Abstract
Background
Despite two decades of promoting systems-based approaches, healthcare safety investigations often fail to produce system-level changes. This study analyzed the quality of recommendations from serious incident investigations conducted by multidisciplinary teams in Finland.
Methods
Serious incident investigation reports (n = 169) from 2018 to 2023 were collected from 11 well-being services counties, with a total of 525 recommendations being extracted and classified as weak, intermediate, or strong according to the US Department of Veterans Affairs’ National Center for Patient Safety criteria. Each recommendation was assigned only to one strength category.
Results
More than half of all recommendations were weak (56%) and focused mainly on staff training (43%) and new procedures or policies (39%). Intermediate recommendations comprised 37%, primarily targeting documentation and communication (63%), while only 7% were strong. Approximately one in four investigations (24%) included at least one strong recommendation. Mentions of recommendations being implemented were rare (20%), and only 7% included follow-up notes. None of the reports specified the analytical method used.
Conclusions
Finnish healthcare's serious incident investigations produce predominantly weak, behavior-focused recommendations that are seldom followed-up or implemented. The absence of standardized reporting formats and limited documentation of follow-up hinder the evaluation of impact. To strengthen patient safety, investigations should yield fewer, but stronger, system-oriented recommendations developed using structured analytical methods and evaluated through consistent national guidance and monitoring mechanisms.
Background
Despite being advocated for at least 20 years, the implementation of a systems-based approach to patient safety investigations in healthcare has progressed slowly. 1 A systems-based approach views errors not primarily as individual failings, but as consequences of underlying system factors, and therefore emphasizes changes to processes, structures, and conditions of care to prevent recurrence. 2 However, one approach that has been widely used to investigate patient safety incidents (PSIs) is root cause analysis. Rather than being one specific, clearly defined method, it is a broad approach aimed at identifying system-level causes and contributing factors behind PSIs.3,4 Many organizations in Finland began implementing investigations of PSIs after the Finnish Society for Patient and Client Safety published the Investigation of Serious Incidents Guide for Social and Healthcare Organizations 2012, which was later updated by the Ministry of Social Affairs and Health in 2023. 5
Since the beginning of 2023, social and healthcare services in Finland have been organized under 21 well-being services counties. In Finland, the reporting of safety incidents is voluntary, and a nationally standardized reporting system is currently lacking. Each well-being services county utilizes its own reporting platform, on which healthcare professionals document safety-related incidents, and typically, these reports are reviewed and managed by the immediate supervisor or a designated safety officer within the unit where the incident occurred. In cases involving serious adverse events, a separate safety investigation may be initiated. In counties where such a procedure is established, the decision to open an investigation is usually made by a chief physician through a formal written directive, with a multidisciplinary internal team then being appointed to carry out an independent investigation, which is recommended but not mandatory in Finland. As in many other countries, in Finland's healthcare system, incident investigations are typically carried out by internal staff, who gather information from multiple sources and construct a timeline of the key events. They analyze contributing factors using tools like fishbone diagrams or the five whys method, and the process should conclude with the development of recommendations and a corresponding action plan.5–9
In Finland, there is currently no comprehensive overview of the types of recommendations generated from investigations of serious adverse events. This study examines serious incident investigations conducted by internal, independent, and multidisciplinary teams within the social and healthcare systems of Finland's well-being services counties and aims to analyze the types of incidents and to characterize the recommendations derived from these investigations by examining their content, strength, and distribution.
Methods
Data collection
The research data were collected in the autumn of 2023 and comprised serious incident investigation reports from 2018 to 2023 that were requested from all 21 well-being services counties, the City of Helsinki, and the Hospital District of Helsinki and Uusimaa. After allowing a period of three and a half weeks for responses, a reminder was sent to each of the organizations. A total of 211 reports were received from 11 organizations (47.8%), representing all reports provided by these organizations for the study period.
