The impact of a standardized incident reporting system in the perioperative setting: a single center experience on 2,563 ‘near-misses’ and adverse events

Patient safety is a global public health issue receiving rapidly increasing attention. Patient safety is the (near) absence of (the chance of) avoidable harm inflicted on the patient through the actions and/or negligence of employees or through flaws in the healthcare system [1]. Numerous medical record studies have shown that unsafe care may result in adverse events (AEs) leading to harm in 3-17% of hospital patients [2]. An AE is “an unintended injury or complication resulting in prolonged length of hospital stay, disability at the time of discharge or death caused by health care management and not by the patients’ underlying disease” [3]. A total of 51-77% of AEs in hospitals are related to perioperative care [2]. A systematic review reveals that 14% of perioperative patients experience some form of AEs, that 38% of these AEs are preventable and that 4% of patients experiencing AEs have fatal outcomes [4]. Preventable AEs are the result of care that falls below current professional standards and the expected performance of practitioners or care systems [1],[2].

To improve patient safety and reduce adverse events, a national safety management system has been implemented in Dutch hospitals (‘Veiligheidsmanagementsysteem = VMS’) between 2008–2012. VMS focuses on the management of risks (the prevention of patient harm and its possible consequences) and is described in the ‘Netherlands Technical Agreement’ (NTA) [5]. The starting point for this NTA was the report “Here you’ll work safely, or you won’t work here” [6]. VMS consists of systems for risk identification, risk analysis, risk evaluation, incident reporting, incident analysis, and managing recommendations and improvement measures.

The operating department (including the clinical operating and recovery rooms) of the Radboud university medical center introduced the Hospital Incident Management System (HIMS) in July 2009. HIMS aimed at facilitating the voluntarily and confidentially reporting of perioperative incidents. After using HIMS for three years, we set out to analyze whether there are specific types or patterns of the reported information and to detect the areas where further efforts are required to improve perioperative patient safety. Specifically, we wished to investigate the number and characteristics of the voluntarily reported perioperative incidents. From the start of recording we noticed that Standard Operating Procedure (SOP) non-compliance was a prominent reported cause of incidents.

Since humans are fallible, systems must be designed to prevent humans from making errors [7],[8]. Standardisation, the use of guidelines and protocols, is generally considered to improve perioperative safety. Preventable patient harm may be the result of guideline non-compliance but unfortunately there is hardly any literature about this problem in perioperative care. Dutch hospital record review studies suggest that at least 10-15% of AEs are directly or indirectly related to procedure non-compliance [9],[10]. The compliance rate of the use of surgical safety checklists ranges from 12% to 100% (mean 75%) [11]. French research revealed that alarms are frequently ignored by the operating theatre staff [12]. English anesthesiologists witnessed that 22% of incidents during anaesthesia are related to protocol violations [13].

There is an association between undesirable perioperative events and subsequent critical perioperative outcomes [14]. Recent studies reveal also that better guideline compliance is associated with better perioperative outcomes [15],[16]. Therefore, adherence to guidelines is an important target for safety improvement programs. This study was designed to provide data on the number, nature and causes of voluntarily reported perioperative incidents.

This retrospective study is performed in the Radboud university medical center, which has 953 beds and 25 operating rooms (ORs). All data were obtained from HIMS, a database that is developed by the Patient Safety Company (http://www.patientsafety.com), licensed to the Radboud university medical center and used by the operating department since July 2009. The study period lasted from the 1^st of July 2009 to the 1^st of July 2012. Incidents in HIMS are reported as a) ‘adverse events’, i.e. any unintended or unexpected event which could have led or did lead to harm of one or more patients receiving hospital care and b) ‘near-miss events’, i.e. events or circumstances that nearly occurred but were prevented (through luck or intervention) and did not lead to patient harm. [1] Harm is defined as any injury to the patient which leads to an extension or increase in the treatment, to temporary or permanent physical, psychological and/or social functional loss, or to death [5]. The worsening of a patient’s condition as a result of the natural progression of an disease is not considered ‘harm’ [5].

