Medical Dictionaries in the 21st CenturyJoe Segen2018-10-05T11:51:51+00:00
Medical Dictionaries in the 21st Century
Part 1 State of the art
Arguably, the most important book in any language is a dictionary, because it defines words and phrases in a standard way that allows Party A to communicate with and understand Party B and vice versa. The medical dictionary’s role is to facilitate communication by defining terminology used in medicine. But today’s medical dictionaries are increasingly out of touch with the needs of professional communication, an assertion that applies to both the printed (paper) as well as the electronic versions of the medical dictionary.
There are currently four major medical dictionaries:
• Dorland’s Illustrated Medical Dictionary (first published in 1890)
• Mosby’s Dictionary of Medicine, Nursing and the Health Professions (1982)
• Stedman’s Medical Dictionary (1911)
• Taber’s Cyclopedic Medical Dictionary (1940)
For the purposes of this opinion piece, I will only refer to the Dorland’s Illustrated Medical Dictionary, as it has been in print the longest (since 1890), is the largest (2097 pages), and thus can be regarded as representative of the group.
Question: Do today’s medical dictionaries serve the needs of physicians? I believe the answer is no, due to two main factors:
• Paper and format
Paper and format: The death knell has sounded on paper-based information products…from encyclopedias to phone directories, all are headed towards extinction. After 244 years of continuous production, in 2012, the venerated Encyclopedia Britannica ended its print run and re-purposed itself as a purveyor of online information. Information products made from paper have at least three disadvantages:
• Time to publication
Size: The Dorland’s was first published in 1890 as the American Illustrated Medical Dictionary. Over the years, it has grown substantially: in 1890, it had some 30,000 entries in 770 pages; the current edition (2012) has 124,000 entries in 2097 pages. It has also grown considerably in physical size, from 6 x 9 inches to today’s 8.75 X 11 inches, and weight.
Based on the changes in page number and size, it’s fair to say that Saunders, Dorland’s publisher, has done everything possible to pack as much information between the Dorland’s covers as possible. However, the increase in size has come at a cost of weight. No matter how thin the paper, the Dorland’s is frankly unwieldy. When a book becomes difficult to comfortably hold and use, many would regard it as too big.
Time to publication: It takes 6 months to a year from the time an author is finished with a manuscript and the printed book is available for purchase. The process rarely takes less than 6 months from author sign-off to shipped product. Today, most would regard the timeframes for producing a paper-formatted reference work as unacceptably slow.
Permanence:A third disadvantage of paper products is their permanent nature. A therapeutic agent billed as a wonder drug today, might prove in post-marketing (Phase 4) studies to have unacceptable side effects and thus would be removed from the shelf…after 50,000 copies of a work extolling that agent’s virtues had been printed.
How would one communicate such important changes to end-users?
Eventually a new edition of the work would be published and the errors present in the previous edition corrected (something that doesn’t always occur).
Electronic products address most of the disadvantages of paper, to wit, size and weight limitations, (long) time to publication, and permanence of misinformation. A dictionary that reaches a ceiling of ± 2000 pages with an 8-point typeface in a paper format, can be 20,000 pages in an electronic format and weigh not an ounce more. Because electronic products have no issues related to printing, warehousing and shipping of hard copy, the work can be as timely as this week’s issue of the New England Journal of Medicine or the British Medical Journal. And if the information in an electronic product today becomes misinformation tomorrow, the end-user would need only to log in and update the App, assuming the correction was made in the master file.
Given that the market is moving away from paper-based information products, does the electronic (e-Book) format used by Dorland’s and other medical dictionary publishers solve all of the above issues? No…see below.
Content:The second issue—and far more important than the challenges of paper-formatted products—relates to content and the type, quality and amount of information that we as physicians should expect from secondary information products. Is the material in today’s medical dictionaries relevant, authoritative, comprehensive, current and valuable to medically-qualified doctors?
For some areas–e.g., anatomy, medical dictionaries get it right all of the time. This is hardly surprising, given that once an anatomic structure has been described and its name agreed upon, it doesn’t change. Yet for most other areas of practice, current medical lexicons fall far short of physician needs. Despite producing new editions in 5 year cycles, they are increasingly out of touch with the needs of medical students, biomedical scientists and advanced health professionals.
