Significant differences in the quality of incident reports – a comparison of four acute hospitals in Finland

Abstract The aim of the study was to compare the quality of the description of the content of patient safety incident reports of ‘near miss’ and ‘adverse event’ occurrences and to examine whether the contributing factors behind the incident were identified. Data were collected from an electronic incident reporting system for a 1-year period (2015) in four acute hospitals in Finland. The analysis framework was based on the incident reporting guidelines, and the data were analysed using statistical methods. The most deficiencies were in records of the consequences of the event for the staff and unit (47%) and the consequences of the event (35%). The description of the content of ‘near miss’ situations did not differ significantly from ‘adverse event’ situations, but statistically significant differences were found between the hospitals in the quality of the description of the content of incident reports. Incident reports did not always identify the processes behind the incident or the factors that contributed to the occurrence of the incident, such as human error. Blaming was still evident in the incident report descriptions.


Introduction
Patient safety incidents are classified as either a 'near miss' or 'adverse event' (WHO, 2020). This study examined the quality of the description of the content of incident reports for these two types of incidents, which were voluntarily reported in an electronic incident reporting system.
In patient care, adverse events are not uncommon, but not all 'adverse events' or 'near misses' are reported by staff (Eiding et al., 2022;H€ ark€ anen et al., 2019;Rashed & Hamdan, 2019). One explanation may be the negative attitude towards reporting and the fear of guilt (Carlfjord et al., 2018;Rashed & Hamdan, 2019). According to previous studies, the system for reporting is complex (Carlfjord et al., 2018) and difficult to report (Sujan, 2015). The most used system for reporting incidents is electronic patient safety event reporting (Gong et al., 2017). Physicians report fewer incidents than nursing staff (Gong et al., 2017), which may be due to a misunderstanding that only nurses can make reports (Varallo et al., 2018) or ignorance concerning who is responsible to report (Rashed & Hamdan, 2019). In addition, physicians' training may lack guidance on how to make incident reports (Louis et al., 2016).
Previous studies have provided information on the reporting of pharmacotherapy-related (H€ ark€ anen et al., 2020;Westbrook et al., 2015) and fallrelated incidents in patients (Richardson & Carter, 2017). The role of human error in incidents has been studied previously as a part of patient safety culture (Duarte et al., 2018). Additionally, factors behind human error (Chatziioannidis et al., 2017), for example, in the implementation of pharmacotherapy (Hsieh et al., 2019) and background variables (Hu et al., 2018;Sendlhofer et al., 2019) in the case of adverse events, have been investigated.
A comparison of the reporting activity for different types of incidents has found that fewer 'near misses' are reported than 'adverse events' (Espin et al., 2019). A comparative study between different countries also identified differences in how the incident reporting system is used (Manser et al., 2017), for example, whether incidents are reported anonymously, how systematic events are analysed and how development measures are implemented (Kellogg et al., 2017;Peerally et al., 2017). Comparing healthcare safety with other high-risk industries such as aviation reveals similarities, like the use of checklists, in ensuring safety and harm prevention. But there are also differences. In health care, an adverse event most often affects only one patient whereas in aviation the question is often about many people. In aviation there are no 'low-level' incidents, instead, any in-flight safety event is reported and thoroughly investigated. In health care, low-level harms, such as CAUTI (catheter-associated urinary tract infection), can be disregarded and only serious adverse events, such as patient injury or death, are investigated using RCA (root-cause-analysis) (Gerstle, 2018;Kapur et al., 2015;Slawomirski & Klazinga, 2022;Thoroman et al., 2018) Moreover, patient safety incidents often incur additional costs (Wulff Risør et al., 2017). A recent report by OECD reveals that the direct cost of treating patients who have been harmed in their care is around 13% of healthcare spending in developed countries (Slawomirski & Klazinga, 2022). The ultimate purpose of reporting incidents is to prevent them from happening again by learning from them at the individual and system levels of the organisation (Kinnunen et al., 2009) (Figure 1). Therefore, from a management and effectiveness perspective, incident reporting should also be combined with cost-effectiveness (Carlfjord et al., 2018) and incident learning (Macrae, 2016;Stavropoulou et al., 2015). The incident reporting process is not only based on the reporter's views and incident description. In addition to that, the supervisor handles the incident report, identifies development needs, as well as the need for training and orientation or the need for changes in the care processes at the unit or at the system level in the hospital. According to Sengen et al. (Peter et al., 1994), a learning organisation is based on five different aspects: individual learning, individuals' own operating models, shared vision, team learning, and systems thinking, such as the contributing factors behind incidents.
The learning organisation includes the knowledge of staff roles and responsibilities (Lalani et al., 2020). A study has suggested that at the individual and group level, incidents can be learned through colleagues' experiences and by discussion of patient safety with colleagues (Sujan, 2015). According to Rashed et al. (Rashed & Hamdan, 2019), it is important to report that all types of incidents and reporting systems can be used to identify errors and learn from them.
This study examined to what extent incident reports contain the elements described in the instructions for completing the report and analysed the quality of incident reports. To the best of our knowledge, only a few previous studies have been conducted on this topic. The study supports the objectives of Finland's national patient safety strategy (Ministry of Social Affairs and Health, 2017).
The aim of this study was to compare the quality of the descriptions of the content of incident reports for 'near miss' reports and 'adverse event' occurrences and to determine whether the contributing factors behind the incident were identified. We hypothesised that the nature of the event was not associated with how the occurrence of the incident was described. The hypothesis was tested when comparing the data there were no differences between hospitals in the description and quality of incident reports.
Four research questions were stated as follows: 1. Is there a difference in the quality of the description of the content of 'near miss' and 'adverse event'? 2. Have the contributing factors related to the incident been described? 3. What percentage of the incident reports led to development actions? 4. Is there a difference in the quality of the description of the content of incidents between the four studied acute care hospitals?

