysis system exists for anaesthesia, although specialty-specific critical incident analyses have been published [7, 8]. A recent publication examined more than anaesthesia-related incidents reported to the National Reporting and Learning Service (NRLS) [9], but interpretation and learning from this dataset is hampered by a number of factors which include the enormous number of incidents reported, the fact that most incidents are not reported by anaesthetists, the high proportion of incidents with either low or no patient impact, and a lack of outcome data. The National Health Service Litigation Authority (NHSLA) was created in 1995 and manages legal claims (both clinical and non-clinical) made against the NHS (NHS Hospital Trusts, Foundation Trusts and Primary Care Trusts) through its risk-pooling schemes. The current clinical scheme is the Clinical Negligence Scheme 706 Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland

2 Anaesthesia, 2009, 64, pages T. M. Cook et al. Æ Litigation related to anaesthesia in the English NHS for Trusts (CNST) which also started in This is a voluntary membership scheme covering clinical claims against the NHS in England. The costs of meeting these claims are met through members contributions. All English NHS Hospital Trusts and Primary Care Trusts are members of the CNST [10]. Between 1995 and 2002 small claims were managed by Trusts and not by the NHSLA, but since 2002 all cases, whatever the size of the claim, have been managed by the NHSLA. Approximately 5000 clinical claims are notified each year. The NHSLA also receives approximately 3000 non-clinical claims each year which are managed outwith the CNST scheme. The CNST has a mandate to settle claims promptly with a current average of 1.46 years from the date of notification to the date when compensation is agreed or the claimant discontinues their claim [11]. The cost borne by the NHSLA includes both damages paid to patients and the legal costs incurred on both sides, where these are met by the NHSLA. In the last three financial years the numbers of CNST claims have been stable ( claims). Claims-related cost has risen from 329 million in to 424 million in (a 29% increase). Annually the cost of the CNST scheme is approximately 60% of the cost of all NHSLA claims. This study analyses data contained in the NHSLA CNST claims reports and its dataset, between 1995 and The report is an overview of the anaesthesia-related claims made during that period. The primary objective of the report is to indicate the financial impact of such claims in English NHS anaesthetic practice. We consider that identification of areas of high medico-legal risk is useful for informing anaesthetic practice, for development of risk reduction strategies and to identify potential areas for future risk management or study of risk reduction. Specific clinical domains (such as obstetric practice, regional anaesthesia) and complications (such as inadequate anaesthesia, drug errors), will be considered in other reports. Methods In May 2007 the NHSLA was contacted and a request for data was filed via their Freedom of Information link which allows access to anonymised data. The initial request was for data relating to epidural and spinal blockade and relating to airway management during anaesthesia. However, due to difficulty in providing that data it was mutually agreed that all data related to anaesthesia would be submitted. The data were delivered in the form of a spreadsheet listing all claims notified to the NHSLA CNST filed under anaesthesia between 1995 and The data included the following fields: financial year of clinical incident, financial year of claim, whether the claim remained open or closed, a brief description of the details of the claim, the cost to the NHS of the claim, the severity of the event in terms of patient outcome, classifications of the nature of the event and the site of the patient incident (for example operating theatre or labour ward). It is important to appreciate that the NHSLA dataset is for claims (financial) management. It is not a clinical or risk dataset and as such the clinical data are very limited. It is also important to note that any claim notified to the NHSLA by an NHS Trust leads to inclusion in the dataset. Inclusion does not infer that the clinical details of the claim are true, whether the claim has been settled or not. In many cases clinical incident detail was inadequate or truncated. More information on this dataset was requested from the NHSLA but this was not available. The data are limited in several respects. Firstly, the clinical incident data describe the claim. This may or may not reflect actual events. Secondly, the classification of type of incident performed by NHSLA for claim purposes was found to be of little use in determining the clinical nature of the event. Third, the cost associated with a closed claim include both legal fees (defence and claimant costs) and the cost of any settlement, but exclude the cost of the NHSLA itself. The dataset does not contain information on which claims were defended (whether successfully or not) or whether a settlement (either out of court or as a result of court direction) was made, although this information will be available for future enquiries. Some further general information on this subject was however available from the National Patient Safety Agency (NPSA) website. The full dataset was examined independently by three investigators, all consultant anaesthetists. Each assessor used the same methodology to examine the details of each clinical incident and used this and other information, such as outcome data and site of event, to classify cases. Extrapolation of data was forbidden; where data were not adequate to allow classification without extrapolation the case was excluded from analysis. Each clinical incident was classified by clinical category and severity. Incidents that were misclassified, contained inadequate detail or were unrelated to anaesthesia (related solely to care in an outpatient pain clinic or intensive care unit) were excluded. Initial examination of the data led to development of a pragmatic clinical classification designed to group together claims of similar clinical origins and enable further clinically relevant analysis of these groups of claims. Clinical categories were obstetric anaesthesia, regional anaesthesia, inadequate anaesthesia, drug related excluding allergy, drug allergy related, central venous cannulation, peripheral venous access, consent, positioning including falls, and miscellaneous. Claims Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 707

