Taking stock of anaesthetic databases: enough data, time for action!

For almost 60 years, the paediatric anaesthetic literature has witnessed multiple publications reviewing epidemiological data with regard to anaesthetic practice, including morbidity, mortality and risk. This edition of Anaesthesia contains the most recent of these, from Strøm et al. looking at the characteristics of those under two years of age undergoing anaesthesia 1. However, after six decades where extraordinary improvements in information technology with regard to databases, processing data and the worldwide sharing of information have occurred, should we not be expecting more from such epidemiological surveys than just reaffirming what we already know, and should we be demanding more from our electronic information systems? The earliest of these publications were retrospective studies that produced estimates of risk, but no assessment of risk factors 2-5. The next 30 years saw the publication of five epidemiological studies 6-10, which Patterson and Waterhouse noted that, despite significant differences in the design of these studies, could be considered as a group because the risk profiles produced were nearly identical, (infants are at greater risk of complications, ASA physical status 3 or greater increases risk substantially) 11. These studies are summarised in Table 1. Tiret et al. 6 Prospective, multi-centre 40,240 < 15 years 0.7:1000 overall (1:250 in infants) (1:2000 in children) 3:10,000 (19:10,000 < 1 year) 1,089,200 < 18 years Tay et al. 7 Prospective, single institution 10,000 Age not stated 8.6% infants 2.1% older children Respiratory 77.4% Murat et al. 8 Prospective, paediatric teaching hospital 24,165 < 16 years 31:1000 anaesthesia 48:1000 recovery Respiratory 53% 3:10,000 (11:10,000 < 1 year) No denominator < 18 years APRICOT 16 Multicentre cohort, across Europe 30,874 < 15 years 5.2% Respiratory 3.1% Laryngospasm 1.2% Cardiac 1.9% Tiret et al. published the first prospective multicentre study quoting an anaesthetic mortality of 1:40,000, and with a significantly higher risk of incidents in infants than in children (1:250 vs. 1:2000, p < 0.001) which increased significantly with higher ASA physical status 6. Tay et al. not only replicated these findings but also highlighted that respiratory events were the most frequent (77.4%), with laryngospasm accounting for 35.7% 7, while Murat et al. recorded an incidence of respiratory events of 53% 8. Both acknowledged that their results were potentially affected by the under-reporting that is endemic within a voluntary system. The first study investigating cardiac arrest in children (paediatric peri-operative cardiac arrest (POCA) registry) was published at the beginning of the millennium by Morray et al.; quoting a rate of 1.4:10,000 anaesthetics; over half of these occurred in infants and two-thirds occurred in ASA physical status 3 to 4 patients (accounting for 95% of deaths) 9. Data for the following six years of the POCA registry were published in 2007 by Bhananker et al. and the main difference was a reduction in medication-related arrests, attributed predominantly to a reduction in halothane-induced cardiac arrest 10. Differences in the design of these studies and definitions of incidents might account for discrepancies between studies, and paradoxically, as practice becomes safer, the harder it becomes to demonstrate further improvements, particularly at any one institution 12. Although beset with problems of under-reporting and how incidents were defined, the analysis of critical incidents can help to determine what improvements might be made. An analysis of critical incidents relating to paediatric anaesthesia revealed a wide spectrum of clinical and organisational incidents. Many of these could have been avoided by the application of existing safeguards, for example, pre-operative checking for wrong-site blocks, allergies, fasting times and weight 13. This knowledge needs to be embedded in our organisations and our practice. In 2010, Varughese et al. published their experience of embedding quality improvement into healthcare delivery. By identifying measures of quality for paediatric anaesthesia and adopting quality improvement methodology, they produced a sustained reduction in respiratory complications in the recovery ward by 50% over a six-month period. Central to their success was using quality measures that had been reported in quality assurance databases and clinical trials within paediatric anaesthesia, that is, delivering the effective therapies that already exist 14. In 2014, Kurth et al. published a serious adverse event rate of 1.4:1000 anaesthetics, concluding that education and application of quality improvement methodology in anaesthetic departments, on a national scale, were key strategies to improve peri-operative safety 15. In 2017, the results of the APRICOT study were published which attempted to get a snapshot of anaesthetic practice across the continent of Europe 16. The primary end-point was the occurrence of peri-operative severe critical events requiring immediate intervention (incidence 5.2%). They revealed some evidence to suggest a favourable effect of an experienced anaesthetist and specialist paediatric centre for more complex children (Table 1). In this edition, Strøm et al. present their findings from analysing the anaesthesia practice and peri-operative complications for children in Denmark of less than two years of age 1. They are to be congratulated on being the first group to do this, and for focusing on such a large number of children in an age group who have, throughout the last 60 years, been identified as a higher risk group. This population-based observational cohort study of 17,436 children integrated three national databases to identify children who received anaesthesia during the study period, and to cross-reference them with the national patient registry and a civil registration system using the unique personal identification number (CPR number) assigned to all Danish citizens at birth. They have also been ambitious in trying to describe the broader anaesthetic management of these children, current practice with regard to anaesthetic airway management and the procedures for which children require anaesthesia. The methodology of their study is robust and the