Below is an array of issues which I feel—as a medical lexicographer with 3-plus decades (from 1984 to 2018) of experience—are major shortcomings that current medical dictionaries have failed to address.
Content Issues in Current Medical Dictionaries
• Terms not in current use
• Aliases and synonyms
• Pronunciation guidelines
• Who named it?
• Redundancy of material
• Trials and their acronyms
• Alphabetical order
• Lingua franca
• Authoritativeness, referencesand resources
• Education and reading level
• Wrong names
• Apples and oranges
(1) Terms not in current useRepresentative pages were selected from the 2012 Dorland’s. One finds an average of 3.2 terms per page that only appear in secondary or non-English language literature, which means that these terms are not used in the working medical parlance and should be retired.
(2) Aliases and synonymsMany medical terms are known by other names, aliases or synonyms. As examples, uninuclear is more commonly termed mononuclear, hypersomnia-bulimia syndrome is better known as Kleine-Levin syndrome, and Christmas disease as hemophilia B.
Aliases and synonyms (A&Ss) have two distinct disadvantages:
The term is listed at least twice (once for the definition and once for the alias—or more if there are multiple aliases)
The reader who lands first on the alias must go elsewhere to find the definition
The average page in the Dorland’s has 15.5 A&Ss. Many of the A&Ss in the Dorland’s–e.g., uninuclear, as above, are hen’s tooth rarities which would not be included in any current publication. One would expect to find common A&Ss in a medical dictionary, but not, for example, all 60+ synonyms for the classic lesion of cutaneous leishmaniasis.
(3) Pronunciation guidelinesMedical dictionaries often include a brief, text-based pronunciation guideline for the term being defined. Noble as is the effort, medical terms are still mispronounced, often creatively, sometimes comically, begging the question of whether space should be used for pronunciation, when the effort is largely ineffective. Eliminating pronunciation guidelines has already been done. Thirty years ago, The International Dictionary of Medicine and Biology (John Wiley & Sons, 1986, NYC), a 3-volume behemoth sidestepped the issue and did not include them.
(4) Who named it?Medical dictionaries typically include a line of text about the doctor, scientist or even the name of the family first associated with a particular syndrome, disease, anatomic structure, test, and so on. Should we honor our professional ancestors and provide a mini-biography in a medical dictionary? The traditional answer has been yes. Given that space in paper dictionaries is at a premium and too many important or widely used terms–e.g., fast-tracking, managed care, tertiary hospital, and too many genes to mention* are ignored, I believe that the answer is no.
*The Dorland’s lists less than 100 of the more than 20,000 genes thusfar identified.
(5) AbbreviationsNo medical dictionary would be complete without translations of abbreviations commonly used in medicine. The abbreviations chosen for inclusion in the Dorland’s fall far short of end-user needs because:
(a)Many abbreviations are no longer used. Latin has not been used for over 50 years, yet the Dorland’s abbreviations include such anachronisms as
abs feb—while fever is absent,
agit vas—the vial being shaken, and
cochl amp—a heaping spoonful.
(b) Abbreviations in common use are not included.
AAA–abdominal aortic aneurysm
CMS–Centers for Medicare and Medicaid Services
The absence of AMA–American Medical Association and the BMA–British Medical Association, is particularly disturbing, given their importance to current medical practice.
(c) Abbreviations are included, but the more common translations are not:
ADA The Dorland’s translates this as adenosine deaminase. It does not include the more common translation, Americans with Disabilities Act
BAL The Dorland’s translates this as either British Anti-Lewisite or biphenotypic acute leukemia. It does not include the more common translation, bronchoalveolar lavage
LFA The Dorland’s translates this as left frontoanterior. It does not include the more common translation, left femoral artery.
(6) Redundancy Given that space is limited in medical dictionaries, one would not expect to see the exact same material in more than one place. However in the 2012 edition of the Dorland’s, the same material that one sees in:
Appendix 1 (pp 2101-2107) Frequently Used Stems,
Appendix 2 (pp 2108-2125) Selected Abbreviations Used in Medicine (discussed in part, above), and
Appendix 5 (p 2133) Table of Elements
is repeated elsewhere, in alphabetical order with the other entries in the body of the dictionary.
