Diagnosis Wording -- simple, consistent, effective ... dare to try?
Sciences behind a good pathology report
Pareto Principle indicates that for a given cohort of events, about 20% of causes accounts for 80% of occurrence of these events. This principle is applicable to a wide variety of natural and social aspects. In diagnostic medicine, 20% of diseases (diagnostic entities) accounts for 80% of cases encountered in our routine practice. Although study on this topic in diagnostic medicine is yet to be published in English literature, epidemiology of human diseases apparently support the validity of Pareto Principle in diagnostic medicine. For example, a handful of tumors in an organ accounts most cases encountered clinically. In the lung, squamous cell carcinoma, adenocarcinoma, small cell carcinoma and carcinoid tumors makes up greater than 80% of primary lung tumors (WHO blue book). More than 200 recognized liver disease (diagnostic entities) afflict human but less than 20% (i.e. 40) of these diseases accounts for the vast majority (i.e. >80%) cases encountered in our practice for any given subspecialty setting in hepatobiliary medicine (e.g., pediatric liver pathology, transplantation pathology etc.). In fact, data of our unpublished study of more than 500,000 surgical specimens indicate that these numbers are further skewed in diagnostic pathology since pathology cases are from selected groups of patients who needs surgical specimen for diagnosis. This conclusion is well inferred in medical literatures.
This provide an evidence-based foundation and rhetoric rationale to build a concise repertoire of diagnostic wordings of a small number of types of diagnostic entities for our daily practice. Although such diagnostic wording repertoire varies with different specific practice setting and subspecialty, its value is indisputable.
A diagnosis is the specific sematic form of a well-defined concept
A pathologic diagnosis represents a unique constellation of morphologic characteristics, immunophenotype, molecular features and closely related biologic behavior. For example, "basal cell carcinoma" (i.e., diagnosis) is used to indicate a neoplastic lesion composed of basaloid cells arranged in basoloid nests / cords with peripheral nuclear palisading, mitosis and apoptosis, frequent cleft around the nests etc. We use these phenotypic features as diagnostic criteria. In other words, a diagnosis is a short sematic form of such a constellation of disease features. This challenge the practice to include "Microscopic Description" for all diagnoses in pathology reports. So is it necessary to say the same thing in two different formats??
When to Apply: justify whether a microscopic description should be included in the report.
Specimen designation and name of surgical procedures have standardized lexicon
All pathology diagnoses contain specimen designation (topography) and name of surgical procedure. But should we use "Breast, left, upper outer, partial excision" or "Left breast, upper outer quadrant, partial mastectomy"? The Systematized Nomenclature of Human and Veterinary Medicine (SNOMED) provides standard lexicons in diagnostic pathology. SNOMED and ICD-9 provide standardized syntax with designated codes that are used for billing and semantics mapping. The specimen designation (i.e. body parts, anatomic site or topography) should follow SNOMED. For example, “Stomach, antrum” (T-57000, T-32100) vs. “Gastric antrum” (T-57600); “Left breast” (T-04030) vs. “Breast, left” (T-4000, ??). Using appropriate nomenclature and syntax helps create uniform reporting standard that is highly desirable for communication across and between institutions, reducing error and improve reporting quality. This is particularly important for data mining. Those who attempted to do complex search for a specific cohort of cases in pathology laboratory system should appreciate the importance.
Key points to take: Follow the established standards of terminology
When to Apply: determine syntax of diagnostic headings.
Mental models affect comprehension and influence readers' focus of attention.
A mental model represents a person's thought process for how something works. (Susan Carey, 1986). A simplistic and relevant version of mental model is the "expectation" of what and how information should be presented in a pathology report. An oral surgeon may want to know the margin status of a resected tumor. He may want status of multiple margins easily visible at the first glance of the report whereas an oncologist may want to find the size and extent of the tumor (i.e., pTNM) easily. It's been well established that mental model is largely determined by prior experience with similar scenario (or in our case, medical documentation). Although it seems to be a grave challenge to structure and formulate pathology report to meet diverse individual mental models, there are some general rules to follow such as sub-grouping diagnostic information by content focus to address the needs of different interest groups; separating key diagnoses from note/comment or checklist. For example, list all specimen parts with margin status consecutively and put tumor synoptic checklist as a separate group for a resected tumor with multiple parts / margins. These measures help categorize reporting content (to meet different mental models / individualized priority)
When to Apply: . guide efforts to create good conceptual models (see next section).
Conceptual models should be consistent with people's mental models
A conceptual model is the actual model that is given to the person through the design and interface of the actual product, which in our case is equivalent to our pathology report. The importance lies in that if there is a mismatch between a person's mental model (i.e., what he expects or is used to) and the conceptual model (i.e., what and how information are presented in the pathology report), then the content of the report will be hard to process, understand, interpret and remember correctly. That is why we frequently find the reports from outside institutions difficult to read. This relationship of mental model - conceptual model seems to be a "chicken-and-egg" one. Fortunate would be it so. Since nearly 80% of information in a typical medical record comes from pathology, and no accepted standards have been well established. Thus, pathologists are well positioned to take a leading role in establishing the standards for future documentation in clinical medicine although we will also conform our report with consideration of mental models by other prevailing media (see below)
Key Points to take: Expected types of information are easier to read and better appreciated than unfamiliar types. Most people report as writer but not also as a reader -- it's necessary to be in the shoe of readers
Mental & conceptual models influence accuracy and effectiveness of communication / interpretation .
