Descriptions of verbal communication errors between staff. An analysis of 84 root cause analysis-reports from Danish hospitals
- Louise Isager Rabøl1,
- Mette Lehmann Andersen2,
- Doris Østergaard3,4,
- Brian Bjørn2,
- Beth Lilja2,
- Torben Mogensen5,6
- 1Danish Society for Patient Safety, Hvidovre Denmark
- 2Unit for Patient Safety, Capital Region, Hvidore, Denmark
- 3Danish Institute for Medical Simulation, Capital Region, Herlev, Denmark
- 4Department of Surgery and Internal Medicine, Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark
- 5Department of Orthopaedics and Internal Medicine, Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark
- 6Hvidovre Hospital, Capital Region, Hvidovre, Denmark
- Correspondence to Dr Louise Isager Rabøl, Danish Society for Patient Safety, Hvidovre Hospital, Kettegard Alle 30, DK-2650 Hvidovre, Denmark;
- Accepted 4 July 2010
- Published Online First 5 January 2011
Introduction Poor teamwork and communication between healthcare staff are correlated to patient safety incidents. However, the organisational factors responsible for these issues are unexplored. Root cause analyses (RCA) use human factors thinking to analyse the systems behind severe patient safety incidents. The objective of this study is to review RCA reports (RCAR) for characteristics of verbal communication errors between hospital staff in an organisational perspective.
Method Two independent raters analysed 84 RCARs, conducted in six Danish hospitals between 2004 and 2006, for descriptions and characteristics of verbal communication errors such as handover errors and error during teamwork.
Results Raters found description of verbal communication errors in 44 reports (52%). These included handover errors (35 (86%)), communication errors between different staff groups (19 (43%)), misunderstandings (13 (30%)), communication errors between junior and senior staff members (11 (25%)), hesitance in speaking up (10 (23%)) and communication errors during teamwork (8 (18%)). The kappa values were 0.44–0.78. Unproceduralized communication and information exchange via telephone, related to transfer between units and consults from other specialties, were particularly vulnerable processes.
Conclusion With the risk of bias in mind, it is concluded that more than half of the RCARs described erroneous verbal communication between staff members as root causes of or contributing factors of severe patient safety incidents. The RCARs rich descriptions of the incidents revealed the organisational factors and needs related to these errors.
Patient safety is still a major problem at many hospitals all over the world. Poor teamwork and communication between healthcare staff are correlated to patient safety and adverse events.1 Team training2 and standardising of verbal communication3 have been suggested as methods to improve staff communication and thereby patient safety. However, the existing descriptive studies of hospital staff communication have been labelled as non-exhaustive and failing to reveal the systemic factors leading to the event. This inhibits the ability to suggest appropriate interventions.4 It has therefore been recommended to add depth to the studies of communication error by exploring the objectives, communication tools, community affiliations, rules and division of labour for all the individuals involved in the patient care team.5
The method originates from aviation and was given a platform in healthcare by the Veterans Affairs National Center for Patient Safety8 and The Joint Commission of Health Care Accreditation.9 It is a systematic interactive process following a prespecified protocol and performed by a multiprofessional team whereby the sequence of events and the organisational factors that contributed are identified. The result is a detailed report (RCAR) based on the incident report, the pertinent written medical documents, interviews with involved staff members, human factors thinking and consensus that describes communication, environment, training and competencies, equipment, safety barriers, procedures and guidelines related to the incident.7 The RCA method can thereby contribute to the broader look at communication factors that is being called for4 5 and which is lacking from individual interviews and analysis of incident reports, for instance.10 11 Furthermore, it has the explorative approach that is lacking from observation studies12 and questionnaires.13
Despite the disseminated use of RCAs in American, British and Australian healthcare systems among others,14–16 there are few indications hereof in the literature.17 This might have to do with confidentiality issues or the bias-issues related to the RCAs. The latter will be addressed in detail in the Discussion section.