At the introduction of HIMS, hospital wide appointments were made for reporting of incidents, based on the national agreements [5]. All employees, including professionals working in the perioperative process were encouraged to report incidents in order to prevent them from happening again. They were requested to fill out a semi-structured digital form of the reporting system. At least the following had to be reported: date, time and location of the incident, a brief description of the event and the circumstances, the type of incident and possible causes, the potential patient impact and the estimated risk of recurrence of that incident, and the measures that may prevent the incident from repeating. The predefined categorization of incident types is illustrated in Figure 1. In Tables 1 and 2 an explanation of the used terms is given, in particular of the concepts estimated risk of recurrence (Table 1) and estimated consequences for the patient (Table 2). This categorization of patient consequences is based on national guidelines as described in the NTA [5]. Based on the estimated patient consequences and risk of recurrence, the incident is automatically classified into four risk categories (Table 3). This risk classification is the basis for actions to be taken by the Operating Room Incident Reporting Committee (ORIRC) as described in Table 4. According to the Dutch consensus classification reported incident causes were classified as human, organizational, technical, patient related and other [1],[2].

The ORIRC consists of a surgeon, two anesthesiologists, a patient safety coordinator, a recovery room nurse, two operating room nurses, an anesthesia nurse, a logistic manager, a logistic employee, and a technician. The ORIRC reviews and discusses incidents weekly with priority to high risk events. Estimating the risk of recurrence and potential consequences for the patient at the time of reporting and before analysis of the incident may be difficult. Therefore, the ORIRC may adapt the risk score after analysis of the incident. Subsequently, the ORIRC undertakes actions, such as gathering additional information, cross-checking with the medical charts, and exploring key contributing factors through informal discussions with care providers, managers, and/or other expert(s), asking ‘why’ and ‘what’ influenced the occurrence of this (near)incident. Besides these analyzing actions, other activities of ORIRC are trend watching, advising to adjust products and procedures, providing feedback and spreading newsletters and safety alerts to all involved in perioperative care. After closing a report, all information is automatically anonymized. The information in HIMS is not used for blaming, shaming or (legal) actions against the reporting professional, but for improving the quality of patient care.

Descriptive statistics were used to describe the number, nature and causes of incidents. Chi-square tests with Yates’ correction and odds ratios (ORs) with 95% confidence limits (CL) were applied to test differences between physicians and non-physicians and between perioperative and hospital wide reporting behaviour. A p-value < 0.05 is considered to show statistical significance.

We studied the characteristics of voluntarily reported perioperative incidents at our operating department in the period July 2009-July 2012. Reported incidents included ‘near-miss’ events as they can be seen as ‘free lessons to be learned’ [17]. Overall, 67,360 operations were performed in the three-year study period and 2,563 perioperative incidents were voluntarily reported. Most (84%) of these incidents were not considered as potentially harmful for patients, but 16% were, which comprised 0.6% of all surgery patients. SOP non-compliance, shortcomings in communication, and mistakes or just ‘having forgotten’, appeared to be frequently reported incident causes. About one third of the reported perioperative incidents was associated with SOP non-compliance; 79% of these SOP-related causes were related to human failure. The SOP-related incidents were not associated with more patient consequences than other incidents. It is noteworthy that the percentage voluntarily reported incidents associated with SOP non-compliance within the operating department did not differ from the hospital-wide incidence.

There is increasing information that better guideline compliance is associated with improved perioperative outcomes [15],[16]. A study on the implementation of a bundle of care to reduce postoperative surgical site infections showed that increased bundle compliance from 10% to 60% was associated with a significant 36% reduction in infection rate [16]. Another study in a Dutch university hospital on the effect of the use of a perioperative safety checklist showed a significant 56% reduction of the in-hospital 30-day mortality after surgery in patients with completed checklists whereas the mortality rate remained unchanged in patients with partially completed or non-completed checklists [15]. The results of our study clearly show that a thorough analysis of the barriers that hinder adequate communication and compliance with SOPs is essential to further improve perioperative safety [18]–[20].