This is not to say that stems, abbreviations and the chemical elements are not useful. They are. Stems/Greek and Latin roots are integral to the language of medicine as are abbreviations. And minutiae about the known elements are central to the science of medicine. But it’s hard to justify repeating 26 pages of material when so many terms in medical English–e.g., biosimilar, impact boosting, medical air, and microsleep, to name a few, are not included.
(7) TrialsA body of terminology that medical dictionaries have traditionally ignored are therapeutic trials and the waggish acronyms sometimes attached to them*. Despite their names, trials play a key role in improving clinical practice and in assessing new therapies. The importance of the TIMI (Thrombolysis in Myocardial Infarction) studies (and countless others) to evidence-based practice cannot be overstated. A medical dictionary that blatantly ignores trials and the related work product fails the test of relevance.
*Examples of interesting acronyms abound and include
VENUS (Very Early Nimodipine Use In Stroke)
EARTH (Endothelin Receptor Antagonist Trial In Heart Failure)
MARS (The Monitored Atherosclerosis Regression Study)
JUPITER (Justification For The Use Of Statins In Prevention: An Intervention Trial Evaluating Rosuvastatin)
(8) Timeliness A dictionary used by health professionals should be current and timely. A dictionary that incorporates information on newly approved drugs, procedures, newly recognised genes, and so on up to the time it goes to press is current. A dictionary in which the content is modified based on new information is timely. Here, the Dorland’s falls short: it contains information on drugs as if they are active agents, when some were pulled from the market up to 11 years earlier.
Even more egregious than including information on drugs long removed from the shelf is the retention of certain eponyms–e.g., Hallervorden-Spatz disease and Reiter syndrome. Julius Hallervorden was a member of the Nazi Party and knowingly carried out experiments on the brains of euthanised prisoners. Even more reprehensible was Hans Conrad Julius Reiter, a member of the SS (Schutzstaffel), who was convicted of war crimes for medical experiments on prisoners at the Buchenwald concentration camp. In one experiment, prisoners were inoculated with an unproven typhus vaccine that resulted in over 200 deaths.
A movement to strip Hallervorden and Reiter of their eponyms began in 1977 and gained momentum in 2001. By 2003*, many peer reviewed communications had abandoned the errant eponyms for acceptable alternatives, to wit, Hallervorden-Spatz disease is now preferably known as neurodegeneration with brain iron accumulation 1, and Reiter syndrome as reactive arthritis. Yet in the 32nd edition of the Dorland’s, which was published in 2012, the offending names were retained. Nearly 10 years on, the Dorland’s had not gotten the message.
*Seminars in Arthritis and Rheumatism 2003; 32 (4): 208–230, 231–236.
(9) BreadthThe world has changed considerably in the 125 years since the American Illustrated Medical Dictionary, Dorland’s predecessor, was first published in 1890. Medical practice at the end of the 19th century typically was that of a simple doctor-patient relationship in a rural setting with few effective therapies.
Multiple factors have contributed to the widening of medicine’s scope:
Changed practice philosophies–e.g., evidence-based practice, defensive medicine
Different funding formats–e.g., managed care, socialised medicine
Medicalization of ancillary fields–anthropology, forensics
Molecular medicine and a widening palette of genes are becoming diagnostically and therapeutically relevant
New technologies–e.g., AI, informatics, robotic surgery
Shrinking of the planet–e.g., human rights, terrorism
Specialization and sub-specialization–e.g., retinal surgery, wilderness medicine
Each of these areas impact on medicine and bring their own vocabulary, very little of which has found its way into major medical dictionaries.
(10) Medspeak Jargon is so integral to the daily practice of medicine that its use would be impossible to quantify. Examples include vasculopath (cardiology), clipase (endocrinology), big guns (oncology), central executive (neurology) and the highly colloquial whoops procedure (surgery). Because of its potentially offensive gestalt, some medical slang is only used verbally, rarely written and certainly never in patient records. Examples of spoken Medspeak include:
CYA (cover your ass), a slang alias for the practice of defensive medicine
CTD (circling the drain), a slang reference to a moribund state of health, and
FLK (funny looking kid), a slang term used in neonatal units to describe an infant who doesn’t have the classic facial dysmorphism of trisomy 21 (Down syndrome) or other common chromosomal defects, but is likely to have one, absent the infant having FLPs (funny looking parents).