3 T. M. Cook et al. Æ Litigation related to anaesthesia in the English NHS Anaesthesia, 2009, 64, pages Table 1 National Patient Safety Agency severity of outcome scale for patient safety incidents [12]. Severity grade None Low Moderate Severe Death Description No harm (whether lack of harm was due to prevention or not) Minimal harm necessitating extra observation or minor treatment* Significant, but not permanent harm, or moderate increase in treatment Permanent harm due to the incident Death due to the incident *First aid, additional therapy or additional medication; excludes extra stay in hospital, return to surgery or readmission. Return to surgery, unplanned re-admission, prolonged episode of care as in or out patient or transfer to another area such as intensive care. Permanent reduction of bodily functions, sensory, motor, physiological or intellectual. were classified as follows: (i) year of claim; (ii) closed or open; (iii) clinical category; and (iv) severity of outcome. Severity of outcome was classified according to the NPSA tool for grading severity of patient incidents (Table 1) [12]. The cost of an incident was not considered when assessing severity of outcome. The severity scale has five defined categories: none, low, moderate, severe and death. Because of the incomplete clinical data presented we added two intermediate categories low moderate and moderate severe. As an example, a case of intraoperative awareness would be classified as low if there were no subsequent psychiatric or post-traumatic sequelae or moderate if there were; in the absence of such details it was classified as low-moderate. Inadequate anaesthesia was sub-classified and included three groups: inadequate general anaesthesia (awareness), inadvertent brief paralysis (for example an error in drug order during a rapid sequence induction such that succinylcholine was administered before a hypnotic drug rather than after it), and inadequate regional blockade. The completed assessments of the three reviewers were combined. Where all three agreed in their assessment, this was accepted. Where only two reviewers agreed (or all disagreed) the case was reviewed further before determining classification. If agreement could not be reached a fourth reviewer was available for consultation, but this was not necessary. There were no unresolved cases. Subsequently, a larger group of investigators was involved in the project to review individual clinical topics. Each of these seven additional investigators also reviewed the dataset and several modifications were suggested. This was an iterative process over several months; the final dataset, from which all investigators then worked, was agreed in April The final dataset was analysed quantitatively and examined qualitatively. The quantitative analysis was to determine the cost associated with particular types of claims. The qualitative assessment was performed in an attempt to identify themes from which learning points might be derived, with the ultimate aim of identifying areas of clinical practice that might be considered of high medicolegal risk. This report describes the quantitative analysis of the final dataset. Additional data contained in NHSLA factsheets [11] and 2007 annual report [10] are presented to provide context for the anaesthesia data. Results General data The NHSLA website states that, since its inception in 1995 until March 2007, the CNST has managed claims of which 1001 are classified under anaesthesia, accounting for 2.5% (one in 40) of all claims. Numerically the number of claims from surgery (15 627; 39% of all claims) and obstetrics gynaecology (8532; 21% of all claims) are substantially greater (Fig. 1). During the same period the cost associated with the NHSLA CNST scheme has been 5.06 billion. The accumulated cost of anaesthesia-related claims is 121 million and represents 2.4% of all claim-related cost (Fig. 2). In the distribution of NHSLA cost as a proportion of overall settlements were: defence legal cost 18.1%, claimant lawyer cost 32.3%, and damages to patient 48.6% [10]. The cost incurred by claimant lawyers is consistently greater than those incurred by the NHSLA lawyers. In addition to settled claims, at the end of March 2007 the NHSLA had unsettled CNST claims with a value exceeding 3.1 billion and a further 1 billion of unsettled claims relating to previous insurance schemes. The NHSLA also makes periodical payments in appropriate cases and in 2007 this liability was 0.66 billion per annum [10]. Therefore the total cumulative cost of claims (both closed and in progress) to the CNST between 1995 and 2007 will likely exceed 10 billion. The outcome of clinical claims notified to the NHSLA in the past 10 years (April 1997 March 2007) are as follows: abandoned by claimant 41%, settled out of court 41%, settled in court 4%, outstanding claims 14%. Cases settled in court include cases where a settlement that was negotiated out of court is required to be ratified in court, which includes all claims relating to minors (children). In the three financial years , and , of the 113 clinical negligence claims litigated in court, 67% were settled in favour of the NHS, 29% in favour of the patient and 4% were settled mid-trial. 708 Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland

4 Anaesthesia, 2009, 64, pages T. M. Cook et al. Æ Litigation related to anaesthesia in the English NHS Figure 1 Total number of claims to the CNST scheme April 1995 March 2007, classified by clinical specialty. Anaesthesia represents 2.5% of claims O&G Surgery Medicine Emergency medicine Anaesthesia Psychiatry/mental health Pathology Radiology Ambulance Paramedical/support services Public health Nursing General practice Figure 2 Total costs to CNST April 1995 March 2007 of closed cases, classified by clinical specialty. Figures are expressed in millions of Pounds. Note: prior to 2002 small claims were managed locally by Trusts without CNST involvement O&G Surgery Medicine Emergency medicine Anaesthesia Psychiatry/mental health Pathology Radiology Ambulance Paramedical/support services Public health Nursing General practice Anaesthesia-related data The dataset contained 1067 anaesthesia-related reports at the time of analysis. The dataset included cases reported to the NHSLA between 1995 and March In total 226 cases were excluded from the final dataset for the following reasons: clearly misclassified (non-anaesthetic, 103 cases), containing inadequate information to either determine whether they were anaesthesia-related or to assess claim (93), an incident related purely to intensive care (13) or to management in a pain clinic (17). Therefore, 841 claims were reviewed. The 841 cases were classified into clinical categories. As claims may relate to more than one category (such as obstetric anaesthesia and regional anaesthesia) totals exceed 100%. Classifications were as follows: regional anaesthesia 366 cases (44%), obstetric anaesthesia 245 (29%), inadequate anaesthesia 161 (19%), dental damage 95 (11%), airway excluding dental damage 71 (8%), Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 709

5 T. M. Cook et al. Æ Litigation related to anaesthesia in the English NHS Anaesthesia, 2009, 64, pages Total cases n (%) Claims closed (%) Claims relating to severe outcome (%) Claims relating to fatal outcome (%) Table 2 Clinical classification of claims and severity of claimed outcome. Regional anaesthesia 366 (44) (18) 8 (2) Obstetric anaesthesia 245 (29) (10) 3 (1) Inadequate anaesthesia 161 (19) 74 1 (1) 0 (0) Dental damage 95 (11) 85 0 (0) 0 (0) Airway 71 (8) (30) 30 (42) Drug related excluding allergy 62 (7) (16) 5 (8) Drug allergy related 31 (4) (32) 5 (16) Positioning 31 (4) 61 1 (3) 0 (0) Respiratory 29 (3) (41) 7 (24) Consent 26 (3) 73 0 (0) 0 (0) Central venous cannulation 21 (2) 76 7 (33) 9 (43) Peripheral venous cannulation 18 (2) 89 1 (6) 1 (6) Wrong side 3 (0.4) 67 0 (0) 0 (0) Miscellaneous 58 (7) (57) 0 (0) Table 3 Cost of closed claims by severity. Costs for 34 claims where severity could not be classified were not determined. Number of claims Percentage of claims leading to cost Total sum paid ( 000) Maximum cost ( 000) > 0.5 million > 1 million Mean ( 000) [median, IQR] Mild [0.5, 0 8] Mild moderate [9, ] Moderate [1 12, 12 41] Moderate severe [1 5, 5 21] Severe [19, 0 57] Death [53, ] Not classified 24 All cases [6, 0 32] drug-related excluding allergy 62 (7%), drug allergy related 31 (4%), positioning 31 (4%), other respiratory 29 (3%), consent 26 (3%), central venous cannulation 21 (2%), peripheral venous cannulation 18 (2%), wrong side 3 (0.4%) and miscellaneous 58 (7%) (Table 2). The severity of claimed injury was as follows: death 10.5%, severe 18.9%, moderate severe 1.5%, moderate 18.0%, mild moderate 17.4%, mild 31.7%, unclassifiable 2.0% the proportions for closed claims are listed in Table 3. The areas with the highest proportion of claims relating to fatal and severe outcomes were central venous catheterisation (43% death, 76% serious injury or death), airway (42%, 72%), respiratory (24%, 66%) and drug allergy related (16%, 48%) (Table 2). Overall 78% of claims were closed. Seventy-three percent of all closed claims led to a financial cost, and within clinical categories this ranged from 42% to 100%. The categories with the highest proportions of closed cases leading to cost were wrong side (100%), drugrelated excluding allergy (94%), inadequate anaesthesia (89%), and drug allergy related (88%) (Table 4). The overall cost of the closed claims in 841 patients was The claims with the highest overall values were regional anaesthesia ( 11.0 million), obstetric anaesthesia ( 7.5 million), airway ( 5.2 million) and drug related excluding allergy ( 4.3 million). The mean and median value of closed claims were and 6050, respectively. The claims with highest mean values per closed case were respiratory ( ), central venous cannulation ( ), drug-related excluding allergy ( ) and airway ( ). The categories with highest median cost per closed case were central venous cannulation ( ), airway ( ), respiratory ( ) and inadequate anaesthesia ( ) (Table 4). Severity of outcome was assessed on the limited clinical data in the dataset, which in turn was a statement of what was claimed to have happened. We did not formally explore the relationship between severity of claim and cost. In general terms cost per case, maximum cost and the number of high cost claims rose with severity (Table 3). Claims relating to outcomes rated no worse than moderate contributed 28% of overall cost. The most reliable outcome in the dataset is likely to be death. The 710 Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland

6 Anaesthesia, 2009, 64, pages T. M. Cook et al. Æ Litigation related to anaesthesia in the English NHS Table 4 Cost of claims by clinical classification. Clinical classification Number of cases Percentage of cases closed Percentage of closed claims leading to cost Total sum paid ( 000) Maximum cost ( 000) Mean [median, interquartile range] ( 000) Cases with cost > 0.5 million Regional anaesthesia [4, 0 26 ] 4 Obstetric anaesthesia [9, 0 29] 2 Inadequate anaesthesia [16, 6 33] 0 Dental damage [0.1, 0 2] 0 Airway [27, 3 83] 2 Drug related excluding allergy [13, 4 42] 2 Drug allergy related [4, 2 14] 0 Positioning [8, ] 0 Respiratory [18, 0 48] 3 Consent [0, 0 9] 0 Central venous cannulation [30, 0 93] 1 Peripheral venous cannulation [3, 0 47] 0 Wrong side [1, 1 2] 0 Miscellaneous [19, 1 129] 3 Claims Financial year Figure 3 Claims by year of event ( ) and by year of claim ( ). Note: until 2002 small claims were not notified to the NHSLA. maximum cost of a claim relating to a severe outcome was 60% more than the highest cost of a fatal outcome. The mean cost of claims relating to severe outcome was 35% more per case than those with fatal outcomes and the overall cost of those with severe outcomes was more than twice the cost of claims relating to death (Table 3). It is likely, therefore, that the cost of fatal outcomes is in general less than those with a severe outcome. Figure 3 shows claims per incident year. The lag time between clinical event and claim has a median of 1 year but 8% of claims were notified to the NHSLA at least 4 years after the event. In addition, the time from claim to closure, although decreasing, is between 1 and 2 years [9]. Finally inflation needs to be included when trying to compare the cost of historic claims to current claims. The UK retail price index (RPI) between 1995 and 2006 ranged from 1.7% (2002) to 3.5% (1995) [13]. Table 5 shows trends in claims by year, both in number and value. All cost is adjusted to reflect 2006 values. Table 5 omits the financial year as our data were guillotined before the end of that year. As we do not know years of settlement for closed claims we have used the year of claim for such calculations; this may produce a little inaccuracy but is likely to affect all data similarly. The number of claims notified to the NHSLA in the three financial biennia (that is, periods of two financial years) after the biennium covering are recorded in Table 5: and show an increase by 40%, 40% and 30%. Table 6 describes the most frequent clinical scenarios seen in claims and Table 7 describes the cases with a RPIadjusted cost exceeding 1 million. Finally, Table 8 Table 5 Trends by year of event, costs adjusted for RPI. Year Number of claims Percentage claims closed Percentage closed cases costing Total sum paid ( 000) Maximum cost ( 000) > 0.5 million > 1 million Mean ( 000) [median, IQR] [43, 2 151] [33, 2 71] [16, ] [4, 0 16 ] [0.5, 0 7] [0, 0 0.4] Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 711

7 T. M. Cook et al. Æ Litigation related to anaesthesia in the English NHS Anaesthesia, 2009, 64, pages Table 6 The most frequent clinical events described in each clinical category. Category compares the distribution of clinical categories of the current data set, and those of a closed claims analysis (the American Society of Anesthesiologists Closed Claims Project) and a clinical incident reporting dataset (the Australian Incident Monitoring System). Discussion Main clinical event Regional anaesthesia Nerve injury inadequate blockade epidural-related problems Obstetric anaesthesia Inadequate regional anaesthesia during caesarean section inadequate general anaesthesia during caesarean section Inadequate anaesthesia Inadequate general anaesthesia inadequate central neuraxial block brief paralysis due to drug order errors obstetrics Airway Tracheal tube soft tissue injury aspiration hypoxia Other respiratory Hypoxia pneumothorax equipment problems Central venous Vascular injury carotid puncture wire cannulation Drug related excluding Drug switches overdose muscle allergy relaxant Drug allergy related Administration of known allergen Positioning Nerve injury while insensate during general and regional anaesthesia Table 7 Closed claims with cost of more than 1 million (adjusted to 2006 values). Year of claim Cost in million Clinical claim Brain damage during cataract surgery Brain damage during general anaesthesia, intubation problem Poorly managed anaesthesia, poor resuscitation, hypoxic brain injury and PVS Neurological damage during obstetric spinal anaesthesia Spinal haematoma in relation to epidural anaesthesia for bowel surgery, paraplegia Cardiac arrest following morphine in anaesthesia, brain damage It has previously been stated that anaesthetists are involved in the care of more than 60% of acute hospital admissions [34]. In the last decade anaesthesia in England has accounted for approximately one in 40 claims notified to the NHSLA and 1 40th of the cost associated with Table 8 Relative frequency of incident types in the first 2000 AIMS incidents, the ASA Closed Claims Project and this report. The three datasets reflect reports from different time periods (see text). Complication NHSLA data ASACCP [14 22] (closed claims) AIMS [23 33] (critical incidents) Ventilation problems < 3% 13% 16% Difficult airway < 8% 5.6% 4% Oesophageal intubation < 0.7% 6.1% 1.7% Regional anaesthesia 44% 20% 8% Awareness 9.5% 1.9% 0.8% Obstetrics 29% 12.3% 4% Central venous access 3% 1.7% 0.9% Drug errors 7% 4% 7.2% Anaphylaxis 4% 3% 2.8% Dental damage 11% Excluded 0.7% Mortality 10.5% 36% 1.5 Morbidity 29%* 64% 6.3* * Morbidity in the NHSLA and AIMS datasets correlates with permanent harm or death (correlating with severe or death in the classifications used in this report). In the ASACCP dataset it is less clearly defined. claims against NHS Trusts. This is considerably less than our closest colleagues, in surgery and obstetrics gynaecology who, when combined, account for 60% of claims and 70% of the cost of those claims. Although the possibility of anaesthesia-related claims being misfiled under surgery cannot be excluded, on the basis of the data available anaesthesia must be considered a medico-legally low-risk specialty in England. Before considering the results of our analysis further we should discuss some limitations of both the data and the study. The data lack clinical detail because they are prepared for claims (financial) management, not clinical (risk) analysis [35]. Weaknesses include lack of: (i) patient demographics; (ii) full claim details; (iii) data describing actual clinical details and outcome (as opposed to clinical claim details); (iv) date and type of settlement or closure; and (v) breakdown of cost of claim into legal fees and patient damages. We considered that NHSLA clinical coding and outcome severity coding were not always accurate. Root cause analysis is not possible. The data are, therefore, far from perfect: it may be that in the future there will be a better system for examining data on adverse anaesthesia outcomes in the UK; we are not aware of one at present. The study also has weaknesses as the data required some interpretation. Interpretation of retrospective data may be problematic; reviewers often differ in their interpretation of the same data [36] and there is evidence that the outcome of an event influences peer reviewers opinion [37 39]. All our primary reviewers were blinded to others opinion and all were 712 Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland

8 Anaesthesia, 2009, 64, pages T. M. Cook et al. Æ Litigation related to anaesthesia in the English NHS instructed to ignore details of monetary cost of cases when classifying them. We erred on the side of eliminating those cases where significant interpretation was required; as a result we excluded more than 20% of cases ( ). We have not made any attempt to determine whether care was appropriate, whether claims were justified or whether cost to the NHS was reasonable. In other closed claims analyses there is evidence that the frequency of payment is linked to appropriateness of care, while the extent of damages is related to both quality of care and severity of injury [40], but our data do not allow us to judge this. More robust analysis and conclusions could be made if a genuine closed claims analysis was established in the UK, which would include detailed study of both the claim and the actual clinical events. Despite these limitations we believe the current data are useful to practicing anaesthetists in a number of ways. The overall cost of the closed claims in the dataset of 841 patients was Seventy-three percent of all claims lead to a financial cost. The mean value for each closed claim was approximately As 22% of claims remain open and our interpretation is that more costly claims appear to take longer to close (see below), these figures for cost per case may be underestimates. We have confirmed that the cost associated with a claim relating to a patient death is generally less than that relating to a severe (permanent) injury. However, it is notable that our data relate purely to financial outcome and must be balanced against the increasing prevalence of convictions for manslaughter amongst doctors (particularly anaesthetists) as noted by Ferner [41]. The data highlight the extent, case-mix and financial risk of anaesthesia-related medico-legal cases in the NHS in England, since This is potentially useful for organisations such as statutory bodies, professional bodies, and defence organisations which may all be responsible for advising the profession on risk avoidance and medicolegal safe practice. The data are also potentially useful to Trusts seeking to manage medico-legal exposure. Second, it is of importance to practicing anaesthetists who may reflect on those areas of their own practice and determine where risks both to their patients and themselves may lie. This analysis may also be of interest to doctors and others involved in medico-legal practice. Anaesthetists may interpret areas of high medico-legal risk to be those with a high number of claims, high overall cost or high cost per closed claim. We have determined that the areas with the largest number of claims are: (i) regional anaesthesia; (ii) obstetric anaesthesia; (iii) inadequate anaesthesia; (iv) dental damage; and (v) airway. The areas with the highest overall cost are: (i) regional anaesthesia; (ii) obstetric anaesthesia; (iii) airway management; (iv) drug error excluding allergy; and (v) respiratory. The areas with the highest mean cost per closed claim are: (i) respiratory; (ii) central venous cannulation; (iii) drug error excluding allergy; (iv) airway; and (v) regional anaesthesia. The data should not be interpreted as indicating the frequency of clinical complications, as only a small proportion of all clinical incidents or patient injuries lead to litigation [1 4]. We also lack denominator data meaning that high representation in this dataset might be due to high denominator rates or a disproportionate increase in claims for some incidents. We also do not know whether those cases we excluded from analysis had the same distribution of clinical categories as those included. Even accepting these caveats it is notable that both regional anaesthesia (44%) and obstetric anaesthesia (29%) appear to represent a disproportionately high percentage of claims. In this report we do not comment in depth on clinical situations described in the claims. We hope to present these data separately. However it is notable (Table 7) that the clinical situations most frequently described in the incident details are descriptions of readily recognised complications of anaesthetic practice. As the data relate to claims analysis and financial risk they can only indicate areas of significant absolute or relative medico-legal risk. This is distinct from clinical risk and from good medical practice; the two should not be confused. It is difficult to report the data on trends over time robustly. In the three financial biennia after (Table 5) claims increase progressively but in the last two biennia appear to reduce. This is likely to be due to delays in notification, such that notification of events for recent years is not yet complete. Seven years after an event more than 95% of clams are closed, but after 4 years only about half are (data not presented). Similarly on initial inspection of Table 5, there is an apparent trend towards lower overall cost of claims in recent years: in the first 6 years of the scheme 82% of closed cases lead to NHS cost, while in the next 6 years only 59%; overall cost appears to reduce eight-fold over the same period, which is at variance with NHSLA claims in general [10, 11]. This might be due to claims yet to be notified, improved clinical practice, improved defence of cases, decreasing numbers of claims (for example, due to a reduction in availability of legal aid), increasing numbers of low value claims, or other factors. However it is also notable that the cost per closed claim in the last 6 years is also notably lower than for the first 6 years of the scheme; it is possible that low cost claims are settled (or dismissed) more rapidly than high cost claims and this may influence both the proportion of cases leading to NHS cost in a biennium as well as per case and overall cost. This is supported by the fact that in the first 6 years each biennium had at least four Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 713

