Research with Large Databases: Our Experience with the National Trauma Databank

Introduction

With healthcare delivery rapidly evolving, large data driven research is becoming increasingly relevant to clinical practice. The National Trauma Database (NTDB) is the largest trauma registry in the United States and is composed of data from over 800 hospitals. The primary purpose of the NTDB is to provide a standardized trauma registry to inform clinicians, policy makers and other stakeholders about current challenges, trends and outcomes in trauma care. As the largest trauma registry, the NTDB offers a broad array of data for trauma patients and can serve as a pivotal tool for clinicians, researchers and policymakers alike. There are also similar databases to the NTDB that exist as well. We share our experience with the NTDB as an example of how powerful big data can be and about the advantages of using large databases for research. With all things being considered, asking research questions in the NTDB can be challenging at first. This brief letter seeks to shed light on the utility of the NTDB and how to approach asking questions within this database.

Size and Scale

With millions of patient records, the NTDB offers a comprehensive view of trauma care in the US and is able to capture the experiences of diverse patient populations across urban and rural landscapes. The NTDB captures key demographic, pre-hospital, hospital and outcome data all of which contribute to the breadth of what the NTDB has to offer. Principally, the NTDB aggregates a wide array of patient demographics, injury data, treatment measures and patient outcomes. These features place the NTDB in a unique position in guiding research, policy and clinical practice in trauma medicine.

Approaching the Research Questions

The NTDB is available through the American College of Surgeons. We were able to obtain access through our trauma surgeons. After acquiring the database, researchers have access to a wealth of data. For example, a dataset for a single year can contain millions of patient records. This presents a significant challenge and for this reason researchers commonly find it helpful to use a programming language like R or Python to both view and extract the relevant data. These programming languages also commonly have IDEs or Integrated Development Environments like R Studio that can aid investigators throughout this process.

The first step to approaching the NTDB is to ask: What is your research question? Who are your patients? How will you define these patients? For example, if you are interested in trauma patients who received a certain procedure, you can identify these patients by querying the database for patients with an ICD Procedure Code that matches the procedure of interest. How you will define your population is critically important as it will determine what parameters in the database can be useful for you.

Another common technique to identify patients is through the use of ICD diagnosis codes. Every patient in the NTDB has their respective ICD diagnosis code entries that occurred during their hospital stay. However, it is important to note this method is imperfect as specifics of a particular diagnosis for a patient are typically not present in the database.

It is important to identify your target population and gradually refine that population based on clearly defined inclusion and exclusion criteria. This is a critical framework to keep in mind: in order for the data to be manageable for analysis it needs to be filtered by both the population of interest and the parameters you wish to measure. If this aspect of the study is well defined, data analysis can be performed with ease.

Advantages of the NTDB

The key advantage to using the NTDB is that it provides easy access to a large, reliable dataset. All data entries conform to the NTDB standard which ensures consistent data entries are made across institutions. Because the NTDB is a large de-identified national database, an IRB is often not required. Due to the robustness of NTDB, you can ask questions about rare disorders in the context of trauma. The strength of the NTDB also comes from its ability to ask high level questions about health outcomes in different populations.

Limitations

It is also important to realize the NTDB can have limitations, including susceptibility to biases and confounding variables due to the NTDB’s retrospective nature. These challenges are common to many retrospective database studies and it is important to keep them in mind when interpreting data from the NTDB. Additionally, yearly variations occur due to a myriad of factors and this necessitates careful interpretation of data across different years. Researchers should keep all of these factors in mind when designing studies with large databases.

Conclusion

The NTDB serves as a vital tool for trauma research and quality improvement in the US. Its expansive scope and scale offer an unparalleled view into the realities and challenges of trauma care across diverse settings. While it is important to recognize the inherent limitations to the NTDB, it remains an indispensable tool to inform clinical practice.