Doctors begin learning Medspeak early in their careers, first as medical student victims of pimping. As trainees, they learn the arts of buffing and turfing patients, trying to avoid a bounce. The language of medicine is replete with slang. For a medical dictionary to be relevant to its target audience, here, physicians and advanced health professionals, it must define terms used by its practitioners. Spoken or written, offensive or banal, Medspeak is an integral part of the language of medicine and must be included in the same work that includes carpal tunnel syndrome, cholecystitis, melanin and radial nerve.
(11) Alphabetical order It is common practice in medical dictionaries to address compound phrases–e.g., iron-deficiency anemia, ischemic heart disease, Marfan syndrome by first alphabetizing the noun, then alphabetizing the adjective.
To repeat: Medical dictionaries aren’t always arranged in alphabetical order.
Iron-deficiency anemia is found under A–for anemia, ischemic heart disease is under D–for disease and Marfan syndrome is under S–for syndrome. The stated advantage of this naming convention is that it allows grouping of related terms.
The obvious disadvantage is that one has to know what noun was attached to the adjective to be able to find the definition. Is it Marfan’s disease or Marfan syndrome? Is it Peters-plus anomaly or Peters-plus syndrome? Is it Western blot or Western blot analysis? A brief flick-through of the current Dorland’s indicates that up to 10% of the entries are not in their natural order, but rather placed in a subgroup.
Many users would find non-alphabetization in a dictionary counter-intuitive.
Perhaps even more disturbing is the fact that the Dorland’s doesn’t even follow its own unnatural rules of alphabetization. On page 882, we find 11-beta hydroxylase deficiency, followed by 17-alpha hydroxylase deficiency, and 21 hydroxylase deficiency, which one might have expected to find closer to page 477, where other deficiencies are defined.
(12) Lingua franca Whilst classic Greek and Latin laid the foundation for the language of medicine, English has become its lingua franca. Virtually all medical conferences and peer-reviewed communications are in English, even in countries where English is not the native language. For those countries in which English is the native language, de-Latinization has the advantage of increased patient comprehension of advice, instructions, and informed consent documents. A few pockets of Latin and Greek-based terminology remain, in particular in anatomy and for dermatologic disorders, but for the most part, English is the language of written and spoken medicine.
The transition away from the classic (Greek and Latin) roots of medicine has been occurring for decades. As an example, the word lemmocyte (page 1020, Dorland’s), defined as a cell of neural crest origin was last used in the primary literature in the 1960s (most of the 227 results on a google search for lemmocyte were in secondary or non-English language literature). The English equivalent, neural crest cell, garnered 127,000 hits on 26.04.2016 and thus is the preferred term. It seems reasonable to assume that a medical dictionary would have evolved to reflect the working parlance and have more English terms and fewer Latin- and classic Greek-based words in the 2012 Dorland’s than in the 1932 edition.
It does not.
(13) Authoritativeness, references and resourcesIt is inconceivable in the 21st century that a peer-reviewed communication could be made without its authors referring to prior work. References reduce the risk of plagiarism and allow readers to follow the authors’ thinking, experimental processes and conclusions, especially in complex areas of immunology, molecular biology and therapeutics. Medical dictionaries are not peer-reviewed works and thus have traditionally been exempt from stringent (read, any) requirement for references. They are regarded as secondary sources and are held to a lower standard for citing source material. They rely instead on consultants and learned authorities to support their legitimacy.
Prior convention (that of not quoting sources) notwithstanding, it is difficult to rely on an information product unable to quote its sources, especially considering the ready availability of regularly updated online resources for:
(14) Comprehensiveness The Oxford English Dictionary (OED) defines comprehensiveness as: the quality of comprising or including much. For a lexicon to be useful, each of its definitions should provide the targeted user demographic (physicians) a satisfactory amount of information (comprehensiveness) about the word or phrase being defined. For a muscle, an adequate definition would include the muscle’s origin, its insertion in bone, blood vessels supplying it or its region, its innervation, function, and types of injury with which it may be uniquely susceptible, if applicable.
Whilst it is impossible to specify the number of words required to meet the comprehensiveness standard for all terms, the relevant details for a muscle require 40 to 50 words, which translates to 3 to 4 lines of text. Few would argue that inflammation of the gallbladder, the Dorland’s four-word definition for cholecystitis is woefully inadequate, as are 6 lines of text for retinitis pigmentosa–RP, when there were, at last count, 82 genes responsible for RP, or 9 lines of text for AIDS (acquired immunodeficiency syndrome).