Incident reporting and investigation was first used in the 1940s to improve safety and performance of military aviation [21] and some decades later incident reporting was introduced in anesthesiology [22]. Meanwhile, incident reporting has become a widely recommended method to gather information about AEs in hospital care [18] and several methods have been developed [13],[23]–[25]. Much information about AEs in hospitals comes from the retrospective study of medical records [2],[4],[26],[27]. However, retrospective record review has the well-known disadvantage that AEs may not have been identified or reported because they were not recorded in the medical file. These retrospective studies also rarely give information about ‘near-miss’events. Patient’s information is also a reliable source for the identification of care-related AEs [28]. Monitoring the performance of professionals by trained observers may provide the most reliable picture of perioperative practice [13],[29]. However, these studies may underreport the actual rate of AEs because the researchers only observed during daytime of weekdays [29]. Smith at al. witnessed that around 20% of the incidents reported were the result of violation from existing protocols [13].

To further improve patient safety it is essential not only to get insight into the number, type, risk and causes of voluntarily reported incidents, but also to reach agreement on definitions and reproducible ways of reporting and scoring. The Netherlands Technical Agreement aims to contribute to the uniformity of the safety management system in hospitals and to create openness about patient safety towards patients and public. In order to improve the effectiveness of the national incident reporting system, we suggest further standardization of incident reporting data, specifically of the categorization of the types and causes of reported incidents. This has the potential to provide us with a system to compare hospital performance and to benchmark performance using a validated system.

The five basic elements essential for the successful translation from incident reporting to learning are 1) an open reporting culture allowing independent non-punitive data input, 2) the opportunity to freely narrate one’s own version of the event, 3) an analysis that turns the incident report into a lesson, 4) adequate feedback [23], and 5) definitions clear enough to be used in other centres in order to be able to compare hospital performances. Investigating and analyzing incidents requires optimal engagement of physicians to get insight into the root cause and prevention of these events [23]. The use of a standardized framework for analysis of events has been recommended and introduced in several centers around the world including our center [23],[30]. A positive safety culture is associated with increased willingness to report errors [31],[32], with increased compliance with SOPs [33] and with fewer AEs [34]. Apparently professionals at our department felt safe enough to report communication failures and non-compliance with SOPs. They knew that the information will not be used to blame or undertake actions against the reporter and that after analysis of an incident the information was anonymized to be used for training purposes and improving the quality of care, if deemed useful for this purpose.

A major weakness of incident reporting in all studies is under-reporting. Incident reporting may capture only 4% to 50% of AEs [25],[27],[32]–[38]. The reasons for not reporting are numerous: clinical factors (e.g. emergency scenario), time constraints, unfamiliarity with the system, problems with the definitions of what constitutes a reportable incident or near-incident, lack of a hospital policy of ‘no disciplinary action’ on incident reporting, lack of anonymity, lack of feedback and confusing aims of the reporting system as such [13],[18]. However, the best policy to improve reporting behaviour is currently unknown [39]. Unfortunately, because of the risk of under-reporting, voluntary incident reporting is not a reliable instrument to monitor performance in e.g. the plan-do-study-act cycle. However, it may be used as qualitative monitor to identify areas that require further efforts to improve perioperative patient safety.

Voluntarily incident reporting provide important and detailed information about perioperative patient safety problems. The most important finding of the present study is that professionals themselves report non-compliance with SOPs associated with human failure is an important area for improvement. Furthermore shortcomings in communication, mistakes, and forgetting were identified as important targets for improvement to reduce perioperative incidents in our hospital. This finding requires acknowledgement of the risk of human attitude, behaviour and failure. Moreover these findings challenge for the development of tools to improve guideline adherence and effective communication, in order to improve perioperative patient safety.