Objectives To systematically investigate the causes and severity of incidents with mechanical ventilation and intravascular catheters in neonatal intensive care units (NICUs) in the Netherlands, in order to develop effective strategies to prevent such incidents in the future.
Design Prospective multicentre survey.
Methods Inclusion criteria were: incidents with mechanical ventilation and intravascular catheters reported to a voluntary, non-punitive, incident-reporting system which had been systematically analysed using the Prevention Recovery Information System for Monitoring and Analysis (PRISMA)-Medical method. The type, severity and causes of incidents reported from 1 July 2005 to 31 March 2007 are described. Local interventions performed as a result of systematic analysis of incidents are also described.
Results 533 of 1306 (41%) reported incidents with mechanical ventilation and intravascular catheters (n=339/856 and n=194/450, respectively) had been PRISMA analysed and were included in the study. Four incidents resulted in severe harm, 18 in moderate harm and 222 in minor harm. Tube-related incidents accounted for the greatest proportion of harm. 1233 root causes were identified, with most being classified as human error (55%). Of the remaining failures, 20% were organisational, 16% technical, 6% patient-related and 4% unclassifiable. The majority of failures were rule-based errors.
Conclusion Incidents with mechanical ventilation and intravascular catheters occur regularly in NICUs, and frequently harm patients. Multicentre, systematic analysis increases our knowledge of these events. Continuous training and education of all NICU personnel is required, together with preventive strategies aimed at the whole system – including the technical and organisational environment – rather than at human failure alone.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Patients in the intensive care unit (ICU) are usually subject to multiple complex interventions. It is known that complex interventions are prone to error because they require several different steps which depend on technical, organisational and human as well as patient-related factors.1 2 To achieve favourable outcomes, these steps should occur in sequence and without error.3 In the last decade, several studies have been performed to assess the extent of such incidents. A recent prospective cohort study in a paediatric ICU found an association between nursing workload and unplanned extubations.4 Another study reported that patients in adult and paediatric ICUs are harmed by unintended and preventable airway incidents, with important contributing factors in the patients being medical condition and age <1 year.5 The same authors reported that patients are harmed by preventable line, tube and drain incidents, and that these incidents tend to occur more frequently in medically complex patients and in the paediatric population.6 A study of incidents relating to arterial cannulation reported multiple contributing factors for each incident report, such as lack of knowledge, rule-based errors (failure to apply existing rules), high unit activity and lack of supervision.7 However, at present the impact of patient safety incidents in neonatal ICUs (NICUs) is not well known. Studies of incidents in neonatal intensive care are scarce and the majority focus on medication errors.8 In previous work concerning voluntarily reported incidents in NICUs in the Netherlands, we identified 414 incidents with mechanical ventilation (9% of all reported incidents) and 245 incidents with intravascular lines (5%) in a 1-year period. In terms of potential consequences and likelihood of reoccurrence, 52% of incidents with mechanical ventilation and 39% of incidents with intravascular catheters were classified as (very) high risk incidents.9
What is already known on this topic
▶ Complex interventions are prone to error because they require several different steps which depend on technical, organisational, human and patient-related factors.
▶ Patients in the neonatal intensive care unit (NICU) are frequently harmed by incidents with mechanical ventilation and intravascular catheters.
What this study adds
▶ Multicentre, systematic analysis of incident reports, makes the errors that (could) occur in the different steps of the processes around mechanical ventilation and intravascular lines more visible.
▶ Tube-related incidents in particular are a threat to the NICU population.
In order to develop effective strategies to prevent future incidents associated with mechanical ventilation and intravascular catheters in the treatment of neonates, we systematically investigated the causes and severity of these incidents in Dutch NICUs.
From February through June 2005, a Neonatology System for Analysis and Feedback on medical Events (NEOSAFE) was implemented in eight of the 10 Dutch tertiary care NICUs (14–24 beds per NICU) and one paediatric surgical ICU (14 beds, 15% neonates). It was a voluntary collaboration supported by the Dutch Association of Medical Specialists. In total, approximately 3500 neonates were admitted annually to the units.
