Monitoring and reporting critical incidents, such as hypotension or a state of awareness, can indicate the quality of clinical practice. Therefore, reporting medical complications voluntarily is encouraged by the World Health Organization and the Dutch Health Care Inspectorate.1,2 Registration of critical incidents not only provides an assessment of the quality of practice but also offers knowledge of the most frequent and most severe critical incidents.

Anesthesiologists should share their experiences with critical incidents in order to increase their knowledge of the potential risks and identify patterns in the development of critical incidents. The gaps and inadequacies found in the healthcare system can be optimized to improve patient safety.3-9 Furthermore, evaluation and feedback constantly encourage clinicians to report critical incidents.3,10

Many countries have developed systems to investigate the number and severity of these critical incidents.7,11-20 Most research has been based on closed claim analysis or anonymous reporting systems; however, these methods have limitations. For example, closed claim analyses will not contain all complications, only those that involve patients and are deemed important. Therefore, in order to evaluate a hospital reporting system and identify topics to improve clinical care and patient safety, we designed the present study to describe critical incidents that anesthesiologists reported voluntarily and non-anonymously through an anesthesia information management system (AIMS) in a tertiary referral hospital.

Methods

Study design

The Institutional Review Board (IRB) of the University Medical Center Utrecht reviewed the study protocol and found that it was not subject to the Dutch Medical Research in Human Subjects Act. Therefore, the IRB waived the need for informed consent (11-271/C; July 5, 2011). This observational study describes prospectively reported critical incidents and complications relating to anesthesia in patients 18 years and older undergoing any type of anesthetic procedure in a tertiary referral university hospital (University Medical Center Utrecht, the Netherlands) from January 1, 2005 to May 18, 2011. Anesthesiologists and anesthesia registrars voluntarily reported complications and critical incidents on a non-anonymous basis via the 20-item complication list of the Netherlands Society of Anesthesiologists. The reporting system was implemented in September 2004; therefore, we chose to evaluate critical incidents reported as of January 1, 2005 to allow an optimization period of three months.

Definitions

We defined a critical incident as an event that could have led (if not discovered or corrected in time) or did lead to an undesirable outcome, i.e., ranging from increased length of hospital stay to death or permanent disability. We included all anesthesia-related critical incidents that occurred at a time when the patient was under the care of an anesthesiologist and were described in clear detail by a person who either observed or was involved in the critical incident. We included critical incidents that not only seemed preventable (i.e., inadequate preoperative screening) or involved human error (i.e., medication error)21 but also were non-preventable (i.e., unexpected difficult intubation).3,15,22

Data acquisition

Critical incidents were reported by anesthesiologists and anesthesia registrars (reporters) in the AIMS on a voluntary and non-anonymous basis. During every anesthetic procedure, a menu item in the AIMS termed complication is presented by pressing the standard event key 〈start skin closure〉, at which time, a reporter can complete a standardized computerized audit form. If a critical incident is reported, a drop down menu displays the 20-item complication list (with miscellaneous as an additional option) developed by the Netherlands Society of Anesthesiologists (Table 1). Thereafter, the incident’s grade of severity can be reported and, if deemed necessary, free text can be added. If the complication report is not entered into the database by the end of the day, the anesthesiologist involved receives a reminder e-mail. Upon completion, the critical incident report is stored in a database within the AIMS along with the patient characteristics. The registry also includes a means to assign a pop-up warning for subsequent anesthetic procedures (i.e., difficult intubation).

Table 1 Classification of critical incidents according to Netherlands Society of Anesthesiologists
Full size table

The currently used 20-item complication list of the Netherlands Society of Anesthesiologists facilitates a generalized classification of critical incidents. After reviewing the critical incident reports, we concluded that we could not base firm conclusions on the classification system as it was too generalized; therefore, we reclassified all critical incidents. Based on the initial classification and comments added by the reporter, we reclassified the critical incidents in keeping with a classification system of the German Society of Anaesthesiology and Intensive Care13 which is a more detailed classification system on which to base our conclusions. If no comment was available or the comment was unclear, we consulted the AIMS to investigate the critical incident in detail. One researcher (K.M.) reviewed all critical incidents. When information was inconsistent, consensus was reached by discussion with two researchers (J.d.G. and B.v.Z.). If more than one category was possible for one critical incident, the most appropriate or most severe category was chosen. If different critical incidents occurred during one anesthetic procedure, these were categorized as separate critical incidents. All reports involving death as grade of severity were discussed with all observers (K.M., J.d.G., and B.v.Z.).

Statistical analysis

All statistical analyses were performed using SPSS® 17.0 for Windows (SPSS Inc., Chicago, IL, USA), except for the calculation of the 95% confidence interval (CI) according to Wilson’s formula (EpiTools: http://epitools.ausvet.com.au). Procedures with more than one critical incident were counted once. Where appropriate, a Chi square test or an independent samples Student’s t test was carried out to display differences between groups. All reported P values are two sided.

