Lesson 13: Clinical Data Repositories

Clinical Data Repositories

Nowadays, most institutions have existing clinical data repositories (CDR), in electronic or
written format, to represent an aggregated database of clinical information. The repositories
usually house a multitude of laboratory results, diagnostic reports, and various clinical
documentation. These data are readily searchable and exportable, often because the information
is gathered from standard clinical care procedures (Robertson and Williams, 2016). The
repositories integrate physician-entered data with data from different existing information systems
including laboratory, radiology, admission, and pharmacy. They are placed where both clinical
data and other data of interest, such as external data sources and financial data, are assimilated
(Carter, 2001).
A clinical data repository can successfully depict the same sample across different points in
time, from varying sources both within and outside the health institution. Common kinds of
available information in the CDR are listed below:

✓ Patient demographics ✓ Immunizations

✓ Patient's primary care provider ✓ Diagnoses
✓ Medication list ✓ Procedures
✓ Allergies ✓ Laboratory results
✓ Hospital in-patient visits ✓ Social history
✓ Emergency department encounters ✓ Vitals
✓ Outpatient practice visits

Maintaining them poses a lot of advantages since the longitudinal view of a patient's medical
record can assist in improving patient experience, and having information about prior test results
and procedures leads to more informed patient care decisions and helps avoid redundant
treatment.

CDR Integration with HIS

Bergeron (2013) describes a clinical data repository as a systematically structured and
gathered "storehouse" of patient-specific data, which is usually mirrored from a clinical
application, or supplemented with data from other clinical systems. Since it is maintained as a
separate database particularly created in aiding decision analysis, the main application avoids
computational loading, and response time to a query is improved. Moreover, because almost all
simple, customized, or complex patient records are mirrored in it, queries are possible without
sacrificing the performance of the source applications. Furthermore, since the data typically
originated from one source with little to no manipulation, near real-time retrieval of clinical data is
possible.
There are different levels of integration in the said repository. These levels depend on
locations, indices, catalogues, semantic translations or equivalences, syntactic structures, and
links to external information. They influence functionality by setting constraints on how easily
someone can make queries for any of the contents. The integration structure from the CDR to the
laboratory information system, for example, may be different for radiology or pharmacy. These
differences might mean that a user may only access certain types of information, and use
particular types of queries from a given information system depending on the restrictions which
were set during the integration process. The aforementioned restrictions vary in terms of the
user's access. While some CDRs are fairly open, others are restricted either to employees of an
institution or to members of some research network. Access control to the CDR functions as a
safeguard to uphold data security and integrity.
Wade (2014) emphasizes that the longitudinal nature of the CDR requires a way of linking
various observations of the same identified subject. Most repositories usually contain personally-
identified data; however, due to privacy issues, they only release de identified data which can
lead to the omission of some data in a dataset. The lack of identifiers could also prevent the linking
of data for some patients. Presented below (Table 13.1) are the different types of clinical data
repositories that Wade classified according to factors described above.
Table 13.1 Types of clinical data repositories
Repository Type Definition
Study A database that collects observations for a specific clinical
research study
Electronic Health Record A database of observations made as a result of direct health care
Registry Observations collected and organized for the purpose of studying
or guiding particular outcomes on a defined population, associated
studies are either multiple or long-term and evolving over time
Warehouse A repository that adds levels of integration and quality to the
primary (research or clinical) data of a single institution to support
flexible queries for multiple uses; is broader in application than a
registry
Collection A library of heterogeneous datasets from more organizations than
a warehouse or more sources than a registry; organized to help
users find a particular data set, but not to query for data combined
across datasets
Federation A repository distributed across multiple locations, where each
location retains control over access to its own data, and is
responsible for making the data comparable with the data of other
locations

These repositories are beneficial in consolidating patient information, a disadvantage is that

most CDRs are only integrated with clinical data. Laboratory results, diagnoses, and
demographics might be available in one platform, but overall patient satisfaction, the amount of
time a patient had to wait before being treated, and other information not directly related to patient
care might be unavailable.