Data collection and handling
Voluntary, non-punitive incident reporting was introduced to establish specialty-based learning from incidents. Patient safety was defined as “the avoidance and prevention of patient injuries or adverse events resulting from the process of health care delivery”.10 An incident was defined as “any event which could have reduced, or did reduce the safety margin for the patient”.11 A multidisciplinary patient safety committee (PSC) was formed in each unit. Personnel were asked to fill out an incident report form immediately after the discovery of an incident. Incidents were either self-reported or reported by the personnel who discovered them. We developed a standardised report form based on recommendations from the Institute of Medicine.12 The report form included a section to be filled out by the PSC during analysis, and included risk scores that were based on potential consequences and likelihood of reoccurrence (see online appendix A).9 An interdisciplinary meeting provided consensus on incident categorisation and classification of (potential) severity. The PSCs managed an electronic database (Microsoft Access) of reported incidents and results of the subsequent analysis, which were forwarded to the central investigator (CS) on a monthly basis. Patient and staff confidentiality was ensured by excluding personal identification from the electronic database. The local medical research ethics committee (METC Zwolle) was consulted and they confirmed that this study did not require approval for implementation as it only involved the registration of incidents.
Reported incidents with intravascular catheters and mechanical ventilation were systematically analysed by the local PSCs using the Prevention Recovery Information System for Monitoring and Analysis (PRISMA)-Medical method.13 The main goal of PRISMA is to build a quantitative database of incidents (including near misses) and process deviations, in order to facilitate the development and evaluation of system-based preventive strategies. The PRISMA-Medical method consists of three main steps: (1) the causal tree incident description; (2) classification of root causes by the Eindhoven Classification Model (ECM); and (3) formulation of structural measures for improvement (classification/action matrix). Causal trees are used because nearly all incidents have more than one cause. By continuing to ask ‘why’ of each event (beginning with the top event), a structure of causes and consequences arises, until the root causes are identified at the bottom of the tree. These root causes are subsequently classified by linking them to one of the categories of the ECM (see online appendix B). In some incidents, recovery factors can also be identified. However, in this study we focus on the failure factors. As both active failures (human error) and latent conditions (technical and organisational failures) surrounding incidents are systematically considered with the PRISMA-Medical method, the results of this analysis can be used to provide a more realistic view of how the system is actually working.14 15 Each PRISMA analysis was conducted by two members of the local PSC, who were PRISMA-trained and familiar with the department and its processes. Test cases were analysed to measure agreement at three levels of root cause classification. Inter-rater reliability was determined by calculating generalised κ values for each level of classification. Substantial agreement (κ 0.70–0.81) was reached at the main level of root cause classification of the test cases (discrimination between technical, organisational and human failure), and agreement among the committees at the second level (discrimination between subcategories of technical, organisational and human failure) was acceptable (κ 0.53–0.59). Discrimination between different types of rule-based errors (the third level of classification) was more difficult to assess (κ 0.40–0.47).16 Therefore, we searched for the first two levels of root causes (see online appendix B). PSCs were encouraged to analyse incidents within 2 weeks after reporting. We describe the type, severity and identified causes of incidents with mechanical ventilation and intravascular catheters reported between 1 July 2005 and 31 March 2007. We also describe the interventions carried out in participating NICUs. Decisions concerning interventions were made by individual units after PRISMA-Medical analysis of incidents. Descriptions of these interventions and the precipitating incident reports were collected monthly by CS.
During the study period, 1306 out of 9107 incident reports (14%) concerned incidents with mechanical ventilation and intravascular catheters. Of these, 533 incidents (41%) were PRISMA analysed and thus included in the study (table 1).
There were 856 reports of incidents associated with mechanical ventilation (variation between units n=42–318), of which 339 (40%) were PRISMA analysed (variation between units n=8–59; a review of all eligible incident reports, including all unselected cases, showed that mechanical ventilation incidents relating to humidification were relatively under-analysed). Three incidents resulted in severe harm, and moderate harm was reported after 12 incidents. For instance, a machine defect during high frequency oscillatory ventilation resulted in repeated bradycardia and desaturation of a patient, and necrosis of the nasal cartilage was reported in several tube-related incidents. Another 149 incidents resulted in minor harm. With respect to risk scores (based on potential consequences and likelihood of reoccurrence), 153 incidents (45%) were classified as (very) high risk incidents.9 A review of all eligible incident reports showed that there was no significant difference between selected and unselected cases with respect to the level of actual harm. However, with respect to risk scores (based on potential consequences and likelihood of reoccurrence), there was a higher proportion of incidents with high risk scores in unselected cases compared to selected cases (58% of unselected cases vs 47% of selected cases).
