admission/discharge/transfer: (introduced in )

(ADT) real-time  data feeds that carry patient demographic and visit information. Whereas claims data are generally retrospective, ADT data can be used to trigger events.

ADT: (introduced in )


all payer claims database: (introduced in )

aggregate claims data set of multiple payers or providers in a single area.

APCD: (introduced in )

all-payer claims database

Back to Top


big data: (introduced in )

any collection of data sets so large and complex that it becomes difficult to process using on-hand database management tools or traditional data processing applications.

Back to Top


Camden Health Database: (introduced in )

a citywide, all-hospital claims database that includes data for all hospital encounters dating back to 2002. the database, and the analysis subsequently produced, was the origins for the hotspotting toolkit.

CHDb: (introduced in )

Camden Health Database

claims data: (introduced in )

historical data obtained from a hospital which includes records for each visit to the emergency department or inpatient stay. Claims data includes patient information, dates of service, provider speciality, place of service, diagnosis and procedural codes, and payer information.

cost curve: (introduced in )

the cost of healthcare over a period of time as displayed on an x-y axis. When a community has a large number of superutilizers associated with a disproportionate amount of spending, the graph has a negative skew and an exponential cumulative distribution.

Back to Top


data sharing agreement: (introduced in )

a formal agreement that allows a hospital to share its data with a third party / outside organization to facilitate care coordination for superutilizers. Often involves a Business Associates Agreement (BAA) and a Collaborative Services Agreement (CSA).

Back to Top


encryption: (introduced in )

the conversion of information and data into a form that cannot be easily understood by unauthorized people. encryption does not prevent information from being intercepted, but denies the message to the interceptor.

Back to Top


geographic hotspots: (introduced in )

high concentrations of cost and hospital utilization by patients in a geographic area. The process is completed by geocoding the patient address from the claims database and probabilistic matching.

geographic information system: (introduced in )


GIS: (introduced in )

geographic information systems are a category of software that capture, store, manipulate, analyze, manage, and visualize geographic and spatially referenced data. This includes but is not limited to the production of maps. In hotspotting, GIS can be used to discover and display the unequal distribution of health variables within a geographic community. It can also be used at the data cleaning stage to standardize and correct patient address data.

Back to Top


health information exchange: (introduced in )

(HIE) the electronic movement of health-related information among multiple  organizations to facilitate access to and retrieval of clinical data with the goal of providing safer, timelier, efficient, effective, equitable, patient-centered care. HIEs can also provide a daily report of currently hospitalized individuals classified as high utilizers to identify potential program participants and ultimately track program processes and outcomes.

HIE: (introduced in )

health information exchange

hotspotting: (introduced in )

a data-driven process for the timely identification of extreme patterns in a defined region of the healthcare system. It is used to guide targeted intervention and follow up to better address patient needs, improve care quality, and reduce cost.

Back to Top


institutional review board: (introduced in )

allows you to conduct research on your superutilizer intervention.

IRB: (introduced in )

institutional review board

Back to Top


matrix clustering: (introduced in )

a method of visualizing data to identify trends and relationships that involves plotting two (or more) variables against one another to find areas of significant overlap.

multifactor authentication: (introduced in )

an approach to identity verification requiring the use of two or more of:

  1. something only you know
  2. something only you have, or
  3. something only you are.

These are called “authentication factors.” A traditional password is type 1, something you know. Websites like LinkedIn, GMail, and Twitter allow you to add a second factor like a code texted to or generated by a verified phone (type 2, something you have) that you must type on a website before you can log in. Some computers have fingerprint scanners to add type 3, something you are. Systems that rely on multiple factors offer stronger protection than passwords alone. It is harder for someone to lose, carelessly share, fake, or steal 2 or 3 types of authentication at once.

Some encryption and password management software can enhance the protection of your sensitive data by providing or approximating multifactor authentication by allowing you to require the presence of a particular USB device ( e.g. YubiKey is a well regarded option, but there are others) or an arbitrary, pre-selected (but otherwise unremarkable or hidden) computer file called a “key file.” Our recommended TrueCrypt 7.1a, KeePass, and Password Safe all provide this feature.

Back to Top


probabilistic linkage: (introduced in )

also called fuzzy matching, refers to the process of identifying common individuals within and across data sets who lack a common ID and would not be identified through deterministic (exact) matching techniques. probabilistic matching involves algorithms that create a probability that two records are plausibly related to one another. there are a number technological tools to help facilitate the process.

Back to Top


relational database: (introduced in )

an electronic data storage structure in which information about different types of entities (e.g., patients, visits, insurance providers…)  is stored in separate tables. Each row in a table is a distinct instance of that entity type and the relationship between entity types / tables is defined by specific rules ( e.g., patients can 1 or more visits; visits can only have 1 patient). Relational data ideally captures more information and keeps it better organized than “flat” data, but it can be less flexible and takes more skill to manipulate.

Back to Top


secure file transfer protocol: (introduced in )

a process for transferring files, SFTP  encrypts your data, preventing passwords and sensitive information as it is shared across a network.

segmentation: (introduced in )

the process of defining and subdividing a large population into clearly identifiable subgroups  having similar needs, wants, or demand characteristics.

superutilizer: (introduced in )

a small subset of the population that accounts for a disproportionate percent of health care expenditures; often these are patients who suffer from multiple chronic medical conditions compounded by an array of social, behavioral health, and environmental factors.