However, for the sake of the fiction created herein, and with the readers’ permission, I will regard 3 lines of text and 20 words as sufficient to adequately define the average term in a medical dictionary. The average page in the Dorland’s has 22 terms with adequate (≥ 3 lines of text) definitions which, multiplied by 2097 (the number of pages) means that the Dorland’s has ± 46,134 entries with at least 20-word definitions. To call a 20-word definition adequate seems generous. What is more disturbing is that the Dorland’s, which claims to be the biggest and best medical dictionary on the market with 124,000 terms, has adequate definitions for less than half.
(15) Currency The OED defines currency as: the fact or quality of being current. The 2012 Dorland’s added 5,500 new terms to the immediately preceding 2007 edition. Based on this, the Dorland’s lexicography staff of four adds ± 1 term/person/day.
First question: what are they doing all day?
Second question: is 5,500 added to a medical lexicon in 5 years enough?
5500 divided by 124,000 (total number of terms) = 4.4% new terms added in 5 years, i.e., ± 1%/year.
It is estimated that the doubling time of medical knowledge in 1950 was 50 years; in 1980, 7 years; and in 2010, 3.5 years. In 2020, it is projected to be 0.2 years—just 73 days*.
Whilst 5,500 might appear to be a big number, some users might be less dazzled by a 1% addition of new information every year, when medical knowledge is doubling every three. To call the Dorland’s current is to take license with the word “current”.
(16) Education level It is likely that any work written at the education level of the general (non-medical) population will fall short of the needs of medical students, junior doctors and specialists. Physicians and trainees, by the very nature of their professional schooling, will be more knowledgeable in health and disease processes and management than a medically unsophisticated audience. It seems reasonable to expect that a reference intended for advanced health professionals would provide definitions that are more useful and specific than:
…elevation of the body temperature above normal (the definition of fever), or
…examination by means of thefluoroscope (fluoroscopy), or
…the study of genes and their heredity (genetics)
The above definitions appear to satisfy the needs of at least some consumers, because they’ve stood the test of time.* But to pretend that these definitions are useful to health professionals with a decade or more of post-university medical education and training is overly generous.
*Eight decades, to be exact. The definition of fever in the 2012 Dorland’s is virtually identical to that found in the 1932 edition (abnormally high bodily temperature). The definition for fluoroscopy in the 2012 Dorland’s is exactly the same as the 1932 edition. And, incredibly, the definition in the 2012 Dorland’s for genetics differs little from that of the 1932 edition (the science that deals with the origin of the characteristics of the individual; the study of heredity), this, in a field that has undergone more changes than any other field in medicine.
This divide could be called the “Goldilocks dilemma”. Like the Baby Bear’s anemic offerings in the fairy tale, Goldilocks and the Three Bears, the definitions found in the Dorland’s are too simple (and too brief) to be useful to a Dr. Goldilocks. On the opposite end of the information spectrum, Papa Bear’s offerings, in this analogy, original reports and reviews in peer-reviewed journals would be excessive for a physician wanting a quick textbite of useful information. A Mama Bear’s bread and bed would be just right. Dictionary definitions are exercises in brevity, but when they’re too brief, as above, they lose an entire readership. Adding a few lines of text to otherwise meager definitions–e.g., providing exact temperatures and management strategies for fever; amount of radiation exposure and indications for fluoroscopy; and internet resources–e.g., genecards.org, omim.org, and uniprot.org would make definitions useful for a professional audience.
(17) Wrong name One would expect a reference in its 32nd edition* to use the best or preferred name for a term being defined. The term chronic granulocytic leukemia (Dorland’s name) got 54,400 hits when googled on 26 October 2016 compared to the widely preferred term, chronic myelogenous leukemia’s 545,000 hits (chronic myeloid leukemia got 461,000 hits, chronic myelocytic leukemia got 119,000). Similarly, congenital megacolon, the term used in the 2012 Dorland’s got 35,300 hits compared to Hirschsprung disease’s 389,000 hits (Hirschsprung’s disease’s 254,000 hits). The author identified these wrongly named entities with the briefest of efforts; how many more wrong names are lurking in a work with nearly 124,000 entries?