Tube-related incidents accounted for the greatest proportion of harm (7% moderate harm and 56% minor harm). They were also assigned the greatest proportion of (very) high risk scores (62% of all tube-related incidents). Most tube-related incidents concerned unplanned extubation (both auto-extubation and accidental extubation by other means, n=44), malposition (either non-tracheal or too deep, n=15) or loosening of fixation (n=13).
Out of 339 incidents associated with mechanical ventilation, 799 root causes were identified using the PRISMA-Medical method (mean of 2.4 root causes for each incident, variation between incident reports: 1–7 root causes). Most root causes were classified as human error (51%). Technical (20%), organisational (19%) and patient-related (5%) failures accounted for the remainder of failures identified, while 5% of failures were unclassifiable. Table 2 shows the distribution of root causes according to the ECM, and proposed actions to prevent their recurrence according to the Eindhoven Classification/Action matrix.13 Most identified root causes were human rule-based failures (47%). Technical external or design failures (9% and 8%, respectively) and protocol-related failures (8%) were also common. Table 3 outlines interventions carried out in individual NICUs as a result of analysing reported incidents. For instance, double-checking of machine set-up was introduced after several reports of incorrect set-up, and education and training on general aspects of mechanical ventilation was intensified after reports of incorrect (written) orders by doctors.
There were 450 reports of incidents associated with intravascular lines (variation between units n=27–102), of which 194 (43%) were PRISMA analysed (variation between units n=11–48). There was no significant difference between the proportion of analysed arterial line incidents and the proportion of analysed venous line incidents. One incident resulted in severe harm: vascular occlusion by an arterial line resulted in foot necrosis and subsequent plastic surgery. Moderate harm was reported after six incidents (for instance, a dislocated intravascular catheter leaking parenteral nutrition into the abdomen, another one leaking parenteral nutrition into the pleural cavity, and excessive bleeding after catheter disconnection). Another 73 incidents resulted in minor harm. With respect to risk scores (based on potential consequences and likelihood of reoccurrence), 82 incidents (42%) were classified as (very) high risk incidents. A review of all eligible incident reports (including all unselected cases) showed that there was no significant difference between selected and unselected cases with respect to the level of actual harm or risk scores.
Out of 194 incidents associated with intravascular lines, 434 root causes were identified using the PRISMA-Medical method (mean of 2.2 root causes for each incident, variation between incident reports: 1–7 root causes). Most root causes were classified as human error (61%). Organisational (22%), technical (8%) and patient-related (7%) failures accounted for the remainder of failures identified, while 2% of failures were unclassifiable (table 2). Most identified root causes were human rule-based failures (55%). Protocol-related failures (11%) were also common. Interventions in individual NICUs as a result of analysing reported incidents included adjusting protocols for intravascular catheters after reports of faulty connections, blood loss after catheter removal and an incorrectly positioned arterial-line transducer causing inadequate blood pressure measurements.
Mechanical ventilation and intravascular catheters are common daily processes in the NICU. Incidents with these treatments frequently harm our patients.
Tube-related incidents in particular are a threat to the NICU population, as they carry a considerable risk for respiratory distress, that is, desaturation and hypoxaemia. Therefore, prevention of such incidents should become part of daily routine in the NICU.
Systematic analysis made the errors that occurred or could occur in the different steps of NICU processes more visible. We identified several weaknesses in the processes necessary for mechanical ventilation. First, many technical external failures were detected, which should be handled at a higher organisational level. Second, we found failures in design that should be discussed with technical experts. Third, there were many deficiencies in protocols which required adjustment. And finally, the majority of failures were human rule-based errors, which should stimulate intensification of training and education. Incidents with intravascular catheters were also often the result of several process weaknesses. As with mechanical ventilation, there were many protocol-related failures. Patient-related failures were also prominent; frequently, unexpected patient movement caused loosening of catheters. However, once again rule-based errors accounted for the greatest proportion of failures. In light of this, we stress the need for continuous training and education, aimed at safer performance of tasks and procedures.