Outline/ hirearchical and logical organization of information establish consistency and promote conceptual model effectively
We all experience difficulties to read pathology reports from other institutions. This is largely because they look different -- the layout, organization, wording etc. -- from what we are used to or anticipate. The differences not only reduce our efficiency to deal with outside cases but can also lead to undue disagreements. This also applies to our pathology reports for our clinical colleagues -- what's presented (conceptual model) fails to match what's anticipated (mental model).
Our mental model is heavily influenced by our experiences with the prevailing media sources such as journal articles, web blogs, newspapers etc. Although each of these media sources may present information in a different style and format, all follow a basic logic ____. For example, most outlines the key points first and then elaborate each points at the nested hiarchical levels and so on, a method now named as "progressive exposure" (ref). Adopting this approach ____ the match of the user's mental models to conceptual model, an important factor in communication (see another section). In fact this has been well adopted in specimen Gross Description of Pathology report . We specify the parts of a multi-parts specimen (e.g., TAH-BSO) or number of lesions in a specimen (e.g., nodules in a mastectomy specimen) before describing each of them. This approach is also favorably / preferably used in this Diagnosis Wording System.
Leading sematic text unit influences reading comprehension and speed
It has been well known that people do not read word by word. Instead, we read and process text information in small group that form sematic unit named chunk, usually 3 - 5 words long. (To learn more, search in Google with "semantics chunking reading comprehension") We all use this semantic chunking to decode the meaning of text information. The semantics of such chunks can be definitive-predictable or neutral-dependent.
"All that glisters is not gold;
Often have you heard that told.… … ”
In this statement from "The Merchant of Venice, Act 2, scene 7" (by W. Shakespeare), "all that glisters" is a neutral chunk since it could be "gold", "not gold" or else whereas "is not gold" is a definitive sematic chunk. Putting a group of neutral (semantic) chunks before a definitive-predictable semantic chunk as in this example can enhance theatric effect by reducing predictability. In technical writings such as pathology report, however, a simple clear-cut and predictable logic should be followed, instead. Specifically, negation sematic unit should lead a sentence to keep predictability of the statement and, thus facilitate reading comprehension and speed.
Thus, "-- No invasive carcinoma is identified in this specimen" is strongly preferred to
"-- An invasive carcinoma in this specimen is not identified" or "-- An invasive carcinoma is not identified in this specimen"
Such is also consistent with the observation in cognitive science that readers believe that things that are close together belong together. It is also consistent with progressive disclosure approach frequently employed to enhance logic coherence and readability of written text (ref)
Although this very issue of negation statement in pathology report was debated, only anecdotes of personal experience were given as the rhetoric basis (ref). Cognitive science, on the other hand, provides evidence to support that negative sematic unit leading a statement makes the statement (e.g., diagnosis line in pathology report) predictable, improves reading speed and reduces erroneous comprehension. For understandable pugenous opinion, exemplary statements in U.S. Constitution may serve as rhetoric defense, one of which is given below.
"No State shall, without the Consent of the Congress, lay any Imposts or Duties on Imports or Exports, except what may be absolutely necessary for executing it's inspection Laws: ... " In this example from the US Constitution, Article 1, Section 10, two negation sematic units (chunks) are placed to lead the statement.
Key point to take: Lead your statement by the most important, definite and unequivocal text component.
When to apply: statement of negative findings (e.g., "-- No dysplasia or carcinoma is identified")
Report format and layout are important in correct communication and data retrieval
Titles and Headlines are critical (Pay attention to literal style and syntactic format)
When a specimen consists of multiple parts, grouping diagnoses under each part is much easier to read, comprehend and remember. This format is used for reporting findings in placenta, specimen of hysrectomy and bilateral salpingoophrectomy, etc.
Similar to the geon theory on object recognition
People remember only four items at once (George A. Miller, 1956). This four-item rule also applies to memory retrieval (Donald Broadbent, 1975).
People process information better in bite-sized chunks. use chunks to turn four into more (placenta report -- disc, cord, membranes).
People believe that things that are close together belong together: This also is consistent with progressive disclosure approach
People read faster with longer line length but prefer short ones: progressive disclosure
Capital letters are inherently hard to read (ref). Thus, except in Headings and warnings, capital letters should be avoided. These days, text in all caps is perceived as "shouting".
Focal usual ductal hyperplasia vs. focal areas of
No carcinoma vs. no evidence of carcinoa v.s. carcinoma is not ...
Fragments of endometrium showing
leading negation vs trailing negation....
implied diagnosis and explicit diagnosis
e.g. EMC of 53-yof: strips of inactive and atrophic glandular epithelium (implying no malignancy)
Kevin Larson: The Science of Word Recognition or how I learned to stop worrying and love the bouma. Advanced Reading Technology, Microsoft Corporation, July 2004
By Flesch-Kincaid readability formular,
You can test the readability at: http://www.standards-schmandards.com/exhibits/rix/index.php
Carey, S. (1986). Cognitive science and science education. American Psychologist, 41, 1123-1130. Reprinted in Open University Press, Readings in the Psychology of Education and in C. Hedley, J. Houtz, & A. Baratta (eds.), Cognition, Curriculum, and Literacy. Norwood, NJ: Ablex, 1990.
(Susan Carey, 1986). (http://rosenfeldmedia.com/books/mental-models/)
Hoover, W.A. & Gough, P.B. (1990). The simple view of reading. Reading & Writing, 2(2), 127-160.
© 2003-2017 NuoNuo Medical Informatics, LLC.