Results

The complication status of 104,133 (94.4%) of 110,310 anesthetic procedures was known (95% CI, 94.2 to 94.5) (Figure). In total, 4,062 events were reported in the AIMS, and 158 (3.9%) reports were classified as not being a critical incident (95% CI, 3.3 to 4.5) because they consisted of surgical complications and warnings for a subsequent anesthetic procedure. The remaining 3,904 critical incidents were found in 3,807 of the 110,310 anesthetic procedures (354 per 10,000 anesthetics; 95% CI, 343 to 365). The 3,904 critical incidents consisted of one single critical incident in 3,715 (97.6%) anesthetic procedures, two critical incidents in 87 (2.3%) anesthetic procedures, and three critical incidents in five (0.1%) anesthetic procedures. Table 2 shows demographic data of the study population; no clinically significant differences were found. The largest critical incident categories were cardiovascular incidents, with 1,164 incidents (106 per 10,000 anesthetics), respiratory problems with 851 incidents (77 per 10,000 anesthetics), and lesions with 820 incidents (74 per 10,000 anesthetics) (Table 3). The cardiovascular critical incidents consisted mainly of hypotension; the respiratory problems critical incidents consisted mainly of difficulties to ventilate (with or without hypoxemia), difficulties to intubate, bronchospasm, and laryngospasm; and the lesion critical incidents consisted mainly of technical difficulties with regional anesthesia (Table 4). The largest groups of reported critical incidents were technical difficulties with regional anesthesia (40 per 10,000 anesthetics) and hypotension (39 per 10,000 anesthetics) (Table 5).

Figure
figure 1

Flow diagram for reclassification of critical incidents

Full size image
Table 2 Characteristics of study population
Full size table
Table 3 Incidence of critical incidents in the different categories with reported consequence
Full size table
Table 4 Incidence of the various critical incidents after reclassification
Full size table
Table 5 Top ten most frequently reported critical incidents
Full size table

Critical incidents with (probable) permanent damage consisted primarily of respiratory and cardiovascular critical incidents. Forty-three (1.1% of all critical incidents) critical incidents led to the death of a patient receiving an anesthetic procedure; forty of those critical incidents comprised a cardiovascular incident ranging from arrhythmia to myocardial infarction (Table 6).

Table 6 Critical incidents with (probable) permanent damage or death as consequence
Full size table

Discussion

The voluntary and non-anonymous critical incident registration system in this study proved to be very effective (response rate 94.4%). This high response was achieved by way of a reminder in the AIMS for reporting during skin closure and an e-mail reminder after completion of the anesthetic procedure. Furthermore, the non-anonymous registration allowed feedback through a twice weekly complication meeting in which action regarding a critical incident was discussed and initiated, thereby encouraging clinicians to report critical incidents. In 3.5% (354 per 10,000 anesthetics; 95% CI, 343 to 365) of anesthetic procedures a critical incident was reported, which is similar to the incidence reported in children using the same methodology.23

The present voluntary and non-anonymous reporting system is unique and has its advantages and disadvantages. Voluntarily reported critical incidents may suffer from underreporting.3,24 Previous studies have shown a low level of compliance when voluntary reporting was compared with automatically detected critical incidents,25-27 and in a different study, an incidence of 28% was reached when researchers completed a retrospective evaluation of all anesthetic procedures.13 Nevertheless, the response rate in the present system was very high (94%), and the advantage of the present system is the fact that anesthesiologists reported only those critical incidents considered to be clinically relevant. The non-anonymous system of reporting may also cause underreporting because a reporter might refrain from reporting due to fear of consequences.6,10,24 Nevertheless, a strong advantage of non-anonymous reporting is the ability to discuss the critical incident with detailed information from the involved anesthesiologist, which can lead to a teaching moment.3

The reporting system used in this study was based on the 20-item complication list of the Netherlands Society of Anesthesiologists. This 20-item complication list was not sufficient for detailed analysis and required extension as 21.6% of reported events could not be classified within the original list and were reported as miscellaneous (Tables 1 and 4). Nevertheless, the limited number of items in the classification system of the Netherlands Society of Anesthesiologists and the large amount of critical incidents in the miscellaneous category might have induced underreporting of the items not in the original list. For example, some might judge certain events as a critical incident, while others might judge the same event as not being a critical incident, and vice versa.24,28 It could be argued that not every critical incident that we present is truly a critical incident, e.g., technical difficulties with regional anesthesia are an inevitable occurrence when performing regional anesthesia. Furthermore, for the present study, all critical incidents were reclassified retrospectively to allow detailed analyses, and lack of information may have caused misclassification.

Cardiovascular incidents (106 per 10,000 anesthetics), in particular hypotension (40 per 10,000 anesthetics), comprised the majority of critical incidents (Tables 3 and 5). Previous studies showed the same level of cardiovascular incidents,13,29 whereas some studies indicated that difficulty with airway management 11,14,16,29,30 or wrong drug/wrong drug-dose/wrong drug-labelling 14 was the critical incident that occurred most frequently. This variance in number and type of critical incident might be due to the diversity of methods in the reporting systems and differences in definitions. For example, closed claims studies report death (26%), nerve injuries (22%), and permanent brain damage (9%) as the most common complications.11

The present study identified the most frequently reported and most severe anesthetic critical incidents in our hospital on which to base future improvements for patient safety. The technical difficulties with regional anesthesia (Table 5) are being addressed in part by implementation of ultrasound guidance,31 but we propose a thorough investigation to determine which regional technique results in the most technical difficulties. Furthermore, the administration of the wrong drug (Table 4) is being tackled by strictly double-checking medication before administration.32

In conclusion, the present study shows that the present reporting system in AIMS along with e-mail feedback leads to a very high response rate in reporting critical incidents. Even so, the complication lists of the Netherlands Society of Anesthesiologists proved to be too limited, and therefore, the present list of complications can be used as an alternative. Cardiovascular complications were reported most frequently.