Concepts, utilization, and perspectives on the Dutch Nationwide Trauma registry: a position paper

Before 1999, the Dutch trauma system was nonintegrated and unorganized. Research described in a PhD thesis in 1987 revealed that—at that time—trauma patients were generally transported to the closest hospital, rather than the most appropriate hospital [9]. This thesis and a mid-city airplane crash in the Netherlands in 1992 were the main catalyst that started the development of regionalized trauma care, designated major trauma centers, and the use of field triage protocols. The Dutch trauma system was reformed in 1999, following the American College of Surgeons Committee on Trauma (ASCOT) guidelines entitled “Optimal Resources for Care of the Seriously injured.” [10]. This structural and logistical optimization has had a major impact on trauma care in the Netherlands, as it did in other countries that implemented the inclusive trauma systems [11, 12]. Along with these organizational changes, the changes in treatment approach of severely injured patients was of significant contributory importance [1, 3]. This led to an absolute region-wide reduction in crude mortality of 50% and an absolute reduction in pre-hospital deaths due to exsanguination of 40% [1].

In 2007, the DNTR was established to determine the performance of the renewed Dutch trauma system. The primary goal was quality registration to improve care for trauma patients in the Netherlands. The data collected within the DNTR provide a comprehensive source of information for quality assessment, quality improvement, policy makers, and research purposes.

There are eleven level-1 Regional Trauma Centers (RTC) which form geographically defined inclusive trauma regions. In each region, these centers fulfill a coordinating role that encompasses multiple level-II and III trauma centers. To assist patients’ triage to the appropriate level of care, ambulance services are assisted by a trauma field triage decision scheme based on vital signs, injury type, and mechanism of injury [13]. To further improve prehospital triage, the Trauma Triage App has been developed, which is currently in trial [14]. The level-I trauma centers are fully equipped to deliver the highest level of emergency and surgical care for the most severely injured with 24/7 coverage of all specialties including thoracic and neurosurgery. Within the regional trauma systems, all trauma-receiving hospitals have a direct linkage to a RTC, to facilitate expeditious transfer of injured patients within the network, to the hospital with the medical expertise and instrumental capacity that matches their alleged resource needs. Additionally, four of these RTCs are equipped with 24/7 Helicopter Emergency Medical Service (HEMS) and a Mobile Medical Team (MMT) which are able to dispatch by helicopter or ground vehicle. Lower-level trauma centers (i.e., level-II and level-III), on the other hand, were established to provide optimal care for moderately and mildly injured patients in a cost-effective manner. The current composition of trauma patients admitted to level-I and level-II and III trauma centers is described in Table 1.

In contrast to most other countries, Dutch trauma surgeons are general surgeons with a specialization in trauma orthopedics. As a result, Dutch trauma surgeons treat both visceral and extremity injuries, and this patient mix is therefore represented in the DNTR [3].

The DNTR includes all injured patients who are directly admitted to the hospital through the Emergency Department (ED), transferred to another hospital, deceased during ER treatment, or within 48 h after trauma. Patients declared dead before hospital arrival or without vital signs upon arrival at the ED are not included.

In 2022, there were 86 trauma receiving hospital in the Netherlands. Thirteen were devoted level-I trauma centers, of which eleven acted as an RTC (Fig 1). Between 2007 and 2021, a total of 1.087.046 trauma cases were registered in the DNTR. In 2007, 64% of all Dutch hospitals with an ED participated. Since 2015, the participation rate increased to 100% as shown in Fig. 2.

In 2007, the Utstein template for Uniform Reporting of Data following Major Trauma was established, which formed the basis for the DNTR. The Utstein template suggested that trauma registries should differentiate between data variables that absolutely need to be collected (core data) and the type of additional data that may be desirable (optional data) [15]. The core data of the Utstein template were divided into three groups (‘Predictive Model,’ ‘System Characteristic Descriptors’ and ‘Process Mapping Variables’) based on the role of the data variable in a registry. The complete set of the currently recorded variables in the DNTR is available in Supplementary Table 1–3. Outcomes measures collected in the DNTR include hospital length of stay, intensive care unit length of stay, duration of mechanical ventilation, need for surgery, in-hospital mortality and 30-day mortality. (Supplementary Table 2–3).

Data collection is done by trained personnel or trained medical professionals that work according to a strict protocol. The DNTR is embedded in a web-based relational database (SQL). A trusted third party secures privacy sensitive information and encrypts personal data. Data can be entered through an online data entry application with plausibility checks or by import of an electronic file. The DNTR organizes data entry training sessions to improve the quality of data and attain a high level of knowledge, for instance, e-learnings are made available for learning AIS coding. The system (although basally) regularly checks data quality and outliers; furthermore, during every annual report, results are thoroughly analyzed, and data are checked and validated in case of unexpected outcomes.

Missing values are an indisputably problem of many trauma registries. Yet, before jumping to conclusions or an imputation method, a researcher needs to be aware on how the variable was recorded and which population it concerns. The percentages per variable in the DNTR are listed in Supplementary Table 1–3. Some variables might seem to have a concerning high number of missing values, yet this specific variable might be clinically irrelevant or only relevant for a small subpopulation.

Methodological support should be consulted as there are various methods (e.g., multiple imputation) to deal with missing values and depending on the type of study one method might be more desirably than another. Both multiple imputation and a complete case analysis are viable methods under certain conditions. An alternative imputation using healthy values (e.g., if the GCS is missing, the highest possible score of 15 is recorded) provides a fast and relatively easy option. Although one should keep in mind that depending on the imputed variables, this results in underestimation of trauma, physiological disturbance, frailty, or mortality, but in some situations such a ‘best-case scenario’ might provide valuable insights.

