The objective of this study was to describe in detail the design and data quality of the Danish HEMS database, and to evaluate the coverage of the variables in Hemsfile according to the QIs proposed by the EQUIPE-collaboration group. The study included 13,392 missions registered in Hemsfile in the first 3 years and 7 months with national HEMS. We have no information on missions not dispatched from the EMDCs due to concurrencies or helicopters/crews being out of service.
To our knowledge it is the first study of its kind dealing with key topics for valid and reliable research. Assessment of data quality is important, and a prerequisite, when interpreting results from register-based studies. The overall data quality is considered high with a high degree of data completeness for the large majority of variables.
However, the study has limitations. Hemsfile includes four report forms each containing a specific set of variables mandatory for registration. Some variables, especially those related to patient status, are prone to inter-rater variability. Evaluating this topic is important but was not the focus of this study.
Moreover, missing or incomplete civil registry numbers was observed in 6,4%. This is in contrast to other pre-hospital or emergency studies reporting on missing values in up to 20% of the cases [18]. Further analyses are needed to assess if these patients represent a random selection of the population or can be explained by e.g. tourists without a Danish civil registry number. Thus, how inter-rater variability and missing data might affect or bias future study results are unknown, but must be addressed in future HEMS study designs.
It appears that by adding the visual warning indication system to Hemsfile the amount of patients with a complete report form increased dramatically. This is a simple way of improving data completeness, and when combined with on-going educational efforts, training and audits, the degree of completed reports reached almost 100% at the end of the study period. However, the layout of the database is not completely intuitive and clear. Multiple-choice menus allowing conflicting registrations, e.g. patient carried and completed on scene at the same time, may lead to misunderstandings and errors when used. Furthermore, manually entered data such as timestamps, which are first registered in a paper-based data form and subsequently entered into Hemsfile, are prone to registration errors. A random sample on 35 of the extreme values observed (response time > 60 min) showed obvious errors in the entering procedure suggesting that electronical data capturing could be beneficial. A well designed electronic data collection tool, intuitive and easy in use, not only increases the data quality by reducing the risk of errors and missing values and eliminating inconsistencies, but also saves time and ensures real-time data when integrated in the helicopters. The technological infrastructure of Hemsfile is presently under reconstruction. When completed, misclassifications and technical errors are likely to be reduced.
Comparing the timestamps registered by the physicians with the timestamps registered by the pilots could provide a more precise estimate of the response-times. This validation requires access to the aviation database, NOLAS, which we did not have. Timestamps have much attention and are a central aspect of EMS data collection, as these are well defined for each pre-hospital unit and easy to evaluate. Accordingly, they are often used as QIs. HEMS offer advanced care for patients suffering time-critical conditions where time to initiation of treatment as well as time to definitive care is considered important for the patient outcome. In these cases precise registration of timestamps is crucial in the evaluation of the service. However, not all patients carried by HEMS are suffering time-critical conditions and thus, timestamps do not always reflect quality of care and should not stand alone as single quality measurements.
About half of the proposed response-specific QIs are available from Hemsfile. Hemsfile was initially designed for mission reporting, but without specific reflections on QIs. The selection of variables was based on experiences from a previous pre-hospital database related to the Danish physician-staffed rapid response vehicles. Knowledge and quality assessments from then formed the design process of Hemsfile in 2009–10. The choice of proposed QIs are based on the latest knowledge in the field. This gap between the first introduction of Hemsfile and the development of QIs may explain the modest number of available QIs in Hemsfile. However, some of the lacking QIs are to be found elsewhere in our system, ex. QI 9 and QI 10.
The QI identification and implementation is an initial step towards monitoring quality. Using the experience from the process described by the EQUIPE-collaboration group forms a basis for further discussions on how and to which extent QIs should be implemented in a Danish setting. The pre-hospital field is characterised by patient heterogeneity and system complexity. Measuring the quality of (H)EMS is a challenging task, which is also underlined in the review by Sayed et al. [19], and a broad and wide-ranging approach may be preferable.
Perspectives and future research
Data from Hemsfile has been used in several studies related to the trial periods (2010–2014) covering different topics, including studies on patient outcome, HEMS effectiveness and socio-economics for selected patient subgroups [1, 2, 20,21,22]. The period with national HEMS has been investigated sparsely [23].
Linkage of pre-hospital data with other national public registries and databases through the civil registry number assigned each Danish citizen [24, 25], offers unique opportunities for research regarding follow-up and healthcare management.
Variables concerning the clinical status of the patients such as severity of illness/injury, pre-hospital diagnostics and pre-hospital interventions performed are characterized by a high degree of data completeness (95.8–99.9%). Further evaluation of these performance indicators may contribute in the assessment of dispatch accuracy, which is a key aspect in the overall evaluation of HEMS.
Increasing our knowledge of the Danish HEMS patient population and the HEMS mission profile is fundamental to improve not only dispatch and resource utilisation, but also patient safety and patient outcome.
Collecting data and being able to compare them with data from other services, e.g. through a set of consensus-based QIs covering the pre-hospital patient pathway, is considered an important step in adding a valuable aspect into performance evaluation, research and collaboration. Therefore, providing an insight into our data source is essential for the purpose of facilitating comparison of services nationally as well as across borders in the future.