By Jonathan D. Gold, MD MHA MSc FAMIA FHIMSS
April 29, 2024
Challenge yourself:
What defines non-recurring events in healthcare, and what are some examples?
How do chronic events differ from finite duration events, and what makes chronic illnesses unique?
Explain the categorization of acute exacerbations of chronic conditions and their representation on a patient's health timeline.
Why is precision important in determining the onset of chronic diseases, and how can unlinked historical dates affect this determination?
Non-recurring versus Recurring Events
For the most part, events may be classified as either non-recurring or recurring. Non-recurring events include one-time events (like procedures that may only be performed a single time—e.g., appendectomy) and the onset of most chronic disease (e.g., diabetes mellitus, type 1). Recurring events include those events that occur or may occur more than once (like acute disease—e.g., upper respiratory tract infection, medication administration, a lab test, and acute exacerbation of a chronic disease).
Finite Duration versus Chronic Events
Finite duration events are those that are completed or that resolve within a given period. They may be procedures, tests, or medications; or they may be acute or sub-acute problems or acute exacerbations of chronic diseases. Some may recur (like upper respiratory tract infections or blood glucose measurements) or may be non-recurring (e.g., menarche or appendectomy). While some chronic illnesses may resolve after a lengthy period (e.g., chronic otitis media), commonly, chronic events do not resolve (although they may be stable or controlled). These include illnesses like hypertension, chronic kidney disease and diabetes mellitus, and may appear as open ended, dynamic and active on the problem list.
Acute exacerbations of a chronic condition (like “acute exacerbation of rheumatoid arthritis”) possess dual elements—in this case, non-recurring onset of ‘rheumatoid arthritis’ and (potentially) recurring ‘acute exacerbation’. Both elements may be plotted independently on the patient’s health timeline even though there is a clear association between the two. Problems do not always align to one-time, chronic, or acute categories. Atrial fibrillation exemplifies this multiplicity—it may occur as an acute event or may develop into a chronic sporadic or continuous problem.
Non-Recurring Events
One-Time Events
Below is a table showing hypothetical entries for a patient who has undergone an elective total splenectomy[1] on 8/13/2007. Facility A is the office of a surgical group practice that performs the pre-op and after care. Facility B is the local hospital where the procedure is performed. Facilities C and D are specialty clinics (endocrinology and cardiology) which see the patient years after the procedure, in 2010 and 2012, respectively.
The derived date is weighted by the precision of the temporal object. The mean of these weighted dates yields a derived aggregate date. The highest precision date or derived aggregate date may be plotted on the patient’s health timeline to determine patient age at event. Additional record sources, including the PHR, may shift the highest precision date and the derived aggregate date.[2]
Chronic Disease
Another example illustrates how a chronic illness might be captured and how the determination of its onset can be established. Chronic illness, for example Diabetes Mellitus, Type 2, is a continuous condition after the initial diagnosis, hence a different strategy might be considered than that used for one-time events. Because a chronic event will typically be recorded as current, most dates will be associated with higher levels of precision (9 or 12). An unlinked historical date (precision of 10 or 13) may provide a more accurate estimation of a chronic disease’s initial diagnosis.[9] An alternative for determining the first record for a chronic disease is to use the earliest recorded date, no matter what the precision is associated with the various later diagnosis entries.
In the absence of an unlinked, historical date, the best estimate for the onset of Diabetes Mellitus, Type 2, for this patient is the earliest record of it, since the medical record date equals the derived date and the chronic disease is current.[11] Upon inclusion of an unlinked historical date (which is prior to the first recorded disease entry), the date of diagnosis can be corrected. One caveat is when the historical date is partially defined or has no greater precision than year. If that occurs and a different record from the same year shows the chronic disease as current, the derived date may be later than the first recorded date of the disease.[12]
Recurring Events
Acute Disease (with multiple occurrences)
Upper Respiratory Tract Infection, an acute disease, may occur multiple times in discrete episodes for a patient. Acute disease differs in that it is not necessarily a one-time event, nor is it a continuous, chronic one. Unlike one-time events, where we focus on a short span in time, and chronic disease, where identifying the start of the disease is a key, critical factor, acute illnesses typically are distinct, short events. They have beginnings and ends. Acute disease is usually recorded while the disease is active, but often the end of the disease is not documented. The beginning of the illness may be approximated anywhere from days to months prior to the diagnosis and that may be included in the record.
