A prospective survey of intra-operative critical incidents in a teaching hospital in a developing country
Abstract
Critical incident monitoring has the advantage of identifying a potential risk to the patient without it necessarily resulting in morbidity. An added advantage in developing countries is the low cost involved in introducing the programme. This paper analyses the incidents reported from the operating room suite in a teaching hospital in a developing country from August 1997 to 31 December 1999. During the period, 20 819 anaesthetics were administered and 329 incidents were reported (1.58% of the cases). Seventy-three per cent of the incidents were reported in patients of ASA grade 1 or 2. Thirty-nine per cent occurred during induction, 51% during maintenance and 10% during emergence. Human error was the cause in 41%, equipment error in 50% and system error in 8.5%. Twenty-two per cent of the incidents resulted in minor, and 13% in major physiological disturbance. The technique has been found useful in identifying trends and selecting issues to be discussed in departmental quality assurance meetings, but requires persistent motivation of the reporting staff.
Modern practice of anaesthesia involves interaction of highly skilled, multidisciplinary professional staff and sophisticated electronic equipment in ever-changing relationships. Any such system in which there is complexity of action and tightness of coupling between components is likely to produce untoward incidents [1].
Collection of data on critical incidents is increasingly being used in quality assurance programmes in anaesthetic departments in developed countries. A critical incident is defined as any untoward and preventable mishap which is associated with the administration of general or regional anaesthesia, and which leads to, or could have led to, an undesirable patient outcome [2]. The frequency of voluntarily reported incidents from individual Institutions varies from 0.28 to 2.8% in different studies [3, 4]. Although the technique has been studied extensively in developed countries there is still a paucity of data from the developing countries.
Methods
After a policy decision to implement the technique as a quality assurance measure in the department, an appropriate form was developed. The form consisted of a single sheet of paper and was double sided (see Appendix). Information requested included the timing of the incident, system involved, the nature of the event, i.e. airway, pulmonary, circuit, equipment, positioning, or miscellaneous. Enquiry was also made of the contribution of the patient's condition, personal and team factors. Outcome was defined according to the sequelae, as no effect, minor or severe physiological disturbance, morbidity or mortality. A tick box method was used for ease of entry. One-third of a page was left for contextual information. There was no provision for identification of the anaesthetist. It was decided to limit the incident reporting to the main operating room suite of eight rooms for the initial phase.
The form and the concept were initially introduced at a departmental meeting. Anonymous reporting was emphasised and a locked collection box clearly marked ‘Critical Incidents’ was kept in the recovery room. The forms were made available in all operating rooms and were subsequently collected from this box on a monthly basis.
The analysis of the first 3 months of critical incidents was reported in a follow-up departmental meeting and every 2 months thereafter in the departmental audit and critical incident meeting.
Each report was evaluated individually and classified according to the nature of the error. The incidents were divided into three broad categories: human errors, system errors or design/equipment errors, as recommended by Runciman et al. [5].
Human errors were further divided into: knowledge based, i.e. the result of making a wrong plan because of inadequate knowledge or experience or surroundings; rule based, i.e. failure to apply a rule designed to avoid error or decrease adverse outcome; skill based, i.e. associated with inattention, haste, fatigue, stress or illness or technical, i.e. failure to execute a sequence of actions required for a technique.
System errors were defined as those in which clinical practice was badly formulated, and equipment/design errors as those in which equipment or workspace design was flawed. Equipment errors were further subdivided into true failure, in which equipment failed to perform in a specified manner, provided it was adequately maintained, used in an acceptable way and checked prior to use. If the equipment problem was related to poor design, misuse or human error, inadequate maintenance or leaks and misconnection, it was categorised separately. The equipment was also divided into anaesthetic, operating room equipment and monitors.
Results
Three hundred and twenty-nine reports were analysed for 29 months from August 1997 to 31 December 1999. There were 20 819 anaesthetics administered in the main operating room suite during this period. The anaesthesia providers were full-time and part-time credentialled faculty members, anaesthesia residents in a structured training programme, and medical officers. The critical incidents reported represented 1.58% of the cases.
General anaesthesia (GA) was administered in 300 of these cases in which incidents were reported, 16 involved a combined GA and regional technique, in six cases regional anaesthesia was given and in one a combined spinal/epidural technique. One incident occurred during a Bier's block, and one during monitored anaesthesia care. In four cases, the technique was not specified.