When developing a team training programme, an assessment of the needs at organisational, team and individual level is necessary.18 We speculated on whether RCARs could be used to explore the organisational needs for verbal communication support. The objective of this article is therefore to review RCARs for descriptions of verbal communication between staff as a part of a needs assessment before developing a team training programme to strengthen patient safety.
Accessing and selecting reports for this study
Hospitals in the Capital Region of Denmark began conducting RCAs in 2001.19 After adjustments, the method was considered stable in 2004. From 2004 to 2006, 94 RCAs were completed at six hospitals in the organisation. Reports conducted after September 2006 were excluded from this study, as they had the risk of being influenced by increasing focus on communication errors in the organisation.
According to Danish law, the reports are considered documents related to organisational development. As the reports do not contain data identifying the patient, involved staff or the RCA team, they can be accessed for patient safety purposes after permission from The Unit for Patient Safety, The Capital Region of Denmark. This permission was obtained before including the reports.
A pilot analysis on 10 RCARs selected at random was conducted to calibrate the data extraction between reviewers. These reports were excluded from the final data set. This left a total of 84 RCARs, which all included a narrative of the sequence of events, a description of standard operating procedures, root causes and/or contributing factors, as determined by the RCA team and a description of the actions to prevent recurrence.
Extracting data from included reports
Two researchers (LIR and MLA) with substantial experience in rating patient safety incidents independently analysed the event, root causes and contributing factors in the 84 RCARs for descriptions of verbal communication error between staff as causing or contributing to the patient safety incident or near miss. Reports with full inter-rater agreement hereupon were further analysed for the following predefined characteristics:
Was there any description of verbal communication errors in relation to handover (eg, sign-off or transferral)?20
The selection of the above characteristics was based on suggested mechanisms of patient safety incidents and suggested methods to improve verbal communication (see the respective references). After independent analysis, the ratings were disclosed, comparisons were made, and κ coefficients were calculated.28 This was followed by an exploratory review of the RCARs where characteristics of the above verbal communicative challenges were identified. The excerpts characterising the incidents were extracted and translated from Danish to English and inserted in table 2.
The raters agreed upon a description of verbal communication error between staff in 44 (52%) of the 84 reports (κ 0.56). These reports stated a median of two root causes (range 0–7) and one contributing factor (range 0–5) per case. All teams included leaders competent of implementing the suggested actions and consisted of a minimum of three different staff groups. In 42 (95%) of the RCARs, frontline staff were part of the team.
The two raters found a description of handover errors (loss of information at sign-out or transfer) in 35 reports (86%) (κ 0.66) (table 3), communication errors between different staff groups in 19 reports (43%) (κ 0.71), misunderstandings in 13 reports (30%) (κ 0.61), communication errors between junior and senior staff members in 11 reports (25%) (κ 0.44), hesitance to speak up in 10 reports (23%) (κ 0.78) and communication errors in teams with more than two members in eight reports (18%) (κ 0.73).
The exploratory review revealed that the incidents occurred where the communication was unproceduralised (31 cases (table 2, eg, nos 12, 13, 14, 16)). Communication was particularly vulnerable when transferring patients between departments or hospitals (11 cases, eg, table 2, nos 6, 8, 21, 41) or when involving other specialties (for instance during consults) (10 cases, eg, table 2, nos 14, 19, 23, 24). Exchange of information was challenging when it relied on telephone conversation (17 cases, eg, table 2, nos 8, 30, 44).
Error in verbal communication between staff was described in more than half of the cases as a factor causing or contributing to severe patient safety incidents. Communication error in relation to handover was the most frequently described characteristic. This is in agreement with others.20 Handovers were particularly risky when there were no procedures for communication between staff, when patients were transferred between departments or hospitals, when information was exchanged between specialties or when the information exchange was conducted via telephone. These aspects of staff communication were previously not well described. This might be explained by the fact that other methods used in the field (mainly observation and interviews) often only describe communication related to one group or setting.10 12 13 The RCA method allows uncovering of communication weaknesses in relation to organisational procedures, barriers, equipment, training and environment, and as such it fills the need for a deeper understanding of healthcare communication.4 5
Communication errors between different staff groups were frequent as well. This can indicate that the different staff groups have different agendas for the patient which can lead to misunderstandings or are trained to communicate differently.29 However, it probably also indicates that communication between nurses and doctors accounts for a large proportion of hospital communication. In any case, our results indicate that the process needs attention during teamwork and communication training.