9 T. M. Cook et al. Æ Litigation related to anaesthesia in the English NHS Anaesthesia, 2009, 64, pages claims costing more than (RPI adjusted) and at least one of more than 1 million, while in the last 6 years there have been no such cases and maximum individual case cost has progressively decreased to only in the most recent biennium. Our cautious conclusion is therefore that trends in payments are difficult to interpret until at least 95% of claims have been settled, which is currently about 6 7 years after an event arises and perhaps 3 4 years after a claim is notified to the NHSLA. How do the data we report compare with other national reports of anaesthesia-related morbidity? Broadly there are two types of datasets for study of patient harm: clinical incident or medico-legal closed claims datasets. Clinical incident datasets generally include, as well as those incidents leading to patient harm, incidents without patient harm such as incipient or near miss incidents. While these are useful and important analyses, particularly for identifying system errors, the case distribution differs (inevitably) from closed claim cases (Table 8). The relationship between clinical incident, patient injury and litigation is complex. In the Harvard Medical Practice studies in the early 1990s the authors studied adverse events from hospital records. They recorded adverse events in 3.7% of hospital admissions and 14% of these lead to death [42]. By cross-checking data from this study with medico-legal claims the same group determined that only 1.5% of those patients experiencing adverse events, caused by apparently negligent care, filed malpractice claims [43]. Further, while there were more than seven times as many negligent incidents as malpractice claims, the majority of the incidents leading to claims were not considered negligent by the authors. In the American state of Utah, Studdert et al. [44] examined medical records and reported that only 3% of patients experiencing negligent adverse events sued. They did record 18 malpractice claims but evidence of negligent practice was found in only four of these, and in more than half of the claims there was no evidence of any adverse event. Low socio-economic status and increasing age reduced the likelihood of litigation. Thus, there is a mismatch between clinical error, negligent error and litigation. More up-to-date data, and data specific to the UK are needed. While both critical incident and closed claim analyses are valuable they are unlikely to represent similar incidents [35, 39, 45]. In Australasia the Australian Incident Monitoring Study (AIMS) [46] was established in 1988 under the auspices of the Australian Patient Safety Foundation. AIMS stated intent is to capture information from a wide variety of sources from near misses to sentinel events (so that) detailed analysis is possible and anaesthetists were invited to report, on an anonymous and voluntary basis, any unintended incident which reduced, or could have reduced, the safety margin for a patient [33]. The data were entered onto an extensive form covering the incident details but also including details of timing, human and system factors, process and mitigating or exacerbating factors; as such its source data are likely to resemble NRLS data rather than NHSLA data, but with considerably increased case detail. The first 2000 reported incidents in AIMS were analysed and reported in considerable detail [33], leading to changes in clinical practice. These 2000 cases were all collected and reported before 1994 so there is no overlap in time between that dataset and the incidents included in our current analysis. The NPSA recently reported that 1% of all incidents caused serious harm and a further 6% were associated with moderate harm [47]. A recent study analysing more than anaesthesia-related clinical incidents notified to the UK NRLS over a 2-year period, also reported that three-quarters of those cases resulted in no harm and only 2% lead to severe harm or death [9]. The authors of that report specifically examined reports associated with epidurals and awareness, but not other clinical categories. The article necessarily focused on process and, therefore, clinical interpretation and learning points were limited; indeed the authors state the lack of detail inherent in generic data fields prohibits translation of these results into robust arguments for immediate change in clinical practice. Perhaps the NHSLA data, despite its limitations, offers a greater opportunity in this respect. Analysis of litigation data has some potential benefits over critical incident analysis and in other ways complements it. In contrast to the data from the NRLS [9], the dataset we report which includes only cases leading to litigation, focuses on patient-perceived major events, which we must presume have often led to harm. Whether or not claims are accurate, justified or successful the process of litigation has personal and financial implications for the profession and its organisations. A weakness of litigation-based analysis is that, although it is known that only a small proportion of patients who suffer harm in hospital then litigate [35], it is not known whether those that do sue are representative of those harmed. It is tempting to speculate that they may not be. Conversely it is likely that incident reporting suffers from bias. In a recent US study it was reported that physicians, statements of intent to report a (hypothetical) clinical error varied by 25% depending on the patient outcome [48]. More notable, however, was that, of those who had experienced such an error, only 50% had reported it, irrespective of outcome. Many doctors also reported lack of awareness of current reporting systems and lack of 714 Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland

10 Anaesthesia, 2009, 64, pages T. M. Cook et al. Æ Litigation related to anaesthesia in the English NHS confidence in reporting methodologies. At present the UK has, in the NRLS, a robust clinical incident reporting system but lacks clinically relevant output from the system. Specialty-specific incident reporting is currently being trialled and it is hoped that this may lead to more focussed and clinically relevant analysis [49]. The reality is that in order to fully evaluate clinical incidents several data sources are required [45]. Cases described in closed claims analyses generally represent greater severity of patient outcome (and financial impact). The current data fall short of being a closed claims analysis, but the analysis is comparable with those from Sweden, Japan and the USA. Table 8 shows that the distribution of clinical categories in the NHSLA is closer to that of the ASACCP than the AIMS data, but still quite dissimilar. Despite differences between the data presented here and that reported in closed claims analyses from other countries, some comparison is worthwhile. Hove et al. [50] examined claims from the Danish Patient Insurance Association, (a national no-fault compensation scheme) over a 9-year period ( ). In Denmark approximately anaesthetic events take place per year. They reported 1256 cases including 24 deaths considered due to anaesthesia, an incidence of approximately 0.06 per anaesthetic cases. The patients who died averaged 46 years old and total compensation was $1.1 million. The deaths were classified as follows: airway 4, ventilation 2, drug or blood errors 4, infusion pump related 4, central venous cannulation 4, regional anaesthesia 4, bleeding 1 and unclear 1. Twenty (83%) of the 24 deaths were assessed as potentially preventable if optimal care was delivered. Eighteen deaths occurred in a theatre complex and these are likely comparable to our data. In contrast to Hove s data, the NHSLA data lacks denominators for all its events. Remarkably, in England it is not known how many anaesthetics are administered per year but estimates of up to 8 million surgical procedures have been made [51] and NHS hospital episodes statistics data [52] report 7.2 million operations performed in However, the proportion of these that involves anaesthesia and whether all anaesthetic episodes are included is not clear. The 3rd National Audit Project of the Royal College of Anaesthetists has now established that approximately central neuraxial blocks are performed in the NHS in the UK each year, and provides approximate denominators for individual procedures and clinical settings [53]. The methodology of the 4th National Audit Project may lead to a clearer estimate of the number of anaesthetics given per year [54]. The 841 cases we reviewed included 76 claims (69 closed) associated with fatal outcomes over a 10-year period. In addition the cases which we excluded included 31 additional deaths. Therefore the cohort of 1067 cases includes 107 deaths. This is approximately five times the number of deaths reported in Denmark but lack of a denominator for English or UK practice prevents comparison of likely incidence of such events. Thirty claims associated with fatal outcome are classified under airway, central venous cannulation 9, other respiratory 7, drug errors 5 and allergy related 5. It is likely that there are clinical and medico-legal lessons to learn from some of these cases. The current UK system inhibits learning and disseminating these lessons. In Japan, Irita et al. [55] studying over 1.4 million elective anaesthetics in ASA 1 patients reported that anaesthesia management was the cause of 4.4% of peri-operative deaths (16-fold fewer than surgical management). Cardiac arrest and deaths attributable to anaesthesia were 1.87 and 0.14 per cases, respectively. The commonest causes were medication problems (48%) and airway ventilation problems (57%). In total more than 70% of critical incidents and deaths were caused by human (surgical or anaesthetic) factors. The same group also showed markedly increased critical incident, cardiac arrest and mortality rates with increasing ASA grades (up to 20-fold) and in emergency cases (up to 3-fold) [5]. In Holland, Arbous et al. [56] reporting on peri-operative mortality after anaesthetics between 1995 and 1997 found that 15% of approximately 800 deaths were anaesthesia-related. Finally, the most well established and widely reported closed claim system is the American Society of Anesthesiologist s Closed Claims Project (ASACCP) [57] which has run since In the two and a half decades since its establishment over 7300 cases have been reviewed. The project has limitations. Firstly, insurers for fewer than 50% of anaesthesiologists are included in the project. Secondly, the project suffers from the same weaknesses as other litigation-based databases due to the mismatch between patient safety incidents and episodes of litigation. Thirdly, the delay between an event occurring and the case being closed, as well as the delay in claims analysis and reporting means that a delay of 8 10 years may arise before publication of the project findings. Notwithstanding these limitations the ASACCP also has considerable strengths. The reviewers have full access to patient claim details: Typically, a closed claim file consists of the hospital record, the anaesthesia record, narrative statements of the involved healthcare personnel, expert and peer reviews, deposition summaries, outcome reports, and the cost of settlement or jury awards [21]. Patient claim and case notes are used to guide a formal case review involving a semi-systematic evaluation and clinician interview; the project now has a wealth of experience in reviewing cases. Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 715