Methods
The data consisted of voluntarily made incident reports related to patient care by the staff at four acute hospitals in Finland. From the beginning of March 2015, 200 consecutive notifications were included from each hospital (total n ¼ 800). Patient reports and reports on social care, primary health care and hospital schools were excluded from the material. Thus, the final data included 739 notifications. A framework for the analysis based on the instructions of the incident reporting system (Awanic Oy, 2016) (http://83.150.87.4/haipro/20/lomake. asp?kieli=ENG) supplier on how to complete the notification (20 criteria) was used. The criteria consisted of the nature of the event, the state of handling of the event, the description of the content of the incident, the type of event and the investigator's assessment. The classification was 'sufficiently', 'partially' or 'not at all' used in the description of the incidents. The analysis framework was peer-reviewed prior to use. In the peer review, 20 cases were analysed by two patient safety experts according to the analysis cluster, and the consistency of the evaluations made by the evaluators was calculated to be 86%.
In the analysis of the actual study data, the categories 'partially' and 'not at all' were combined into a single category labelled 'no' and the category 'sufficient' into the category 'yes' because some incident reports contained less than five entries.
The quality of the description of each incident report was analysed. The adequacy of the content of the incident report was assessed based on how the report had been recorded, what had happened and what the consequences were.
The data were also used to assess the quality of the content of the notification, which could have caused the incident and what the contributing factors were.
The following scale was used to classify the adequacy of the content of incident reports: more than five criteria described in the notification (i.e. 75% or more described ¼ content described well), three to five criteria described (50-74% ¼ content described moderately) and less than three criteria described (<50% ¼ poor content description) (Rockville et al., 2018).
The quality of the analyses was ensured by the fact that the criteria used in the analysis of the data followed the instructions of the notification.
IBM SPSS 26 software was used for data analysis using cross-tabulation and Pearson's Chi-square test.
The study protocol was approved by the Hospital District Ethics Committee, and research permission for each hospital under study was obtained.