Regarding local interventions carried out during the study period as a result of systematic analysis, several units had already focused on the training and education of personnel on the theory and use of mechanical ventilation and intravascular catheters. This underlines the value of systematic analysis. It also demonstrates that these failures are not limited to one unit. Therefore, we propose that specialty-wide incident analysis should lead to specialty-wide interventions rather than just local interventions. Multidisciplinary, multicentre focus groups may contribute to more thorough investigations to accomplish powerful interventions. This may also lead to better compliance with future preventive strategies. For instance, the results of our study could be used to initiate a collective (re-)education programme for paediatricians and residents in paediatrics on the theory and practice of mechanical ventilation; and critical review of protocol-related failures might contribute to the creation of uniform procedures across NICUs. On the other hand, specialty-wide analysis forestalls superfluous searches for preventive actions that are already used in other units. For instance, one unit mentioned the implementation of different colours for the securing of catheters, after reports of accidental removal of a silestic catheter instead of the peripheral infusion catheter (table 3). This type of intervention has already been used in other NICUs and has also been described previously.17 Although our study would allow comparison of incident types and identified root causes between different NICUs, we did not report this information for reasons of confidentiality.
This study has some limitations. First, several units reported a shortage of time in handling the large number of incidents reported after the introduction of the voluntary reporting system. As a result, only 41% of all eligible reported incidents underwent PRISMA-Medical analysis. Although the units that had expected time-management problems had been instructed well ahead of time to analyse every third report so as to obtain a representative sample of PRISMA analyses, selective analysis may have affected the final results. For instance, a review of all eligible incident reports showed that mechanical ventilation incidents relating to humidification were relatively under-analysed, which may have affected the final profile of root causes. Moreover, there was a higher proportion of mechanical ventilation incidents with high risk scores in unselected cases compared to selected cases. Thus, in our final study, the severity of mechanical ventilation incidents in the NICU is possibly underestimated.
Second, using the PRISMA-Medical method results in retrospective incident analysis. We acknowledge the possibility that post hoc analysis of incidents is subject to confirmation bias as well as other types of biases.
Third, in spite of the systematic collection and analysis of incidents, there are many (known or unknown) co-factors, which make it difficult to assess the beneficial effects of system changes on patient safety in neonatal intensive care. However, improving even one of those factors might lead to safer conditions for patients. Finding and preventing these factors remains an important issue for further studies.
Incidents with mechanical ventilation and intravascular catheters occur regularly in the NICU, and as a result patients are frequently harmed. Multicentre, systematic analysis increases our knowledge of these incidents. Although we stress the need for the continuous training and education of all NICU personnel, preventive strategies in the NICU should be aimed at the whole system – including the technical and organisational environment – rather than at human failure alone.
The authors would like to thank all unit employees for reporting incidents, and in particular the unit patient safety committees for their contribution to the collection and analysis of incident reports. The authors thank the Dutch Association of Medical Specialists for their support in funding this study.
The NEOSAFE study group Neonatal intensive care units: Academic Medical Centre, Amsterdam: JH Kok MD PhD, E te Pas RN; Erasmus MC-University Medical Centre, Rotterdam: H Pas RN, C van der Starre MD; Haga Hospital, The Hague: E Bloemendaal RN, RH Lopes Cardozo MD PhD, AM Molenaar RN; Isala Clinics, Zwolle: A Giezen RN, RA van Lingen MD PhD, HE Maat RN, A Molendijk MD PhD, C Snijders MD; Maastricht University Medical Centre: S Lavrijssen RN, ALM Mulder MD PhD; Máxima Medical Centre, Veldhoven: MJK de Kleine MD PhD, AMP Koolen MD, M Schellekens RN; Radboud University Medical Centre Nijmegen: W Verlaan RN, S Vrancken MD; VU University Medical Centre, Amsterdam: WPF Fetter MD PhD, L Schotman RN, A van der Zwaan RN. Paediatric surgical intensive care unit: Erasmus MC-University Medical Centre, Rotterdam: C van der Starre MD, Y van der Tuijn RN, D Tibboel MD PhD; Other departments: Division of Patient Safety, Hasselt University, Diepenbeek, Belgium; and Faculty of Technology Management, Eindhoven University of Technology: TW van der Schaaf PhD; Research Bureau, Isala Clinics, Zwolle: H Klip PhD, BJ Kollen PhD.
Funding CS was supported by an unrestricted grant from the Dutch Association of Medical Specialists.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.