use of national databases in anaesthesia is novel. There were some changes to the data collected within the databases during the study period which has meant that some datasets are incomplete. The authors acknowledge the inherent limitations in database-driven research and that those children receiving anaesthesia in institutions that do not report into the database are unknown, and may be a source of selection bias. Overall, the collection of data was very robust. However, there was no information regarding whether or not a peri-operative adverse event had occurred in 28% of cases and this may be a source of error. The problem of missing data affects almost all studies and how to handle the missing data requires careful thought. If the data are ‘missing not at random’ then it is better not to perform any analysis which will produce erroneous results. If the data are ‘missing at random’ then multiple imputation (estimating) of data may be used 17. The results describe anaesthetic practice in children under two years of age as occurring predominantly in University hospitals (79% of cases) and being conducted by specialist paediatric anaesthetists on the majority of occasions (85%, 95%CI 84.6–85.5%). Non-surgical indications for general anaesthesia, including diagnostic radiology, account for nearly one-third of cases. There was still some occasional paediatric anaesthetic practice, with seven institutions performing less than four cases per month. There are broad similarities to UK practice, although the use of regional anaesthesia in the study group is less than in other surveys 18, but, as in Europe and North America 18, 19, caudal anaesthesia was the commonest central neuraxial block. However, some dissimilarities are apparent, for example, that infraclavicular, popliteal sciatic and femoral nerve blocks were the commonest peripheral nerve blocks and that peripheral nerve blocks were used in 23.1% vs. 35% in a North American survey 19. This may represent a cultural difference in regional anaesthesia and trauma, or differences in surgical technique. In terms of its broader international and European relevance, this study compares well in terms of size to the APRICOT study (17,436 less than two years of age vs. 3273 less than one year of age, and 13,463 between one and 5 years of age in the APRICOT study). Also, the critical incident rate was higher in the APRICOT study, (5.2% vs. 1.7%) but the authors acknowledge that the significant amount of missing data and potential under-reporting may have underestimated the prevalence. With regard to airway management, the commonest technique used to secure the airway was tracheal intubation, although the use of supraglottic airway devices increased over the study period. More than 50% of the peri-operative complications were related to complications with airway management (284 cases), with a difficulty in intubation incidence of 1% (272 cases). This is similar to the rate found by Fiadjoe et al. 20, but they found fewer airway complications (1.3% vs. 20%), and Fiadjoe's study included more airway specific complications, for example, minor airway trauma and laryngospasm, which were not included in the Danish national database. They state that their incidence of serious adverse events (0.13%) is similar to previous studies, citing the paper by Kurth et al. (0.14%) 15. However, different definitions were used in comparison with previous studies, and with regard to those used by Kurth and colleagues they were very much narrower. What can we expect in the future? We have seen that over the last 60 years many studies that have produced similar results. These have undoubtedly led to improvements in safety, but, if we want the incidence of mortality and critical events to continue to fall, even greater numbers are required to power these studies. In another 60 years' time will we still be seeing the same studies, showing similar results and with the same criticisms of design and data capture? Although many of the findings in Strøm et al.'s study are in accordance with previous research and are not new, what is novel is that they have demonstrated that the use of large national databases can provide a comprehensive data collection set, albeit within the legal frameworks safeguarding personal liberty, security, confidentiality and information sharing. It is interesting to note that in the Report of the Children and Young People's Health Outcomes Forum (CYPHOF) 21, it was suggested that the NHS Number should be used as a unique identifier to bring together health, education and social care data for all children and young people. This would create a way of linking information, on health interventions and their impact and stop databases being passive repositories for information. If there is to be new data, the template of large national databases which are related, used by Strøm et al. should be considered, but with agreed standardised principles, questions and definitions that allow for meaningful comparison and detection of rare, but clinically important events. Kurth et al. have shown how quality improvement can produce results, by delivering the effective therapies that already exist 14, and the anaesthetic literature contains many examples of possible candidate interventions 22-25. What is now required is for the worlds of quality improvement, risk management and health informatics to come together to facilitate continued improvements in practice and safety. The peri-operative risk factors, and the effective therapies that we already know, need to be integrated into electronic healthcare records to provide guidelines for best practice, and clinical 'nudges' to change behaviour, to ensure best practice is followed, and for these records to allow real-time data capture. Innovation on this scale will not only influence training and practice but also has the potential to influence health policy, and will therefore require strong leadership to influence policymakers to provide the significant investment this 21st century standard will require. We have sufficient data already, perhaps now we will be able to convert this information into knowledge that will at last be retained in the institutional memories of our organisations and that can serve our patients well. No external funding or competing interests declared.