Results
The nature of the event ('near miss' or 'adverse event') was used as the background variable. Altogether, 42% of incident reports were categorised as a 'near miss'.
The incident reports were examined to identify whether the notification was made in accordance with the system supplier's instructions on what to record in the notification ( Table 1). Descriptions of what and how the incident happened, what was being done, and the circumstances at the time of the event were well recorded (over 75%; more than five criteria described) in both the 'near miss' and 'adverse event' reports.
The descriptions of how the situation was handled, what were the consequences for the patient, properly recorded suggestions on how the event could be prevented from recurring and what were the consequences for the staff and unit, were recorded moderately (50-74%; three to five criteria described).
A more detailed analysis of the adequacy of the content relative to the nature of the event was conducted using cross-tabulation and Pearson's chi-square test. The results showed that there was a significant relationship between the nature of the incident ('near miss' or 'adverse event') and the incident description what and how the event happened (v 2 (2) ¼ 3.891; p ¼ 0.049). A similar significant relationship was observed between the circumstances at the time of the event and other factors that, in the opinion of the notifier, contributed to the occurrence of the event (v 2 (2) ¼ 6.030; p ¼ 0.014). (Table 1).
Many incident reports (59%) included the notifier's suggestions on how to prevent a recurrence. Contributing factors were identified in 79% of all notifications. As a remedy, care and attention were proposed in 11% of all notifications. A small proportion of incident reports (6%) blamed someone or some people for the event ( Table 2).
The incident reports were also examined whether the incident was recorded as an anomaly related to the operational process or information flow or was attributed to human error. Based on the results, the recording of human errors was low. The comparison of the results of the original descriptions (health care professional) with the evaluation of the description in this study (researcher) is shown in Table 3.
Less than half (38%) of the incident reports led to development measures. The highest number of development measures was observed in Hospital B (54%), and the lowest was observed in Hospital A (27%). Among the development measures, changes in operating methods and renewal of guidelines were mentioned.
Whether the descriptions of incidents differed between the four hospitals was also investigated (Figure 2). All the hospitals had a good record (over 75%) of what was being done and what had happened, whereas the recording of the treatment of situations and consequences for patients was moderate (50-74%). In other event categories, the quality of records varied between hospitals.
Poor records (less than 50%) of how the incident had affected the staff was found in Hospitals B and D. Hospital A had the fewest records of what was being done, and in this respect, the hospitals also differed statistically significantly (p ¼ 0.014). Descriptions of what had happened in the emergency were best recorded in Hospital B, and in this too, the hospitals differed statistically significantly (p ¼ 0.001). There were also significant differences in the recording of the incident treatment between different hospitals, with the highest number of entries in Hospital C (p ¼ 0.016). There was no statistically significant difference between hospitals in the description of the consequences of the incident, but the description of the consequences for the staff and unit in the records differed significantly between the different hospitals. The lowest number of registrations was in Hospital D (p < 0.01). The conditions at the time of the event and other factors that contributed to the occurrence of the event were best described in Hospital B and worst in Hospital A. Again, the hospitals differed statistically significantly (p < 0.01). Records to prevent recurrence of the incident were best in Hospital B, and there was a statistically significant difference between hospitals (p ¼ 0.05). A detailed description is given in Figure 2.  Table 3. Identification of the underlying variables reported by a healthcare professional and evaluated from the researcher's point of view (n ¼ 739).

Healthcare professional Researcher
This is an error due to incomplete / incorrect information 59% 61% This is a process error 10% 36% Human error in the background 11% 28%