Of the 211 reports, nine were excluded either because they fell outside the specified time frame, or because the description of the event was extremely unclear, or because there were no details at all. In addition, 27 reports were excluded because they did not include any recommendations.
One hundred and twenty-seven recommendations were removed because they were not actual recommendations, but rather statements—for example: “It is difficult for doctors to follow the results of blood glucose measurements” or “Preventing the incident would have been difficult”—or questions, e.g. “Did the patient's IV drug use history influence the response to the situation?”, or because they had been written in a conditional form, using words such as “could”, “would”, or “should”, including “There should be an adequate number of hospital beds in relation to the need for care” or in terms such as “it is suggested to consider / It is recommended to consider….”. Some recommendations were also removed because they were not related to the cause of the event or contributing factors but rather to other issues such as follows: “multidisciplinary development of adverse event report processing as a team effort” or “organizing an immediate debriefing so that all parties involved can participate.” Following these exclusions, the reports were re-examined, and six additional reports were excluded because no eligible recommendations remained. Thus, the final research data for the analysis comprised 169 serious investigation reports with 525 recommendations. The data selection process is illustrated in Figure 1.
Process of data selection.
Data analysis
The data were analyzed using a deductive content analysis approach guided by predefined classification frameworks. 10 In this study, incident types were classified using an established framework developed in a Finnish patient safety project coordinated by the VTT Technical Research Centre of Finland which underpins the HaiPro incident reporting system used in Finnish healthcare. 11 The framework includes predefined categories presented in Table 2 and is broadly aligned with international taxonomies such as the World Health Organization International Classification for Patient Safety. 12
Recommendations were classified using a deductive approach based on the framework developed by the US Department of Veterans Affairs National Center for Patient Safety, which categorizes recommendations according to their expected effectiveness and sustainability. Each recommendation was assigned only to one strength category (weak, intermediate, strong) and further into subcategories (Table 1). In this framework, the strength of recommendations reflects their anticipated effectiveness in preventing the recurrence of incidents. Strong recommendations typically involve system-level or technical solutions, such as engineering controls or forcing functions, which reduce dependence on individual performance and are therefore considered more robust and sustainable. Intermediate recommendations focus on modifying existing processes, practices, or organizational arrangements, for example through standardization or changes to guidelines. Their effectiveness is considered moderate, as they still require consistent implementation in practice. Weak recommendations primarily aim to influence individual behavior, such as through training, education, or reminders. As these approaches rely heavily on human performance, they are generally considered less reliable in achieving sustained safety improvements.4,12
The classification of recommendation by incident type.
Each report was treated as a single analytical unit and assigned to one primary incident type based on the main event described. Recommendations were analyzed as individual units; a single report could include multiple recommendations, each of which was extracted verbatim and analyzed separately while retaining its link to the corresponding incident type. When analyzing the investigation reports, we also examined the investigation methods used in the regions participating in the study and noted any references to the planned implementation and follow-up of recommendations.
Prior to classification, each author conducted a thorough review of the data set. Both the incident type classification and the strength of recommendations evaluation were performed independently by the two authors, with any ambiguous cases flagged for joint discussion. Final classifications for both processes were determined through consensus and subsequently rechecked independently to ensure accuracy and consistency.
Descriptive statistics were used to summarize the data. Recommendations were analyzed by incident type (Table 1), and frequencies and percentages were calculated for recommendation categories and subcategories (Table 2).
Recommendations by category.
Note: In the intermediate category, no recommendations were made in the classes “Eliminate look-alikes and sound-alikes” or “Eliminate or substitute system/device.” In the strong category, no recommendations addressed “High reliability training.”