In contrast to previous findings, our analysis could not confirm a strong hierarchy and failure to speak up as a major cause of communication errors.24 This can indicate a different culture in Danish hospitals compared with other cultures. As this could mean a limited effect of assertions tools, which aims at enabling staff to speak up, further analysis is needed to confirm this.18
The study has helped to clarify the need for intervention. In order to support teamwork and communication, the organisations need to provide staff knowledge, skills and attitudes about safe information exchange especially during handover, information exchange via telephone, between staff groups and specialties. In a human factors perspective, this will have a larger effect if supported by standardised techniques and checklists.21 30 31 However, targeting staff alone will be insufficient: as these data indicate, a lack of organisational procedures and guidelines establishing who communicates what to whom and when affects patient safety. Establishing and implementing such procedures will increase the chance of team and communication training success.
Except for the findings about hierarchy, which might be a mainly Scandinavian phenomenon, the findings might be applicable to hospitals in general. Healthcare is becoming more complex, and few organisations have the necessary procedures in place to account for this.32
The validity of the review is underscored by the fact that all RCA-teams were multiprofessional, all teams included local leaders, and nearly all had frontline staff members in the teams.
Based on these results, and the fact that RCARs are widely available in many healthcare organisations, we recommend including RCARs in needs assessments for communication and team training curricula and—where necessary—review organisational procedures and guidelines.
Hindsight bias is the major risk factor when working with RCARs: the RCA team focuses on understanding the systemic factors leading to the decisions and actions of the staff members involved but has no direct observations of the event. The analysis relies on frontline staff's memory and written records. And because the analyses are uncontrolled, a verification of the conclusion is difficult.33 The conclusions can further be influenced by leading team members. In this study we therefore excluded studies from late 2006 and onwards, as these had a risk of being influenced by new communication tool agendas.
A second important bias is the risk of confirmation bias: it is easy for both RCA team and reviewers to conclude that an incident could have been prevented with improved communication.4 In this study, this effect was attempted limited by letting two independent reviewers rate the RCARs and select relevant excerpts for others to interpret (table 2). Kappa values between 0.44 and 0.78 show moderate to substantial agreement between the raters extracting the data. However, the ‘less-than-perfect’ value can be explained by the fact that the original purpose of the RCARs was local use: the exact nature of some involved units and the experience of involved staff members were often described knowingly. Furthermore, details about ancillary services and paraclinical specialties were often excluded. If RCARs are to be systematically reviewed for quality and research purposes, thorough descriptions of organisational details must be included, along with a description of the discussions that took place in the team: what causal relations were considered by the team but rejected, and why? This will increase the validity of RCARs.
Finally, there is the problem of selection bias: the selected RCARs are not representative of all patient safety incidents. In the Capital Region, approximately 1% of the reported incidents are considered severe or frequent enough to consider a RCA. Of these, approximately 50% undergo RCA. The numbers are therefore not absolute but can serve as input to a priority list for future patient safety interventions.
The most important strategy to limit the influence of all three bias types, to uncover needs at individual, team and organisational level, and reveal both quantitative and qualitative aspects, is the use of the mixed method design.34 In this case, the RCAR review can for instance be supplied by staff interviews, direct observation and analysis of cultural surveys.
More than half of the included RCARs described erroneous verbal communication between staff members as root causes or contributing factors. Loss of information during handover and between staff groups was described as the most frequent characteristic of the incidents. The related organisational factors were lack of communicative procedures during transfer, telephone communication and involvement of other specialties. With the risk of bias in mind, it is concluded that RCARs holds rich descriptions of patient safety incidents that allows outsiders to gain insight into organisational factors leading to the events.
Funding Det Kommunale MomsfondBredgade 54—Postboks 21811017 København K, The Pharmacy Foundation of 1991.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.