*The first edition of the Dorland’s was published in 1890 as the American Illustrated Medical Dictionary
(18) Typographical errors Typos are the bane of all published work. It would be delusional to think that they can be completely eliminated, but they can be minimized by obsessive attention to detail and adequate budgets for copy editing. Little typos slip through; it would be surprising for a work as large as a major medical dictionary with an estimated 2.7 million words* to have no mistakes at all.
* 144 lines of text/page X ± 9 words/line X 2097 pages
Still, one would expect a book in its 32nd outing to be free of glaringly obvious typos. Such is not the case: the caption on page 361 of the Dorland’s reads Electron micrograph of ciliated columnar pseudostratified epithelium in the respiratory tract, for what is clearly an H&E stained medium power photomicrograph. I found that typo purely by accident, which raises a slightly different question than above…what are the Dorland’s consultants doing all day, given that three are professors of pathology and one is a practicing pathologist who looks at glass slides all day and would know the difference between an H&E-stained slide and an electron micrograph?
(19) Apples and oranges Sequencing of the human genome in 2003 ushered in a new—and far more complex—era in medicine. It’s no longer enough to separate apples from oranges; now apples must be separated from other apples, as Cripps Pink, Fuji, Gala, Granny Smith, Honeycrisp, and Red or Golden Delicious. Before 2003, it was enough to provide a clinical diagnosis of a disease–e.g., retinitis pigmentosa (RP), given the generic nature of RP management strategies.
It is now known that RP is caused by defects of over 80 different genes, the latest identified forms of RP are RP81 (OMIM:617871), which was described in 2017 by Biswas et al and caused by a defect of IFT43, the gene that encodes asubunit of a transport complex involved in retrograde ciliary transport along axonemal microtubules.
Whilst the management options for RP remain unchanged for the near term, identifying the specific gene causing a determined form of a disease lays the groundwork for gene knockout therapies which are now peeking over the horizon with the likes of CRISPR and other gene manipulation strategies likely to appear in the future.
The 2012 Dorland’s may not acknowledge that most inherited conditions are in fact groups of disorders. As examples, it defines
Achromatopsia with a mere two words, monochromatic vision, leaving it to the reader to discover the six genetic forms of achromatopsia in the OMIM (Online Mendelian Inheritance of Man) database
Amyotrophic lateral sclerosis (ALS) as a …motor neuron disease marked by progressive degeneration of the neurons that give rise to the corticospinal tract and of the motor cells of brainstem and the spinal cord, resulting in a deficit of upper and lower motor neurons; it usually ends fatally in two or three years. The opening phrase, a motor neuron disease…would suggest to a reasonable person that ALS is a single disease, not a group of at least 22 genetically distinct causes (again, per the OMIM database) of the same clinical disease, and finally,
Senior-Loken syndrome is defined as a rare autosomal recessive syndrome of tapetoretinal degeneration and familial juvenile nephronophthisis…Senior-Loken syndrome is in fact a group of 9 (at last count) genetically distinct forms of the same clinical disease.
This last point may be the most problematic of the content-related issues challenging current medical dictionaries. Despite sharing similar or even identical clinical pictures, most inherited conditions are now known to be groups of genetically disparate diseases. For a medical dictionary to remain relevant in the post-genomic 21st century, it would need to tabulate or catalogue the genetic forms of retinitis pigmentosa, spinocerebellar ataxia, and amelogenesis imperfecta–to name a few of the thousands of genetic forms of inherited diseases. The work would also need to provide details for thousands of genes linked to those syndromes. This seems unlikely for many of the above noted reasons.
Conclusion The present opinion piece asks the basic question:
Is the text version of the medical dictionary–taking here as representative of the group, the Dorland’s Illustrated Medical Dictionary–a fixture in health professionals’ libraries, still a useful source of information for physicians and advanced health professionals, or is it an anachronism?
Douglas M Anderson, The Dorland’s Illustrated Medical Dictionary’s chief lexicographer, writes in the Preface of the 2012 edition: “We are confident that the 32nd edition will continue to provides its users with the same authoritativeness, comprehensiveness and currency that they have come to expect from Dorland’s” (Preface, page vii).
One might question this use of the words authoritativeness ,comprehensiveness, and currency. The author believes that current medical dictionaries, in particular, the Dorland’s Medical Dictionary, are none of these.