Multiple Views for Patient Medical Record

Information on patients is typically scattered across multiple subsystems. A clinical data
repository standardizes data from disparate sources into a cohesive format. It comprises
numerous tables which offer a partial view of patient information (Gensinger, 2014). Its structure
allows data to be extracted along dimensions such as time (by year, month, week, or day),
location, or diagnosis. These data can often be accessed in smaller units within the same
dimension. For instance, a user can view the number of patients with a certain type of diagnosis,
laboratory result, or prescription within a year, then a month in that year, and further into a day in
that month. One can also access how many times a particular procedure has been performed at
all locations within a health system, and then see the aggregate amount per region or per facility.
It helps organizations to transform large amounts of information from distinct transactional files
into a unitary decision-support database (Wager, Lee, & Glaser, 2013).
Ball and Douglas (2013) elaborate that a well-deployed clinical repository has multiple
advantages. First advantage is the CDR function to provide longitudinal views of patient
information. It is often organized primarily around patients and secondly around visits or
encounters, a method that easily accommodates views that span multiple visits. This allows
clinicians to trend and chart results independent of the visits and test panel organization. For
example, a clinician could study the trend of a patient's blood sodium levels over the past six
months independent of other factors.
Another advantage is its capability to provide access to information when needed. Since it
receives information from a multitude of feeder systems, it can create a "one-stop shopping
environment. This is done by allowing the clinical staff to access a variety of patient-focused
information through a consistent and easy-to-use graphical interface (GUI). The GUI access can
be deployed through hand-held devices, bedside computing devices, computers in physician's
offices, or computing devices in nursing stations. In any case, this wide variety of information
access moves far closer to deployment of information at the point of care.
Finally, CDR provides a cross-continuum view of information since it allows information to be
gathered and viewed from sources other than an acute setting. This type of ambulatory-focused
information combines with the acute information to give clinicians a new level of insight into the
wellness of their patients.

Data Visualization of Laboratory Results and Vitals

Data collected through an electronic health record system may be retrieved at the request of
an authorized user, whether a physician, medical technologist, nurse, or radiologist. The
electronic health record may present patient care information as text, tables, graphs, sounds,
images, full-motion video, or signals on an electronic screen, phone, pager, or paper (Bronzino &
Peterson, 2014).
Unfortunately, analyzing trends and patterns from large data sets can be a challenging
process. This is where data visualization, the art of representing data in a pictorial or graphical
format, becomes useful. It helps in simplifying a wide array of information, and it allows decision-
makers to derive analytical results from the information presented visually. Through this,
correlations, patterns, and trends which might be undetected from text-based clinical data can be
revealed and recognized with more ease.
For example, the physician can easily review the results of multiple chest x-rays obtained over
the course of months or years when deciding if a nodule has grown. This information would not
have been accessible when basing on textual clinical data alone. Moreover, patients suffering
from specific diseases that require careful charting or monitoring of laboratory values, including
anticoagulation or blood sugar values, can understand what is expected in their care because of
better means of data presentation (Figure 13.1).
 Figure 13.1 Sample Blood Sugar Level Chart

Visualization of clinical data is increasingly becoming an important tool in decision-making.

The graphical representation feature of most clinical data repositories enables scenario analysis,
which helps users use different kinds of filters in order to change the level of information that may
be seen. Common filters include age and gender, in order to assess outcomes of certain
interventions based on isolating certain factors. This kind of analysis is a good opportunity in
empowering the frontline staff by giving them straightforward data which will efficiently and
effectively facilitate the performance of their tasks (Rains & McCuistion, 2018).

Key Points to Remember

✓ Clinical data repositories (CDR) integrate physician-entered data with data from different
existing information systems including laboratory, radiology, admission, and pharmacy.

✓ A CDR is a systematically structured and gathered "storehouse" of patient-specific data,

which is usually mirrored from a clinical application, or supplemented with data from other
clinical systems.

✓ Repository types include study, electronic health record, registry, warehouse, collection,
and federation.

✓ CDRs offer a cross-continuum view of information since they allow information to be

gathered and viewed from sources other than an acute setting.

✓ Data visualization helps in simplifying a wide array of information, and it allows decision-
makers to derive analytical results from information presented visually.