Although the DNTR has already demonstrated its added value for multiple purposes, further development is required [4, 5, 13]. The DNTR aims to record data from the entire acute care chain, starting at the site of injury until hospital discharge or assessment of the 30-day mortality. Prehospital data are subject to improvement, for which collaboration with prehospital personal [e.g. Emergency Medical Technicians (EMT)] should be refined. Furthermore, personal patient data are required in order to follow a patient through the chain (pre-hospital, (inter)hospital, and after discharge). To ensure the quality of the trauma registry, it was decided that individual patients’ consent was not feasible for several reasons; it is time-consuming for the patient and physician, it would greatly increase administrative tasks, and it could potentially result in selection bias. In particular, this selection bias should be avoided if the register is used to asses quality of the entire trauma system. Therefore, it was decided that no individual consent was needed for patients to be registered in the DNTR; patients do, however, have the possibility to ‘opt out.’ [16]. Naturally, the DNTR complies with the general data protection regulation (GDPR). Previously, the DNTR data were pseudonymized at a regional level and uploaded in the DNTR database, making it practically impossible to link data from different health-care providers to a specific patient. Gladly, new Dutch legislations allow the DNTR to include the required citizen service numbers. Only the DNTR has access to the non-pseudonymized data, and privacy is warranted. Researches requesting data from the DNTR for research purposes will not receive reducible data. This recent development offers new opportunities to present individually collected data on an aggregated level, assess the weaknesses, strengths and make substantiated decisions on the future direction of the acute care chain. Furthermore, it might enable us to link citizen service numbers to different data sources, hereby enlarging the available data and further increase the versatility of the data.

The Dutch registry differs from other European national registries by capturing all acute trauma-related hospital admissions regardless of their age, injury type or severity, resource use, or length of stay. Inclusion of all acute trauma admissions has value over more strict inclusion criteria, which results in a very restricted view on the magnitude and impact of injury [4]. Focusing solely on injury severity would result in exclusion of a large percentage of especially elderly and infants. Furthermore, we found that selection of only severely injured and intensive care admitted patients encompasses only 5% of all admitted trauma patients in the Netherlands and would leave out almost 70% of fatalities [4]. Finally, using these criteria would result in a gross underestimation of medical resource utilization in the Netherlands. Capturing all acute trauma admissions is a prerequisite to effectively evaluate our trauma systems’ performance, and it enables policy-makers to make weighted decisions on trauma prevention and control and allocation of workforce and resources [4].

It must be said that registering all acute trauma admissions results in a higher workload and thus more expenses. The demand on data managers for the DNTR is already high, while compared to, for instance, the Trauma register of the German Trauma society (TR-DGU), the number of participating hospitals is rather small. It begs the question whether an all-inclusive system as ours is feasible or at the least cost-effective for larger trauma registries.

All data in the DNTR are gathered prospectively, however, most conducted studies are observational retrospective cohort studies. Until recently, one of the biggest flaws was the inability to include follow-up parameters (e.g., PROMS) in research projects. Most RTCs have started performing routine EQ5D-based health status follow-up for their trauma admissions, but with the new Dutch legislation, the completeness of these follow-up questionnaires is expected to improve even more. These developments will be a great tool to evaluate the quality of trauma and give rise to new possibilities for research to be conducted on data from the DNTR.

Prediction models were already feasible from the DNTR. The scope of possibilities that have now opened up need to be explored in the near future. Studies will always be limited to the included patients, selection criteria, and variables available. Furthermore, 100% of the key variables should be available; otherwise, records should be excluded. Several different methods for data handling were discussed previously.

Recently, the funnel plot methodology was introduced in the Netherlands to monitor and evaluate trauma care on a hospital level [17]. Using a graphical presentation of the standardized mortality ratio (SMR) (i.e., the ratio of observed deaths to the expected number of deaths or the observed mortality rate to the expected mortality rate), the hospital specific quality of trauma care can be monitored [17, 18].

The prediction of survival probabilities for individual trauma patients is essential for trauma system evaluation. Various models have been developed since the introduction of the Trauma Injury and Severity Score (TRISS) [19]. In contrast to other models, the Dutch model (mTRISS-NL) was developed to accurately predict the survival of all acutely admitted trauma patients using widely available core variables. The mTRISS-NL model includes the variables: sex, American Society of Anaesthesiologists physical status and nonlinear transformations of age, systolic blood pressure, Injury Severity Score and the Glasgow coma scale-derived best motor response [20].

All research proposals on the DNTR need to be approved by the Dutch scientific research board. DNTR never provides exports of data and conducts the research analysis themselves. A well-planned study script describing the variables used in the analysis needs to be provided by the researchers, preferably tested in their regional data. It is recommended to consult an epidemiologist before admitting a research proposal and to assist during the (methodological) process of the study.

For the DNTR a board, a scientific advisory committee, a data manager platform, and a program manager have been appointed. Furthermore, the Dutch Trauma Centre Council, composed of leading trauma surgeons from the 11 RTCs, provides their advice for future development and research questions of interest to be performed on the trauma registry. The RTCs receive annual governmental funding to cover expenditures of DNTR infrastructure and wages, providing continuity in sustaining and developing the registry system. One data manager per trauma center, responsible for the coordination of the regional trauma registry, participates in the national data manager platform. Quarterly, the platform discusses cases and definitions of data items to ensure consistency across the regional trauma registries. Furthermore, an online reporting tool is available for the participants including hospital, regional and national benchmark data. Annually national and regional reports are published and handed out at a national conference about the trauma registry results.