The purpose of this lesson is to summarize the ideal data that health care analysts want to inherit and work with. Data that are in harmony as result of centralization and standardization. Given that this ideal is rarely achieved at least early on, it is important to think about why data that seemingly should be the same are often fragmented and unstandardized. After this lesson is complete, you will be able to describe why one particular burn registry had data fragmentation issues, and how a variety of standardization and centralization processes help the team achieve more harmonized data environment. A little work involving health care data often requires that the data are centralized and standardized. For example, consider a health care plan that receives money from the government for improving flu vaccination rates among this population of patients. The problem for one health system was at least 33 percent of the patients received the flu shots outside of the office practice. So, documenting adherence to flu vaccination was problematic. Many of these patients went to non system pharmacies and other external health care organizations. Thus documenting their vaccination status was a significant problem. Generating a unified and complete record on a patient for research or quality improvement will typically require accessing records from multiple systems, clinic systems, pharmacies, and nursing homes. This is even the case for integrated health care delivery systems like tertiary care academic medical centers. This type of health system might seem like a closed system, but many of them have nearly 50 percent of their patients referred to them, and once they receive specialty care, they go back into the community. The same thing applies with the Veterans Administration, where often it is considered a closed system. Many of the veterans receive care from specialists in the community where the VA has referred the patient to the specialist, and the specialists is now documenting in a separate system. As a result, the VA records are often incomplete. To sum up, analytics usually requires complete data from the population, and data that is standardized so that the data elements have the same meaning among all of the patients. Given that this is often not the case, work needs to be done to identify the problems, and document how disparate data can be combined and standardized. Let's look at an example. The National Burn Repository was developed by the American Burn Association. This registry summarizes and compares cases submitted by burn centers internationally as well as in the United States. The concept behind the NBR is simple. Burn centers send a central organization or a repository a standard set of data elements regarding their burn cases. The central repository receives the data, and summarizes the information, form formative reports showing how a hospital system compares with their peers. The hospital system where I work participated in the ABA registry, however, the ABA was in the process of rolling out a new web-based product that was supposed to make it easier for organizations to send them. After the team knew that they wanted to participate in this new web-based process, they decided that they wanted to go back and look closely at the burn patient process, so that they could do better at understanding the data that they were currently collecting. To do this, they look closely at their clinical work flows. This is an example of one of the work flows that data team developed. The entire work flow is too large to show on one screen, but hopefully you can get an idea of the amount of data work flow diagrams can have. The team spent hours talking with those responsible for burn care confirming how documentation is received, and how it's entered throughout the patient's hospital stay. The more the team talk with people, the more they learned how there were many more departments involved in the work flow they were initially believed. By looking at the work flow, the ABA team found 10 entry points into the health system for burn patients. That means that there are 10 different ways that a patient can come into their health system to receive treatment for a burn related injury. The 10 entry points include external points as well as internal points of entry. External patients are transferred from different hospital systems, internal patients came to the emergency department or went directly to the operating room once received in the emergency department. For each of the entry points external as well as internal, the team analyzed what type of documentation was completed or received. In the work flow documentation, the team categorized the type of information that was received. They then organized by the point of entry as well as by key steps in the work flow. As shown here, they found many instances of duplicated data, and the more they investigated, the more they found that duplicate data was recorded using inconsistent formats. They found across the board that the patient's name was entered into multiple areas of the system. The name was written on many paper documents. This is duplication of data because it's being entered or documented multiple times. As the work flow was documented, the ABA team also created a spreadsheet to organize the data and information that they were learning about through their numerous interviews. This was critical because with 10 entry points and dozens of documentation opportunities it was essential to have a very well organized document so that others could discover what had been learned. They also took time to take inventory of where the data was entered, and in many cases they noticed that it was entered in multiple locations. With this document, they added more details. Instead of just saying that something was entered in the HR, they wanted to identify that something was entered in a history and physical note or something was entered in a progress note or perhaps something was entered in a lab section of the HR. Next, the team also wanted to show variation with regards to the providers who created the data. They wanted to show the different people who entered data because sometimes you'll have different roles or a variety of roles entering the same information in different places within the HR. This type of documentation also makes it simple to identify data fields missing in the source data or data fields that are absent from the source data. Later in the analyses, the ABA team used this documentation to help identify the source system for each data element. The summary shown here called the clinical view by the team, was a discussion tool on working with stakeholders. They wanted to have a working document that would help the stakeholders be involved with the process of analyzing data without intimidating them, without losing them in the details. This view is also very important because you'll notice that the ABA team indicated whether or not the data source was structured or unstructured. Later on in the project, the view is immensely helpful in helping them find opportunities for improving data collection by converting those unstructured fields into structured opportunities. There are many ways that data can be documented and organized. This is just one example of something that the team developed to help them in a particularly challenging projects, and in a particularly large project with lots of stakeholders. This documentation also worked because it looked at different views that they had on a spreadsheet, and anyone at anytime could pick up this document and understand what the information means. Even today, a completely new group of people can look at this and understand the project. They can look at this and see opportunities for continued improvement.