Future and Conditional Temporality
Medical records often refer to future events or conditional contingencies. Future events, like conditional events, commonly include orders and scheduled procedures. While capturing these remains important, the precision related to these must be zero, since they have not occurred. When the event finally occurs, it is no longer “future” or “conditional” and can be regarded in its appropriate event type category.
Conclusion
Classification of events into non-recurring versus recurring, finite duration versus chronic, and future versus conditional categories provides a structured approach to understanding temporal aspects within medical records. While one-time events like elective procedures and chronic diseases present distinct challenges in determining onset and managing data precision, recurring events such as acute illnesses require great care in distinguishing which accounts are for the same event. Future and conditional temporality add layers of complexity, requiring careful documentation and adjustment as events unfold. By navigating these temporal dimensions effectively, we can better interpret patient histories, plan interventions, and provide more accurate data for predictive modeling.
[1] Splenectomy is a point-specific, one-time event.
[2] Incorporating a source veracity modification to the specificity value, reflects the source generating the date (medical facility records, medical claims records, pharmacy records, personal monitoring devices, personal health record, etc.). This allows for patient input when the element already exists from an objective source. In particular, while much of the earlier medical record remains on paper and the start of many chronic diagnoses are not captured fully, adding an additional half point to the precision score from a patient when the element exists in the record gives additional weight to the patient’s documentation of the event.
[3] Future and conditional dates for procedures (or other events), can be plotted on the health timeline, but have no confidence value assigned before or after the event has occurred or the scheduled date has passed since the event may not have occurred when scheduled or may not have occurred at all. Nonetheless, as a reference point, it is important to mark on the age line as an anticipated event.
[4] As Facility B has two entries, one with a confidence level of 9 and the other with a lower confidence level (1), the lower number is dropped from the equation (and subsequently equals 0). Note that this also happens for two of the Facility C entries.
[5] To prevent skewing, when the temporal object for an event is only defined as “occurred”, the first recorded date at a facility is considered the date of occurrence with a Confidence Level of “1”. If later entries in the patient’s EMR for that event have no higher confidence level, then they will not be included when interpolating the occurrence date. Alternatively, if greater specificity is given to the historical date, the unlinked date will have a higher confidence level and the other “0” and “1” confidence level dates will be updated to reflect a “0” confidence level.
[6] The “highest precision date” is simply the date that has the highest level of specificity. If more than one date shares this specificity, it is the average of these. In the above example, this = 8/13/2007.
[7] Interpolated date for event based on derived dates and the precision for each date. The “derived aggregate date” is determined by finding the average for derived dates multiplied by their respective time/date precision. In the above example, the average date is derived by multiplying the derived date times the precision divided by the precision total, i.e., (8/12/2007*7) + (8/13/2007* 9) + (12/22/2010*1) + (7/2/2007 *4) / (7 + 9 + 1 + 4) = 9/29/2007.
[8] Date for event based on first date cited (using all tethered or unlinked results). The “first derived date” is the earliest derived date found.
[9] For instance, when historical information from a paper record is incorporated into an EMR.
[10] Facility A shares an electronic medical record between a hospital and a number of ambulatory clinics (including family practice and endocrinology). Facilities B and C use two different systems, but are linked through the health information exchange with Facility A. The patient-generated personal health record (PHR) also records diagnoses.
[11] For Facility A, this is 4/3/1997; for Facility C, this is 5/17/1998.
[12] For instance, if a facility record states, “Diabetes Mellitus, Type 2 (1997)” (which has a derived date of 7/2/1997), while a linked and current entry for 4/3/1997 says “Diabetes Mellitus, Type 2”.
Comentarios