Results
The reports (n = 169) included 12 different event types and a total of 525 recommendations, with a mean of 3.35 (SD = 1.49, range = 1–23) recommendations per report. The most frequent incident types—medication-related (n = 33) and treatment or monitoring (n = 27)—generated the highest numbers of recommendations (113 and 96, respectively). In contrast, lower frequency categories, such as aseptic or healthcare-associated infections (n = 4) and violence (n = 3), showed the highest recommendation intensity, with means of 6.5 and 6.0 recommendations per investigation. Overall, while high-frequency incident types contributed the majority of recommendations, certain rare incident types produced more recommendations per investigation (Table 1).
Each recommendation was classified into a single strength category. Most of the recommendations were categorized as “weak” (n = 292, 56%) and were focused mainly on staff training (n = 126, 43%) and new procedures/memoranda/policies (n = 113, 39%), with these two classes together representing 46% of all recommendations. Only weak recommendations were made in 27 investigations. The second most frequent category was “intermediate” (n = 192, 37%), and the vast majority of these recommendations targeted “Enhanced documentation/communication” (n = 120, 63%). Only a minority of the recommendations were categorized as “strong” (n = 41, 8%), with the largest proportion (n = 12, 29%) targeting “Engineering control or interlock (forcing functions, environment, work area design)” (Table 2).
Most of the strong recommendations were found in medication-related reports (n = 11). Conversely, the fewest strong recommendations were made in the categories Related to an operative procedure (n = 2), Related to organization of care (n = 2), Asepsis/Healthcare-associated infection (n = 2), and Related to violence (n = 2). No strong recommendations were made in the Other category (Table 1).
Approximately one in four investigations (24%) included at least one strong recommendation. In six investigations, more than one strong recommendation was issued, up to a maximum of four. Of these six cases, two addressed care follow-up, one medication safety, one suicidality, one device safety, and one patient accidents.
In 33 reports, mention was made about the implementation of recommendations, and in 12, at least one recommendation included a note regarding the follow-up of its implementation.
None of the investigation reports (n = 169) described the method that was used for the investigation.
Discussion
This study examines serious incident investigation reports (n = 169) conducted by internal, independent, and multidisciplinary teams within the social and healthcare systems of Finland's well-being services counties. Approximately half (11 out of 23) of the requested organizations provided reports, representing well-being services counties from across Finland. Although broadly representative, the reports differed in quality and completeness. None of the reports specified the investigation methods used. Analysis revealed that most recommendations were categorized as weak (56%) and predominantly addressed staff training and new procedures or policies, with only a minority being categorized as “strong” (n = 41, 8%). These findings are consistent with previous research showing a predominance of weaker recommendations in incident investigations.13,14 From the perspective of the hierarchy of controls, this distribution is notable, as system-level interventions are considered more effective than person-centered approaches, suggesting that current recommendations may have limited potential to achieve sustained improvements in patient safety. 1
Recommendations should be linked directly to the root causes identified. In simpler or shorter investigations, recommendations and solutions can often be developed simultaneously; however, in more complex cases, recommendations may serve as the basis for later action planning and the development of solutions, which may be carried out by a different or newly assembled team. 15 In our data, contributing factors are described in almost all of the reports, and the method of analysis is not specified in any of them. In this study, contributing factors were not systematically analyzed as separate variables, but this inconsistency was noted during data review. It was very difficult to understand how the investigation team had arrived at the recommendations made, and there was not always a clear connection between the description of the incident and the recommendations. These findings are consistent with the report Recommendations but no action: improving the effectiveness of quality and safety recommendations in healthcare by the Health Services Safety Investigations Body (HSSIB) which similarly identified challenges in linking analysis to recommendations and highlighted limitations in the effectiveness and implementation of recommendations. 16
Our study supports previous studies in showing that most recommendations were classified as weak and predominantly focused on staff training or the introduction of new procedures, memoranda, or policies.13,14 Emphasizing staff training and policy revisions is increasingly regarded as a weak and insufficient response to the complex challenges of the modern healthcare system, and it is therefore crucial that action plans are not built solely on weak recommendations, but that they also include at least some measures that can be considered intermediate or stronger in order to ensure more sustainable improvements.4,9,17 These findings may also be influenced by broader organizational and behavioral factors identified in previous research, such as limited resources, hierarchical decision-making structures, and challenges in translating analytical findings into system-level actions.8,9,17
In our study, investigations yielded a mean of three recommendations per case, with wide dispersion. A few investigations generated substantially more—up to 23 recommendations, the majority of which were classified as weak. This pattern raises concerns about the appropriate volume of recommendations, the need for prioritization, and the realism of implementation in routine practice, with similar observations having been documented in earlier research.8,17,18 Because a single adverse-event investigation typically requires dozens of person-hours, 3 producing large numbers of predominantly weak recommendations is not consistent with the reality of constrained resources and ongoing reform and cost containment pressures in Finnish health and social care. Raising the quality of recommendations—fewer, stronger, system-based, and clearly prioritized—thus becomes essential to ensure that limited investigative efforts translate into feasible implementation and measurable safety gains.