Discussion
The main findings of this study were the following: The quality of the descriptions of the incidents was in general good or moderate, but differed between near-misses and adverse events, with adverse events being better described. Additionally, less than half of the incident reports included a description of the development measures to prevent incidents in the future. Furthermore, the quality of the descriptions of the incidents differed in the four study hospitals. However, the nature of an event, that is, whether it is a near-miss event or an adverse event, should not affect the reporting activity or event description (Hewitt et al., 2016).
The factors and circumstances that contribute to each incident should be identified. If near-miss notices are carefully reviewed with staff, necessary protection mechanisms can be identified and devised together.
In a learning organisation, patient safety and incidents should be reviewed daily and addressed systematically as part of the development of the patient safety process with staff, organisational management, trust staff and patients (Sujan, 2015). The quality of incident reporting can also be viewed from the perspective of the development of the learning organisation (Leistikow et al., 2017). Although the activity of notifications is constantly increasing, the number of development measures taken based on notifications is small (Ahlberg et al., 2020;Liukka et al., 2019;Macrae, 2016). It is the responsibility of staff to report every 'near miss' events related to a patient's care and any 'adverse event' that has occurred to the patient. Supervisors should process the notifications and plan development measures together with the staff. Development measures proposed by staff, such as increased accuracy or diligence, are difficult to implement without addressing why accuracy was inadequate during the event and what factors contributed to it. It is important that incident reports are processed and learned from together (Hewitt et al., 2016).
Incident reports do not necessarily identify the practices and related processes that contribute to the occurrence of an event, such as patient identification or the different stages of medication implementation (Aaronson et al., 2019). However, identifying them could help to improve development measures and assist the supervisor in analysing the reports. The processes behind an incident usually involve different professional groups and may be related to, for example, the flow of information and its shortcomings (Aaronson et al., 2019). When examining the quality of notification, the root causes behind the occurrence of an incident are often left unidentified, even though they contribute to a deeper analysis of the occurrence of the incident and to consideration of how incidents could be prevented, for example, through checklists or training (Eiding et al., 2022). The incident reporting system needs to output information to support management quickly in a user-friendly manner (Carlfjord et al., 2018;Denecke, 2016). In addition, it should include a variety of analytical tools to study learning processes (Hewitt et al., 2016). Analysis tools are needed when dealing with different stages of processes and the course of an event. The system should also notify if the descriptions of the event are insufficient (Gong et al., 2017). If the system was connected to the patient information system, an incident notification (Mitchell et al., 2016) could be made automatically in the event of an incident, but this could be challenging in terms of data protection, security and anonymity. According to the results, guilt and blaming are still reflected in incident reports. However, the aim is that staff should also be able to report incidents without fear of being blamed (Carlfjord et al., 2018;Mjadu & Jarvis, 2018;Rashed & Hamdan, 2019).
The four studied hospitals had the same incident reporting system in place. Thus, the underlying instructions for completing the report were the same. There were qualitative differences in the incident reporting between hospitals but also within them. There were shortcomings in identifying the human error. If underlying human factors can be identified, they can be addressed, for example, by looking at job descriptions or the content of shifts, as well as the adequacy of human resources. Patient safety culture has an impact on how human error is reported (Duarte et al., 2018), as reporting and filling in reports may be influenced by the different organisational cultures of hospitals (Burlison et al., 2020). In this study, it was not possible to determine what was the root cause for the discrepancy between occupational groups.

Conclusion
The quality of the descriptions in incident reports differed between hospitals. Notifications did not always identify the factors or the underlying processes that contributed to the occurrence of the event. Deficiencies in identifying the contributing factors suggest a need for additional training and guidance in utilising the patient safety incident system. Discussion of the incidents in multi-professional teams is recommended to improve reporting activity and to eliminate underreporting of problems. The incident reporting system should be easy to use and provide analytical information to support management and prevent the occurrence of similar events in the future.
Currently, head nurses, head physicians and experts in various fields have been named as handlers of patient safety incident reports in hospitals. As a result, notification processing is often decentralised, which poses challenges for consistent processing. One solution in hospitals could be the centralised processing of reports, which could focus on the quality and consistency of processing to help identify risks associated with patient care and necessary development measures.
In further research, it could be examined in more detail whether there is a qualitative difference between different occupational groups in the reporting of incidents.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
The present work was financially supported by the State Research Funding, Finland.

Notes on contributors
Tuula Saarikoski is a Patient Safety Co-ordinator.
Kaisa Haatainen is a patient safety manager at the Kuopio University Hospital.
Risto Roine is an emeritus professor of patient safety at the University of Eastern Finland.
Hannele Turunen is professor (nursing science) at the University of Eastern Finland and nurse director at Kuopio University Hospital, Finland. Hannele's research interest is to examine patient safety culture from different perspectives, and together with her international research group she has published more than 100 scientific articles on the topic.