11 T. M. Cook et al. Æ Litigation related to anaesthesia in the English NHS Anaesthesia, 2009, 64, pages The size and detail of the dataset adds strength to the analysis of the cases, for example the ASACCP was able to perform detailed analysis of over 370 respiratory cases [14] and more than 170 cases of difficult intubation [58]. The size and thoroughness of the ASACCP have allowed robust conclusions to be drawn that lead to changes in practice over time [21]. In contrast, in analysing the current dataset, our very limited clinical data do not even enable us to determine patients pre-morbid states or an accurate picture of clinical care or outcome. Robust clinical conclusions are likely to be harder to extract. Comparing the clinical categories and severity of outcome in the databases of the NHSLA and ASACCP, it appears that the ASACCP dataset is generally of cases of greater severity (Table 8). The ASACCP reviews approximately 300 cases per year, with a mortality rate in the cases above 30%. In England the number of cases reported to the NHSLA annually is approximately 100 with a mortality rate of close to 10%. Inclusion of cases from the other countries of the United Kingdom would no doubt increase the cohort of cases. Our dataset also contains a higher proportion of claims relating to regional anaesthesia, awareness and obstetrics than the ASACCP. These are important differences and, while useful lessons may be learnt from other countries s closed claims analyses, differences in medical and legal practices are likely to mean that extrapolation from other countries s data should be done with caution. There is ever increasing awareness and expectation of the potential harms caused by medicine and numerous initiatives to minimise such occurrences [59 63]. A co-ordinated closed claims analysis of such UK cases would undoubtedly lead to an insight into local practice and learning points that cannot be gained from analysis of other countries s data. If such a system were to exist it would benefit from extension outside the NHS to the independent sector and would require the co-operation of defence organisations as well as the NHSLA. Such a project has been proposed in the past and perhaps data such as these will provide further stimulus. Finally, there is much current interest in human factors and how these influence patient safety. AIMS considered human factors early in their reports and reported accidents cannot be abolished. However, an understanding of the factors underlying (incidents) can lead to the rational direction of resources and effort to prevent them and minimise their effects [64]. In Arbous s [56] study, human factors accounted for more than twothirds of anaesthesia-related deaths (and organisational factors for around 10%). Atul Gawande, the American surgeon involved with the World Health Organization s Safe Surgery Saves Lives campaign [62] has written that progress in medicine will not be made through improved technology but rather through improved application of current knowledge and activity: in short doing it better [65]. In order to work out how to do things better it is important first to determine where we are doing things poorly. The data presented here offer some insight into areas where significant problems are arising. The lack of detail hampers an analysis of the contribution of system and human factors but it is inevitable that both will exist. In some areas low-tech solutions do indeed suggest themselves. These might include: regional anaesthesia (checklists to improve risk: benefit analysis in its use, improved post-operative surveillance to identify and address sequelae); problems with inadequate anaesthesia, particularly regional anaesthesia (improved pre-operative explanation, attention to detail before surgery, increased alertness to the possibility of inadequate anaesthesia and prompt response to it); drug errors and allergy related errors (improved pre-administration checking systems, perhaps including checklists); consent (improved systems and communication). A significant challenge is to develop and introduce such systems that improve safety without so protocolising the processes that anaesthesia staff are disengaged or demotivated. If poorly constructed or introduced, such systems might harm anaesthesia safety or even hinder recruitment of high quality staff, to the ultimate detriment of patient care and safety. In conclusion, we have classified the existing data held by the NHSLA on medico-legal claims related to anaesthesia. The data provide a considerable insight into the distribution of such claims and is useful in identifying areas of high medico-legal risk. However, the lack of detailed clinical data hampers robust clinical conclusions. The UK does not at present have an anaesthesia closed claims analysis system; we believe that such a system would be clinically and financially beneficial to patients, anaesthetists and the healthcare system itself. Acknowledgements We would like to acknowledge help we received from Ms Ruth Symons and the National Health Service Litigation Authority and from Dr Stuart White (Consultant Anaesthetist, Brighton) for his critical review of an earlier draft of the paper. References 1 Brennan TA, Leape LL, Laird NM, et al. Incidence of adverse events and negligence in hospitalized patients. Results of the Harvard Medical Practice Study I. New England Journal of Medicine 1991; 324: Wilson RMcL, Runciman WB, Gibberd RW, et al. The quality in Australian healthcare study. Medical Journal of Australia 1995; 163: Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland