The goal of this video is to go over general concepts about clinical data and how clinical data are generated. This will give us a common vocabulary and framework for understanding specific types of clinical data. The very first piece of vocabulary we should tackle is the difference between EHRs and EMRs. You often hear them used interchangeably, even in this specialization. But there are some important distinctions you should know. An electronic health record, contains records about the entire health of a patient. This includes medical records from multiple providers, radiology images, and laboratory data. Non-medical data like insurance information, patient demographics, genetic sequencing, or even censored data from fitness trackers can be part of an EHR. An electronic medical record, on the other hand, is far more limited. Think of this like an electronic version of the paper chart of the past. These records are usually limited to a single set of providers and typically cannot be easily shared with other medical offices. So, in summary, an EHR is a much broader and more useful tool that an EMR. Regardless, whether we talk about EHRs or EMRs, it's important to understand why we have these tools and what purpose they serve. The first thing that probably comes to mind is for patient care. We need to know what diagnoses a patient has, whether we prescribed any medications for them, and if they have any medication allergies, or if they've had laboratory tests performed or radiology images taken. We want to know how providers made their diagnoses or if the patient has been hospitalized, what happened during the night while the nurses were taking care of them. But that's not the only purpose that EHR serve. EHRs are also used to support billing and payment for the medical care provided. For providers and hospitals to get paid, they must document what medical care was provided, who provided those services and justify why those services were necessary. Similarly, EHRs serve as an important legal record. They document what the provider knew about the patient at any given time by recording who enter data, who reviewed the data, and when. They record the history of everyone who accessed the record to ensure that health privacy rules are being followed. So, EHRs actually serve a double or even triple rule. As such, when we use EHR data for our analyses, we need to understand not just what those data are but how those data were generated. A useful framework you can use are the four W's of each data type. Who recorded the information. Think through the last time you went to the doctor. You probably checked in with someone at the front desk, and then were escorted back to the exam room by a nurse or nursing aide. They likely took some vital signs, things like your weight or blood pressure. Then you saw your provider for your actual visit. If you were at a teaching hospital, you may have even seen a resident before seeing your actual doctor. Perhaps, you needed lab work and interacted with a lab technician, or were given a new prescription that you dropped off with a pharmacist. All of these individuals and more are interacting with and entering data into your EHR. When did they record the information? It's also important to know when information is recorded. For example, many providers don't write or finalize our note about your visit until days later. Laboratory data often have multiple dates associated with it; when it is ordered, when you got the lab drawn, when the result was available, and perhaps even when the provider saw the result. Why did they record the information? Was it for clinical care like the laboratory test order? Or was it a billing code to get reimbursed for the visit or both? Clinical notes are used not only for clinical providers to record what happened in the visit but also for billing to justify why they performed the services they're charging for. They also serve as a legal record to document the providers plans and discrepancies between provider recommendations and patient choices. What did they record? Was it's structured data? Things like a single number or one of only a few options. Or perhaps semi-structured data. Semi-structured data is like a spreadsheet. Data stored in a specific location but that data could be a number, a word, or even a paragraph. Finally, there's unstructured data, completely free form text that doesn't have any obvious structure. Regardless of what the datatype is, remember that the what is almost always impacted by the previous three W's. As we go through the different types of clinical data you can expect to see as a clinical data scientist, we will cover each of these four W's.