Applying the SMART criteria more systematically offers a pragmatic route toward strengthening the quality of investigation recommendations and shifting them from predominantly weak, behavior-focused measures toward system-based interventions. 15 Although the Finnish national guidance for investigating serious adverse events highlights the use of SMART criteria, 5 uptake in practice seems limited.
Only 33 of the 169 investigation reports mentioned the implementation of recommendations, and only 12 included any note on follow-up. This raises questions about the extent to which investigations lead to concrete changes in practice. However, implementation and follow-up may be documented outside the reports, and responsibility for these actions often lies with operational management rather than the investigation team. In addition, the absence of standardized national reporting guidance leads to variation in report content and limits the ability to assess the impact of investigations. Overall, the available data do not allow reliable conclusions about whether serious incident investigations result in system-level changes, highlighting the need for more consistent national practices to ensure the visibility of recommendation follow-up.
Strengths and limitations
To our knowledge, this is one of the first studies to systematically examine the strength and characteristics of recommendations arising from serious incident investigations in Finnish health and social care.
The study was based on serious incident investigation reports obtained from well-being services counties across Finland and covering a wide range of incident types and care settings. Although reports were received from approximately half of the organizations contacted, the participating organizations represented different geographical regions of the country, and the incidents examined are not unique to any specific organization or specialty. Therefore, the data provide a broad and realistic picture of current practices related to recommendations arising from serious incident investigations in Finnish health and social care.
Several limitations should nevertheless be acknowledged. The investigation reports varied considerably in their structure, level of detail, and overall quality. As the analysis relied on documented information, some recommendations, implementation plans, or follow-up activities may have been undertaken in practice without being recorded in the reports. Furthermore, the study examined recommendations as documented in investigation reports rather than their actual implementation or effectiveness. Finally, because the investigation of serious incidents is not mandatory in Finland, the number of reports included in the study should not be interpreted as reflecting the true incidence of serious adverse events nationally.
Conclusions
This study reveals that most recommendations emerging from serious incident investigations in Finland's health and social care system are weak and focus on individual behavior rather than system-level change. Implementation and follow-up are rarely documented, and the lack of a standardized investigation framework limits learning and impact assessment. Strengthening investigation practices requires the use of systematic analytical methods, the prioritization of strong system-level actions, and consistent follow-up mechanisms to ensure that investigative findings lead to measurable safety improvements.
Footnotes
Ethical considerations
In Finland, the final report of a serious adverse-event investigation is usually a public document, unless it contains confidential information. The reports for this study were requested from various organizations in such a way that they did not contain confidential information and were fully anonymized. Consequently, no ethical approval or research authorization was required.
Consent to participate
Not applicable.
Consent for publication
Not applicable.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research has received funding from the State Research Funding (VTR), granted by the Research Committee of the Western Finland Cooperation Area.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability
When requesting the material, the researchers committed to using it solely for research purposes, meaning the material is not publicly available.