The phase of anaesthesia during which the incidents occurred is shown in Table 1. Tables 2–5 show the ASA status, the nature of the errors, the specialty involved and the immediate patient outcome. Eighteen errors (5.47%) were reported in emergency cases.
| Phase | n = 320 | (%) |
|---|---|---|
| Induction | 124 | (38.95) |
| Maintenance | 162 | (50.62) |
| Emergence | 34 | (10.62) |
- Information about the phase of anaesthesia not available in nine of the forms.
| ASA | n | %) |
|---|---|---|
| I | 108 | (34.17) |
| II | 125 | (39.55) |
| III | 65 | (20.56) |
| IV | 18 | (5.69) |
- Information about ASA status was not available on 13 forms.
| Nature of errors | n | (%) |
|---|---|---|
| Human errors | 136 | (41.3) |
| Knowledge based | 24 | (17.6) |
| Skill based | 38 | (27.9) |
| Rule based | 39 | (25.7) |
| Technical | 9 | (6.6) |
| Insufficient contextual details | 30 | (22.0) |
| Equipment error | 165 | (50.1) |
| System error | 28 | (8.5) |
| Specialty | n | (%) |
|---|---|---|
| General surgery | 63 | (19.1) |
| Urology | 12 | (3.6) |
| Obstetrics & gynaecology | 30 | (9.1) |
| Orthopaedics | 30 | (9.1) |
| Paediatric surgery | 24 | (7.3) |
| Neurosurgery | 42 | (12.7) |
| Dental surgery | 11 | (3.3) |
| ENT surgery | 53 | (16.1) |
| Plastic surgery | 10 | (3.0) |
| Vascular surgery | 01 | (0.3) |
| Thoracic surgery | 14 | (4.2) |
| Information not available | 39 | (11.8) |
| Outcome | n | (%) |
|---|---|---|
| No effect and no physiological response | 150 | (45.5) |
| No physiological effect but temporary sequelae | 51 | (15.5) |
| Minor physiological disturbance | 74 | (22.4) |
| Major physiological disturbance | 42 | (12.7) |
| Event resulted in immediate morbidity | 11 | (3.3) |
| Resulted in immediate mortality | 01 | (0.3) |
Equipment errors
From a total of 329 errors, 165 were related to equipment; 78% were related to anaesthetic equipment, 16% to monitoring, 4% to operating room or other equipment and in 2%, details were not available. Further details are shown in Table 6. True failure occurred in 24.2% of these, and 72.7% related to misuse or human error. In 3% of the reported incidents, insufficient contextual information was given to allow further classification.
| Type of equipment | No. of cases reported | (%) |
|---|---|---|
| Anaesthetic equipment | 129 | (39.2) |
| Tracheal tubes | 22 | (17.0) |
| Laryngeal mask airway | 2 | (1.5) |
| Ventilators | 15 | (11.6) |
| Circuits | 33 | (25.5) |
| Central supply | 4 | (3.1) |
| Flow meters | 8 | (6.2) |
| Vaporisers | 16 | (12.4) |
| Intravenous cannulae | 23 | (17.8) |
| Intravenous infusion | 1 | (0.7) |
| Syringes | 2 | (1.5) |
| Gum elastic bougie | 1 | (0.7) |
| Scavenging tube | 1 | (0.7) |
| Epidural catheter | 1 | (0.7) |
| Monitoring equipment | 26 | (7.9) |
| Datex | 6 | (23) |
| Temperature | 3 | (11.5) |
| Arterial transducer | 1 | (3.8) |
| Non-invasive blood pressure | 2 | (7.6) |
| Pulse oximeter | 4 | (15.3) |
| Central venous pressure monitoring | 2 | (7.6) |
| End-tidal CO2 monitor | 7 | (26.9) |
| ECG | 1 | (3.8) |
| Operating room equipment | 7 | (2.1) |
| Operating room table | 1 | (14.2) |
| Tourniquet | 2 | (28.5) |
| Blood warmer | 1 | (14.2) |
| Infusion pumps | 3 | (42.8) |
| Miscellaneous | 1 | (0.3) |
| Chest tube | 1 | (0.3) |
| Details not available | 2 | (0.6) |
Human errors
One hundred and thirty-six human errors were reported during the period, which represented 41.3% of total critical incidents. Twenty-four (17.6%) were classified as knowledge based, 38 (27.9%) as skill based, 35 (25.7%) as rule based and nine (6.6%) as technical. Thirty (22%) contained insufficient contextual information to classify them any further.
If the incidents relating to misuse or human error in those classified as equipment errors (120) were also added to the above group, human errors formed 256 (77.8%) of the total critical incidents.
System errors
Twenty-eight (8.5%) of the critical incidents were classified as system errors. The errors were further classified as drug related, surgical team related, inadequate anaesthetic assistance related and policy related. The term ‘policy related’ referred to errors for which a policy should have existed but was not available or was not followed.
Discussion
The critical incident monitoring was first used in aviation by Flanagan, a psychologist, in 1954 [6] and was modified and applied to anaesthesia by Cooper in 1978 [7]. When studying errors, the first step is to compile a database of incidents with sufficient contextual details to allow subsequent identification of mechanism and contributing factors, which often take a predictable form [8]. The largest database has originated from the Australian Incident Monitoring Study (AIMS) [9]. Several papers have been published which categorise and describe the initial collection of data, but there is still a paucity of data reported from the developing countries. The availability of equipment, difficulty in intra-operative maintenance, lack of adequately trained personnel and less than ideal work environment can all influence the reported data in these countries.
Use of critical incident monitoring as a quality assurance measure has several advantages. It is useful in detecting new problems, identifies near misses which can be instructive for trainees, may reveal clusters of incidents or previously undiagnosed sources of errors, and is economical [5]. It is seen as an objective method, has a potential for changing attitudes [10], negative outcomes are not required, validity is unaffected by advances in anaesthesia, focus is on prevention and all aspects of anaesthetic care can be monitored [11]. It is also seen as non-threatening because information is volunteered. However, the method does have some limitations. Reporting varies according to an individual's perception of an incident, and depends on motivation and universal acceptance that reporting will have a beneficial result [12]. If an incident is not immediately reported, the memory of events will be altered with time, and an anaesthetist's account of a critical incident has been shown to differ significantly from what actually occurred [13].
The economy of the method and the voluntary reporting have certain advantages in developing countries. There is minimal cost involved in initiating and setting up the programme and voluntary reporting makes it attractive in Eastern cultures which are seen as less ‘open’ and hierarchical compared with the West.
The frequency of incidents reported from individual institutions has varied from 0.28% to 2.8% [3, 4]. This percentage may not reflect the actual number of critical incidents occurring within a department, since reporting is voluntary. A previous study has estimated a return rate of 30% [14].
The incidents reported in our study represent 1.58% of the cases undertaken in the main operating area. The majority of incidents (77.8%) were related to human errors. Forty-one per cent were true human errors and 36% were equipment errors related to human errors. This figure is similar to other major studies reported from developed countries [2, 15–17]. Some variation in data in comparison to these studies may have arisen from variations in coverage of anaesthetic services. Our study was limited to the operating room and excluded incidents arising in the waiting area and the recovery room.
More incidents were reported during the induction or maintenance period compared with emergence, and this pattern was consistent with data from previous studies [2].
The critical incidents, when presented to the department, resulted in the establishment of departmental quality assurance issues meetings and critical incident meetings every 2 months. Critical incidents were first presented in the critical incidents meetings, at which any issues could be highlighted by any department member for emphasis, information and discussion. Policy issues were identified and designated to a trainee, who would work in collaboration with a faculty member. These policies were then presented in the departmental quality assurance issues meetings.
One of the problems encountered by us when analysing incidents was the lack of standardisation of the terms used. Some of the discrepancies in data generated from different countries may reflect this difference. An attempt has been made by Banks and Tackley to standardise these terms [18] and this could also form the basis of future qualitative research.
In contrast to other studies such as AIMS [9], in which highly detailed data collection forms were used, we used only a single page front/back form, for reasons of cost. This may have resulted in cruder data collection. A more sophisticated system would require more money, cumbersome data handling and more man-hours for data analysis. On average, 40 forms were generated each month and this amounts to a yearly cost of less than US$25. Even the crude data collected were found to be helpful as a learning experience and allowed development of policies that improved patient safety.
In summary, we found the technique to be useful in revealing trends, as an educational tool and as a method of quality assurance to develop policies to prevent recurrence. It was particularly attractive to us because of low costs.
The spectrum of incidents reported in our study is similar to that of other critical incident reporting studies. This suggests that the anaesthesia system, rather than type of training, equipment or culture, is to blame. We are now expanding the activity to other areas in our institution, and anaesthesia departments in other hospitals have also shown an interest in its application to their own settings. This simple, low-cost quality assurance measure may help to improve the standard of anaesthesia care in developing countries. However, it requires at least one dedicated, identified faculty member to constantly motivate the staff to return the forms.
